Knowledge Management in Organizations : 12th International Conference, KMO 2017, Beijing, China, August 21-24, 2017, Proceedings 978-3-319-62697-0, 3319626973, 978-3-319-62698-7, 3319626981

Table of contents :
Front Matter ....Pages I-XIII
Front Matter ....Pages 1-1
Activity Theory Based Approach for Requirements Analysis of Android Applications (Kandiraju Sai Ashritha, T. M. Prajwala, K. Chandrasekaran)....Pages 3-15
Machine Consciousness of Mind Model CAM (Zhongzhi Shi, Gang Ma, Jianqing Li)....Pages 16-26
The Knowledge Communication Conceptual Model in Malaysian Public Sector (Rohaizan Daud, Nor Zairah Ab Rahim, Roslina Ibrahim, Suraya Ya’acob)....Pages 27-38
Applying Process Virtualization Theory in E-HR Acceptance Research: Testing and Modifying an Experiment (C. Rosa Yeh, Shin-Yau Hsiao)....Pages 39-48
Front Matter ....Pages 49-49
Virtual Teams Stress Experiment Proposal: Investigating the Effect of Cohesion, Challenge, and Hindrance on Knowledge Sharing, Satisfaction, and Performance (Andree E. Widjaja)....Pages 51-63
Knowledge Sharing on YouTube (Eric Kin Wai Lau)....Pages 64-71
Internal Knowledge Sharing Motivation in Startup Organizations (Jouni A. Laitinen, Dai Senoo)....Pages 72-83
Knowledge Sharing in Constructability Management (Marja Naaranoja, Joni Vares)....Pages 84-94
Front Matter ....Pages 95-95
Skills Sets Towards Becoming Effective Data Scientists (Wardah Zainal Abidin, Nur Amie Ismail, Nurazean Maarop, Rose Alinda Alias)....Pages 97-106
E-Learning Platforms Analysis for Encourage Colombian Education (Lizeth Xiomara Vargas Pulido, Nicolás Olaya Villamil, Giovanny Tarazona)....Pages 107-118
Using a Simulation Game Approach to Introduce ERP Concepts – A Case Study (Marjan Heričko, Alen Rajšp, Paul Wu Horng-Jyh, Tina Beranič)....Pages 119-132
Knowledge Creation Activity System for Learning ERP Concepts Through Simulation Game (Horng-Jyh Paul Wu, See Tiong Beng, Marjan Heričko, Tina Beranič)....Pages 133-143
Entrepreneurship Knowledge Transfer Through a Serious Games Platform (Dario Liberona, Cristian Rojas)....Pages 144-156
Front Matter ....Pages 157-157
Knowledge for Translating Management Innovation into Firm Performance (Remy Magnier-Watanabe, Caroline Benton)....Pages 159-169
Product vs. Service War: What Next? A Case Study of Japanese Beverage Industry Perspective (Zahir Ahamed, Akira Kamoshida, H. M. Belal, Chris Wai Lung Chu)....Pages 170-181
Comparing the Perceived Values of Service Industry Innovation Research Subsidiary Between Reviewers and Applicants in Taiwan (Yu-Hui Tao, Yun-An Lin)....Pages 182-189
Validation Tools in Research to Increase the Potential of its Commercial Application (Anna Závodská, Veronika Šramová, Anne-Maria Aho)....Pages 190-199
Front Matter ....Pages 201-201
Global Studies About the Corporate Social Responsibility (CSR) (Sahar Mansour)....Pages 203-213
Enhancing Work Engagement Towards Performance Improvement (Chris W. L. Chu, Reuben Mondejar, Akira Kamoshida, Zahir Ahamed)....Pages 214-227
Knowledge Management and Triangulation Logic in the Foresight Research and Analyses in Business Process Management (Kaivo-oja Jari, Lauraeus Theresa)....Pages 228-238
Corporate Knowledge Management, Foresight Tools, Primary Economically Affecting Disruptive Technologies, Corporate Technological Foresight Challenges 2008–2016, and the Most Important Technology Trends for Year 2017 (Kaivo-oja Jari, Lauraeus Theresa)....Pages 239-253
Non Profit Institutions IT Governance: Private High Education Institutions in Bogota Case (Yasser de Jesús Muriel Perea, Flor Nancy Díaz-Piraquive, Rubén González Crespo, Ivanhoe Rozo Rojas)....Pages 254-269
Front Matter ....Pages 271-271
Decision Making in Cloud Computing: A Method that Combines Costs, Risks and Intangible Benefits (Yuri Zelenkov)....Pages 273-285
Study on Network Online Monitoring Based on Information Integrated Decision System (Fan Yang, Zhenghong Dong)....Pages 286-294
Marine Geochemical Information Management Strategies and Semantic Mediation (Tenglong Hong, Xiaohong Wang, Jianliang Xu, Shijuan Yan, Chengfei Hou)....Pages 295-306
Implementation of a Text Analysis Tool: Exploring Requirements, Success Factors and Model Fit (Giorgio Ghezzi, Stephan Schlögl, Reinhard Bernsteiner)....Pages 307-317
Front Matter ....Pages 319-319
A Workflow-Driven Web Inventory Management System for Reprocessing Businesses (Huanmei Wu, Jian Zhang, Sunanda Mukherjee, Miaolei Deng)....Pages 321-335
Towards Information Governance of Data Value Chains: Balancing the Value and Risks of Data Within a Financial Services Company (Haifangming Yu, Jonathan Foster)....Pages 336-346
A Dempster Shafer Theory and Fuzzy-Based Integrated Framework for Supply Chain Risk Assessment (Yancheng Shi, Zhenjiang Zhang, Kun Wang)....Pages 347-361
Front Matter ....Pages 363-363
A Minimal Temporal Logic with Multiple Fuzzy Truth-Values (Xinyu Li, Xudong Luo, Jinsheng Chen)....Pages 365-377
Modified Similarity Algorithm for Collaborative Filtering (Kaili Shen, Yun Liu, Zhenjiang Zhang)....Pages 378-385
Healthcare-Related Data Integration Framework and Knowledge Reasoning Process (Hong Qing Yu, Xia Zhao, Zhikun Deng, Feng Dong)....Pages 386-396
Front Matter ....Pages 397-397
Algorithms for Attribute Selection and Knowledge Discovery (Jorge Enrique Rodríguez R., Víctor Hugo Medina García, Lina María Medina Estrada)....Pages 399-409
P-IRON for Privacy Preservation in Data Mining (G. Arumugam, V. Jane Varamani Sulekha)....Pages 410-423
Education Intelligence Should Be the Breakthrough in Intelligence Science (Chuan Zhao)....Pages 424-434
An Improved Method for Chinese Company Name and Abbreviation Recognition (Lei Meng, Zhi-Hao Wei, Tian-Yi Hu, Qian Cheng, Yu Zhu, Xiao-Tao Wei)....Pages 435-447
Front Matter ....Pages 449-449
A Big Data Application of Machine Learning-Based Framework to Identify Type 2 Diabetes Through Electronic Health Records (Tao Zheng, Ya Zhang)....Pages 451-458
A Case Study on API-Centric Big Data Architecture (Aravind Kumaresan, Dario Liberona, R. K. Gnanamurthy)....Pages 459-469
Survey on Big Data Security Framework (M. Thangaraj, S. Balamurugan)....Pages 470-481
Towards Integrated Model of Big Data (BD), Business Intelligence (BI) and Knowledge Management (KM) (Souad Kamoun-Chouk, Hilary Berger, Bing Hwie Sie)....Pages 482-493
A Review of Resource Scheduling in Large-Scale Server Cluster (Libo He, Zhenping Qiang, Wei Zhou, Shaowen Yao)....Pages 494-505
Front Matter ....Pages 507-507
Mobile Application for the Reduction of Waiting Times Through IoT (Prototype) (Pedro Santiago Sierra, Jeyson Bolivar Siculaba, Giovanny Mauricio Tarazona)....Pages 509-519
Design Science and ThinkLets as a Holistic Approach to Design IoT/IoS Systems (Reinhard Bernsteiner, Stephan Schlögl)....Pages 520-533
Research on DV-HOP Algorithm for Wireless Sensor Networks (Xiao-Li Cui, Wen-Bai Chen, Cui Hao)....Pages 534-546
Dynamic Charging Planning for Indoor WRSN Environment by using Self-propelled Vehicle (Wei-Che Chien, Hsin-Hung Cho, Chin Feng Lai, Timothy K. Shih, Han-Chieh Chao)....Pages 547-559
Back Matter ....Pages 561-562

Citation preview

Lorna Uden · Wei Lu I-Hsien Ting (Eds.)

Communications in Computer and Information Science

731

Knowledge Management in Organizations 12th International Conference, KMO 2017 Beijing, China, August 21–24, 2017 Proceedings

123

Communications in Computer and Information Science Commenced Publication in 2007 Founding and Former Series Editors: Alfredo Cuzzocrea, Orhun Kara, Dominik Ślęzak, and Xiaokang Yang

Editorial Board Simone Diniz Junqueira Barbosa Pontifical Catholic University of Rio de Janeiro (PUC-Rio), Rio de Janeiro, Brazil Phoebe Chen La Trobe University, Melbourne, Australia Xiaoyong Du Renmin University of China, Beijing, China Joaquim Filipe Polytechnic Institute of Setúbal, Setúbal, Portugal Igor Kotenko St. Petersburg Institute for Informatics and Automation of the Russian Academy of Sciences, St. Petersburg, Russia Ting Liu Harbin Institute of Technology (HIT), Harbin, China Krishna M. Sivalingam Indian Institute of Technology Madras, Chennai, India Takashi Washio Osaka University, Osaka, Japan

731

More information about this series at http://www.springer.com/series/7899

Lorna Uden Wei Lu I-Hsien Ting (Eds.) •

Knowledge Management in Organizations 12th International Conference, KMO 2017 Beijing, China, August 21–24, 2017 Proceedings

123

Editors Lorna Uden University of Staffordshire Stoke-on-Trent, Staffordshire UK Wei Lu Beijing Jiaotong University Beijing China

I-Hsien Ting Department of Information Management University of Kaohsiung Kaohsiung Taiwan

ISSN 1865-0929 ISSN 1865-0937 (electronic) Communications in Computer and Information Science ISBN 978-3-319-62697-0 ISBN 978-3-319-62698-7 (eBook) DOI 10.1007/978-3-319-62698-7 Library of Congress Control Number: 2017945722 © Springer International Publishing AG 2017 This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, express or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Printed on acid-free paper This Springer imprint is published by Springer Nature The registered company is Springer International Publishing AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

Preface

The 12th International Conference on Knowledge Management in Organizations: Emerging Technology was held during August 14–21, 2017, at the Beijing Conference Center, China, hosted by Beijing Jiaotong University, China. Knowledge management (KM) is facing a challenging time with the advance of big data and the Internet of Things (IOT) as well as cognitive learning. It is important for KM to take up the challenges to help companies reach their goals and weather the storms ahead. Given the rise of big data, IOT, disruptive technology, and current market demands, KM research must be fully aware of the new challenges. Traditional KM approaches have failed to meet the challenges posed by Big Data, mobility, social media, and customer demands. There is a strong relationship between innovation, technology, and KM. ICTs have great influence on KM, and therefore any innovation in ICTs is directly linked to creativity in KM. Collaboration is essential in an economy based on highly specialized knowledge. There is a need to focus on fostering collaboration between individuals, teams, divisions, and organizations. It is important that we develop the skills and culture that enable high-value collaboration. Implementing a whole new set of business processes is also required to unlock the full potential of collaboration for KM. KM is not only limited to technology, but it is the integration of business strategy and process, organizational community and culture, expertise and technology. How do we develop products and services that will meet the values of the customers? To do this requires that we look into the new emerging discipline of service science and especially service-dominant logic. Co-creation of value is essential to provide services and products that will offer value to users. There are many research issues that need to be addressed to incorporate these new technologies in KM. A number of questions are to be addressed: What does the emergence of social media, big data, and IOT and related mobile media mean for KM initiatives in companies? What is the relevant research needed for academia? Which of our established KM models are still valid, where do we need new models or frameworks? This conference calls for contributions in light of these contexts. The KMO 2017 conference aimed to bring together leading academic researchers and scholars to exchange and share their experiences and expertise in all aspects of KM challenges. It also provided an interdisciplinary platform for researchers, practitioners, and educators to present and discuss their most recent work, trends, innovations, and concerns as well as the practical challenges encountered and the solutions adopted in the fields of KM in organizations. The conference welcomes contributions from researchers and scholars to contribute to original and unpublished results of conceptual, constructive, empirical, experimental, or theoretical work in all areas of KM in organizations at the conference. The conference solicits contributions of full papers that address the themes and topics of the

VI

Preface

conference. We are also interested in case studies that demonstrate how KM research strategies have been applied and the lessons learned. Research contributions on the aforementioned aspects will enlighten industry on how to handle the various organizational and technical opportunities and challenges in KM. KMO 2017 aimed to encourage research into the various aspects of KM especially in new research involving emerging issues, challenges, and trends. These and other related topics were discussed and explored at the 12th KMO 2017 conference. This year, we received 45 papers. All published papers have undergone a rigorous review process involving at least three reviewers. The authors of these papers come from 20 different including, Austria, Chile, China, Colombia, Finland, Hong Kong, India, Indonesia, Japan, Malaysia, Russia, Singapore, Slovakia, Slovenia, Spain, Taiwan, Tunisia, UAE, UK, and USA. The papers are organized into 11 thematic sections: • • • • • • • • • • •

Knowledge Management Models and Behavior Studies Knowledge Sharing Knowledge Transfer and Learning Knowledge and Service Innovation Knowledge and Organization Information Systems Research Value Chain and Supply Chain Knowledge Re-presentation and Reasoning Data Mining and Intelligent Science Big Data Management Internet of Things and Networks

Besides the papers, we also had invited keynotes and four tutorials. We would like to thank our (a) authors, reviewers, and Program Committee for their contributions and (b) Beijing Jiaotong University for hosting the conference. Special thanks to the authors and participants at the conference. Without their efforts, there would be no conference or proceedings. We hope that these proceedings will be beneficial for your reference and that the information in this volume will be useful for further advancements of KM in both research and industry. June 2017

Lorna Uden I-Hsien Ting Wei Lu

Organization

Conference Chair Lorna Uden

Staffordshire University, UK

Program Chairs Wei Lu I-Hsien Ting

Beijing Jiaotong University, Beijing, China National University of Kaohsiung, Taiwan

Local Chair Zhenjiang Zhang

Beijing Jiaotong University, Beijing, China

Program Committee XiaoTao Wei Jari Kaivo-Oja Houn-Gee Chen Reinhard C. Bernsteiner Hilary Berger Paolo Ceravolo Dario Liberona I-Hsien Ting Akira Kamoshida Costas Vassilakis Dai Senoo Eric Kin-Wai Lau G.R. Gangadharan George Karabatis Guandong Xu Hércules Antonio do Prado Lorna Uden Luka Pavlič Marja Naaranoja Marjan Heričko Paul Horng-Jyh Wu Remy Magnier-Watanabe Richard Self Takao Terano Victor Hugo Medina Garcia

Beijing Jiaotong University, Beijing, China FFRC, Turku School of Economics, Finland National Taiwan University, Taiwan Management Center Innsbruck, Austria Cardiff Metropolitan University, UK Università degli Studi di Milano, Italy Universidad Santa Maria, Chile National University of Kaohsiung, Taiwan Yokohama City University, Japan University of the Peloponnese, Greece Tokyo Institute of Technology, Japan City University, Hong Kong, SAR China IDRBT, Hyderabad, India University of Maryland, Baltimore County, USA University of Technology Sydney, Australia Catholic University of Brasília, Brazil Staffordshire University, UK University of Maribor, Slovenia Vaasa University of Applied Sciences, Finland University of Maribor, Slovenia SIM University, Singapore University of Tsukuba, Tokyo, Japan University of Derby, UK Tokyo Institute of Technology, Japan Universidad Distrital Francisco José de Caldas, Colombia

VIII

Organization

Wu He Yu-Hui Tao Whai-En Chen William Wang Carmine Gravino Yuri Zelenkov K. Chandrasekaran Marta Silvia Tabares B Jonathan Foster WeiWei Xing Chuan Zhao Albena Antonova Xudong Luo

Old Dominion University, USA National University of Kaohsiung, Taiwan National Ilan University, Taiwan University of Waikato, New Zealand Università degli Studi di Salerno, Italy Financial University under the Government of the Russian Federation, Russia National Institute of Technology Karnataka (NITK) Mangalore, India Universidad EAFIT, Medellín, Colombia University of Sheffield, UK Beijing Jiaotong University, Beijing, China Chengdu University of Technology, Chengdu, China Sofia University, Bulgaria Guaangxi Normal University, China

Contents

Knowledge Management Models and Behaviour Studies Activity Theory Based Approach for Requirements Analysis of Android Applications. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Kandiraju Sai Ashritha, T.M. Prajwala, and K. Chandrasekaran Machine Consciousness of Mind Model CAM . . . . . . . . . . . . . . . . . . . . . . Zhongzhi Shi, Gang Ma, and Jianqing Li The Knowledge Communication Conceptual Model in Malaysian Public Sector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Rohaizan Daud, Nor Zairah Ab Rahim, Roslina Ibrahim, and Suraya Ya’acob Applying Process Virtualization Theory in E-HR Acceptance Research: Testing and Modifying an Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Rosa Yeh and Shin-Yau Hsiao

3 16

27

39

Knowledge Sharing Virtual Teams Stress Experiment Proposal: Investigating the Effect of Cohesion, Challenge, and Hindrance on Knowledge Sharing, Satisfaction, and Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Andree E. Widjaja

51

Knowledge Sharing on YouTube . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Eric Kin Wai Lau

64

Internal Knowledge Sharing Motivation in Startup Organizations . . . . . . . . . Jouni A. Laitinen and Dai Senoo

72

Knowledge Sharing in Constructability Management . . . . . . . . . . . . . . . . . . Marja Naaranoja and Joni Vares

84

Knowledge Transfer and Learning Skills Sets Towards Becoming Effective Data Scientists. . . . . . . . . . . . . . . . Wardah Zainal Abidin, Nur Amie Ismail, Nurazean Maarop, and Rose Alinda Alias

97

X

Contents

E-Learning Platforms Analysis for Encourage Colombian Education . . . . . . . Lizeth Xiomara Vargas Pulido, Nicolás Olaya Villamil, and Giovanny Tarazona

107

Using a Simulation Game Approach to Introduce ERP Concepts – A Case Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Marjan Heričko, Alen Rajšp, Paul Wu Horng-Jyh, and Tina Beranič

119

Knowledge Creation Activity System for Learning ERP Concepts Through Simulation Game . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Horng-Jyh Paul Wu, See Tiong Beng, Marjan Heričko, and Tina Beranič

133

Entrepreneurship Knowledge Transfer Through a Serious Games Platform: The Venture Creation Game Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Dario Liberona and Cristian Rojas

144

Knowledge and Service Innovation Knowledge for Translating Management Innovation into Firm Performance . . . Remy Magnier-Watanabe and Caroline Benton

159

Product vs. Service War: What Next? A Case Study of Japanese Beverage Industry Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Zahir Ahamed, Akira Kamoshida, H.M. Belal, and Chris Wai Lung Chu

170

Comparing the Perceived Values of Service Industry Innovation Research Subsidiary Between Reviewers and Applicants in Taiwan . . . . . . . . . . . . . . Yu-Hui Tao and Yun-An Lin

182

Validation Tools in Research to Increase the Potential of its Commercial Application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Anna Závodská, Veronika Šramová, and Anne-Maria Aho

190

Knowledge and Organization Global Studies About the Corporate Social Responsibility (CSR) . . . . . . . . . Sahar Mansour

203

Enhancing Work Engagement Towards Performance Improvement . . . . . . . . Chris W.L. Chu, Reuben Mondejar, Akira Kamoshida, and Zahir Ahamed

214

Knowledge Management and Triangulation Logic in the Foresight Research and Analyses in Business Process Management . . . . . . . . . . . . . . . Kaivo-oja Jari and Lauraeus Theresa

228

Contents

Corporate Knowledge Management, Foresight Tools, Primary Economically Affecting Disruptive Technologies, Corporate Technological Foresight Challenges 2008–2016, and the Most Important Technology Trends for Year 2017 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Kaivo-oja Jari and Lauraeus Theresa Non Profit Institutions IT Governance: Private High Education Institutions in Bogota Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Yasser de Jesús Muriel Perea, Flor Nancy Díaz-Piraquive, Rubén González Crespo, and Ivanhoe Rozo Rojas

XI

239

254

Information Systems Research Decision Making in Cloud Computing: A Method that Combines Costs, Risks and Intangible Benefits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Yuri Zelenkov

273

Study on Network Online Monitoring Based on Information Integrated Decision System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fan Yang and Zhenghong Dong

286

Marine Geochemical Information Management Strategies and Semantic Mediation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tenglong Hong, Xiaohong Wang, Jianliang Xu, Shijuan Yan, and Chengfei Hou Implementation of a Text Analysis Tool: Exploring Requirements, Success Factors and Model Fit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Giorgio Ghezzi, Stephan Schlögl, and Reinhard Bernsteiner

295

307

Value Chain and Supply Chain A Workflow-Driven Web Inventory Management System for Reprocessing Businesses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Huanmei Wu, Jian Zhang, Sunanda Mukherjee, and Miaolei Deng

321

Towards Information Governance of Data Value Chains: Balancing the Value and Risks of Data Within a Financial Services Company. . . . . . . . Haifangming Yu and Jonathan Foster

336

A Dempster Shafer Theory and Fuzzy-Based Integrated Framework for Supply Chain Risk Assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Yancheng Shi, Zhenjiang Zhang, and Kun Wang

347

XII

Contents

Knowledge Re-presentation and Reasoning A Minimal Temporal Logic with Multiple Fuzzy Truth-Values . . . . . . . . . . . Xinyu Li, Xudong Luo, and Jinsheng Chen

365

Modified Similarity Algorithm for Collaborative Filtering . . . . . . . . . . . . . . Kaili Shen, Yun Liu, and Zhenjiang Zhang

378

Healthcare-Related Data Integration Framework and Knowledge Reasoning Process. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hong Qing Yu, Xia Zhao, Zhikun Deng, and Feng Dong

386

Data Mining and Intelligent Science Algorithms for Attribute Selection and Knowledge Discovery. . . . . . . . . . . . Jorge Enrique Rodríguez R., Víctor Hugo Medina García, and Lina María Medina Estrada

399

P-IRON for Privacy Preservation in Data Mining . . . . . . . . . . . . . . . . . . . . G. Arumugam and V. Jane Varamani Sulekha

410

Education Intelligence Should Be the Breakthrough in Intelligence Science . . . Chuan Zhao

424

An Improved Method for Chinese Company Name and Abbreviation Recognition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Lei Meng, Zhi-Hao Wei, Tian-Yi Hu, Qian Cheng, Yu Zhu, and Xiao-Tao Wei

435

Big Data Management A Big Data Application of Machine Learning-Based Framework to Identify Type 2 Diabetes Through Electronic Health Records . . . . . . . . . . Tao Zheng and Ya Zhang

451

A Case Study on API-Centric Big Data Architecture . . . . . . . . . . . . . . . . . . Aravind Kumaresan, Dario Liberona, and R.K. Gnanamurthy

459

Survey on Big Data Security Framework . . . . . . . . . . . . . . . . . . . . . . . . . . M. Thangaraj and S. Balamurugan

470

Towards Integrated Model of Big Data (BD), Business Intelligence (BI) and Knowledge Management (KM) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Souad Kamoun-Chouk, Hilary Berger, and Bing Hwie Sie A Review of Resource Scheduling in Large-Scale Server Cluster . . . . . . . . . Libo He, Zhenping Qiang, Wei Zhou, and Shaowen Yao

482 494

Contents

XIII

Internet of Things and Network Mobile Application for the Reduction of Waiting Times Through IoT (Prototype) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pedro Santiago Sierra, Jeyson Bolivar Siculaba, and Giovanny Mauricio Tarazona Design Science and ThinkLets as a Holistic Approach to Design IoT/IoS Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Reinhard Bernsteiner and Stephan Schlögl Research on DV-HOP Algorithm for Wireless Sensor Networks . . . . . . . . . . Xiao-Li Cui, Wen-Bai Chen, and Cui Hao Dynamic Charging Planning for Indoor WRSN Environment by using Self-propelled Vehicle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Wei-Che Chien, Hsin-Hung Cho, Chin Feng Lai, Timothy K. Shih, and Han-Chieh Chao Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

509

520 534

547

561

Knowledge Management Models and Behaviour Studies

Activity Theory Based Approach for Requirements Analysis of Android Applications Kandiraju Sai Ashritha(B) , T.M. Prajwala, and K. Chandrasekaran Department of Computer Science and Engineering, National Institute of Technology, Surathkal, Karnataka, India {14co121,14co133}@nitk.edu.in, [email protected]

Abstract. This paper aims at providing a detailed explanation on the necessity of an alternative approach based on Activity theory for the requirements analysis of a restaurant automation application. In the recent past, android platform has turned out to be one of the most userfriendly platforms for the development of application software. Also, the number of devices on which android applications can be used outrun the other platforms. Automation is one such domain which takes advantage of this platform. One such scenario is the application of automation in restaurants. This helps in efficiently reducing manpower, improving efficiency, accuracy and quality of the system. However, the applications developed in this domain fail to meet all of an average customer’s requirements. Thus the requirements analysis phase is critical to the development of such applications. The traditional methods of requirements engineering do not ensure that a majority of the requirements are captured and hence turn out to be unsuitable for restaurant automation application. Thus, in this paper, the use of activity theory for requirements analysis has been proposed for capturing the non-functional requirements which play a major role in the evaluation of performance characteristics of the system. Keywords: Activity theory · Restaurant automation analysis · Non-functional requirements

1

·

Requirements

Introduction

Software Engineering process can be broadly distributed into four main stages: (i) Requirements Engineering, (ii) Design and development, (iii) Testing, verification and validation and (iv) Deployment and Management [1]. Although, all of the above phases are essential for the development of the software, requirements engineering plays a crucial role. The outcome of the Requirements engineering phase is a compilation of the user and system requirements. Hence, it is necessary to ensure that this stage is carried out in a systematic fashion. This phase of c Springer International Publishing AG 2017  L. Uden et al. (Eds.): KMO 2017, CCIS 731, pp. 3–15, 2017. DOI: 10.1007/978-3-319-62698-7 1

4

K.S. Ashritha et al.

analysis might seem no more complicated than the expansion of the information obtained from the customer during inception and elicitation, but there maybe a number of possible issues that might arise due to the change in requirements and customer views over time. Thus, it is necessary to ensure that the requirements analysis is carried out carefully with utmost patience as this is the only opportunity for the developer to directly communicate with the customers [2]. Even a small error that might be incurred during the analysis phase will lead to serious issues during the development and testing phases. If this occurs, the developer will have to deal with problems of customer dissatisfaction and it might become inevitable to develop his project all over again. In other words, Requirements Analysis is the direct reflection of the success of the project [3]. The requirements for the development of any software belong to three major categories namely: Functional requirements - These requirements describe the general behavior of the system in relation to the system functionality. They provide an insight into the the goals the system aims at achieving. Non-functional requirements - These requirements are the constraints on the functional behavior of a system. While the functional requirements describe what the system should do, the non-functional requirements elaborate on how well the system will do so. They maybe broadly classified into three categories namely, product requirements, organizational requirements and external requirements. Domain requirements - These requirements are constraints from the operational domain. The classical methods of requirements engineering consider only the structural and behavioral aspects of the system and the need for performance aspect of the requirements is often overlooked at this stage. The reason being, these requirements come into the picture only when the users directly interact with the software. These requirements play a major role in meeting the expectations of the customers, as they involve the “human aspects” while dealing with the software [4]. This condition can be overcome by viewing the system from a social perspective, unlike a pure technical or a procedural context. However, figuring out organizational requirements seems to be a difficult task as they are not easily predictable or observable. Thus, these additional requirements that deal with the way humans handle software, stress the need to shift our focus from fixed requirements to those which are of a more dynamic nature. The current world scenario demands a highly frequent usage of mobile/android applications. Although the user requirements in the name of structural and behavioral aspects are entirely captured, the complete user satisfaction might not be achieved after the deployment. This is due to any compromise in non-functional requirements which directly affects the sole purpose of the software development. Since no definite approach for the complete analysis of non-functional requirements exists as of yet, there is a dire need of a new alternative for the same [8]. We attempt to solve this with the usage of Activity theory in this context.

Activity Theory Based Approach for Requirements Analysis

5

Section 2 explains the past work done in the field of activity theory. Section 3 provides details about bridging the gap between the past works. Section 4 gives a description of the restaurant automation android application. It also provides details regarding the motivation behind the choice of activity theory for the requirements analysis of the problem at hand. Section 5 gives a brief introduction to knowledge management and its relevance to the context. Section 6 contains the proposed solution. Section 7 provides implementation details of the android application and Sect. 8 applies the principles of activity theory to the application. Section 9 concludes the paper and Sect. 10 provides a scope for future work.

2

Literature Review

The basic unit of Activity theory is “human activity”. An activity is simply an interaction between users(actors) and the environment. The main concept of the theory is that every human activity is judged or in other words mediated. 2.1

History

Activity theory was first proposed by Vygotsky in the year 1978. His idea back then, was that every human activity is mediated which results in changes in its outcomes. However, his theory focused on an individual rather than a user community. This was overcome by Engestroms idea which treats the tools or artifacts as components that affect the functioning of human activities. According to him, mediation is a result of the entire components of the system and is not restricted to a single individual [9]. He further specified that an activity cannot be isolated. An activity is taken up (performed) by a subject, going through series of mediation as suggested by tools to achieve certain objects (objectives) which are then transformed to outcomes. Engestrom also put forward the basic structure of an activity system as illustrated in Fig. 1. It is also known as Engestroms activity system. According to his theory, an activity system is a collection of elements, the relationships that exist between the elements and the mediating artifacts (tools). The third generation of the activity theory considers a joint activity system as the basic unit of analysis [6]. Also they specify that an activity system always strives for evolution with time (Fig. 2). 2.2

Activity Theory Diagram

The components of the activity system are as follows: 1. Subject - Subject may be an individual or a group of individuals. A subject may also be called an actor. An activity is undertaken by the user. Relationship between the subject and object is mediated by an artifact or tool. 2. Object - An object is performed by the subject to result in an outcome after transformation.

6

K.S. Ashritha et al.

Fig. 1. Activity system structure

Fig. 2. Hierarchy of activities

3. Tools/Mediating artifacts - A subject interacts with an object through a tool. A tool/artifact is anything which mediates subjects actions. These may have existed from before or may have been created and transformed during the process of activity development. 4. Rules - The relationship between the subject and community has been established with the help of a set of well defined rules/norms. 5. Community - Community is a group of users sharing the same objective. 6. Division of Labor - The relationship between community and object is mediated by division of labor. This basically describes how work related to the activity is divided among the members of the community. 2.3

Activities, Actions and Operations

It has been proposed that activities consist of a three level hierarchical structure comprising of activities, actions and operations [10]. An activity corresponds to a motive, an action corresponds to a goal and an operation corresponds to a condition respectively. The dynamic nature of activity theory comes into picture during this decomposition of activities into different levels. The motives of an activity system tend not to undergo any changes, whereas the changes in conditions or goals, affect changes in operations and actions respectively [7]. The three levels of activity in a system contain well defined relationships between them and are not considered as separated and individual identities. 2.4

Contradictions

Engestrom defines contradictions or conflicts as those conditions that create a tension in an activity system. Contradictions are broadly classified into four types: (1) Primary contradictions: These arise within an element of an activity system, (2) Secondary contradictions: These occur between different elements of the same system, (3) Tertiary contradictions: These occur between objects and motives, (4) Quaternary contradictions: These occur between a central activity system and its neighboring systems [4]. These instabilities and contradictions result in the evolution of the activity theory with as minimum controversies as possible.

Activity Theory Based Approach for Requirements Analysis

2.5

7

Application to Requirements Engineering

Most of the problems solved psychologically, take human action as the basic unit of analysis in most of the real life situations. But human action should be analyzed into a context so that it is completely understood. This is the reason why we define action as being analyzed inside an activity and present activity as the basic unit of analysis. This is how activity theory is closely related to our real life situations [1]. Also, the different percepts of activity theory related to human aspects, support the use of the theory in requirements analysis of the software.

3

Gap Analysis

Activity theory is an efficient technique in splitting a task at hand such as an activity into sub-tasks which include activities and operations. The current theory does not take into account the method of predicting the next activity after the current activity. The advantages of incorporating the predictability concepts include minimizing the probability of occurrence of cases in which the software might fail if a particular precautionary measure is not taken care at the current stage. Another added benefit is that the software functionality will seemingly become more logical and take the right path of execution with minimum failure. Thus, the future scope of improvement of the activity theory concept lies in this area.

4 4.1

Problem Statement Description

This paper considers the Restaurant Automation android application as a case study. The motive behind the design of the software is to introduce automation in restaurants and reduce the number of employees and their workload. Automation of the operations in the restaurants will make the entire process time and cost-effective, as the orders given by the customers will reach the kitchen instantaneously. The customer experience, will also be enhanced as they would be given more control over their order. An increase in the revenue from the younger generations can also be expected as they find the use of technologies more convenient and adaptive. In the traditional system of practice, the customer is not always given the privilege to make a choice regarding the table, and is expected to wait patiently till the waiter arrives and writes down the order manually using pen and paper. Inevitable human errors can also be made by the waiters. These pitfalls in the conventional system along with the motive to improve customer satisfaction have driven the need for automation. Our application automates most of the services in a restaurant. The application replaces the conventional methods that are followed in a restaurant, with on click features like the customer being able to select a table provided on the user interface and choosing food items from the menu available on the phone screen,

8

K.S. Ashritha et al.

being able to enter his feedback and suggest further improvements. Through the application, the only functions of the chef would be to receive an order notification from the customer and to notify the waiter after the food is ready to be served. Similarly, the functions of the waiter would be to constantly check and clear tables and to deliver the food to the appropriate table on receiving a notification from the kitchen staff. In this way, the application helps in minimizing the work load on the restaurant staff while co-ordinating their work, apart from saving the customers time and providing him with efficient services. The subsequent section describes the need to have an activity theory based requirements analysis approach pertaining to the android application. 4.2

Motivation

As stated earlier, though the traditional methods of elicitation and analysis capture the structural and behavioral aspects of requirements excellently, we need a better method that also includes the non-functional requirements. The ‘softer or ‘human aspects of the requirements come into picture only when the users actually handle the software. And if these requirements are overlooked during the analysis phase, though the user would have an application as per his structural requirements, when deployed, he might still dislike the software while using it, which eventually affects the usage of the application. More importantly, the interaction between android applications and their users is enormous. So, for this purpose we should not only take care of the look and feel of the application but also about the other problems that a user might face during the usage of the software. This paper proposes the use of activity theory for requirements analysis. The reasons for choosing activity theory as an alternative are: (i) We get a clear picture of the objective that the user intends to achieve using the software. Also, by mapping the functional requirements that were gathered during the requirements elicitation phase to systematic models, we develop a greater insight into how the different components of the model impact the achievement of the user objective. (ii) According to Leont’ev [13], every activity is driven by a human motive, which differentiate unique activities. He also stated that “There can be many actions which seek to reach the same goal. Operations determine the way to execute the actions which are performed to meet goals”. Thus, the concept of hierarchy in activity theory divides every activity into three levels namely: activity, action and operation. This division helps in shifting our focus from the completion of larger tasks to sub-tasks. This makes sure that we have made a note of all the requirements at each level, and to be more precise, at each task. Successful completion of all the actions and operations will indirectly mean the successful completion of an activity. (iii) Contradictions or Conflicts in activity theory reflect the needs or requirements of different customers and help us prepare a consolidated list of requirements. Contradictions also help an activity system in evolving from time to time and become better.

Activity Theory Based Approach for Requirements Analysis

5

9

Knowledge Management

The success of the development of any application software depends primarily on the effectiveness of the requirements engineering phase. A requirement is a software capability that must be provided for the achievement of the domainspecific goals. Hence, the benefits of the product under development depends entirely on the capabilities it can provide within the domain under consideration. This underlines the importance of the application domain in requirements engineering. Every domain comprises of certain formal and informal aspects. The informal part is termed as tacit knowledge, which is characterized by understandability and implicitness. This part of the domain knowledge is assumed to be implicitly understood by the receivers and hence, not explicitly conveyed which makes it hard for being captured during requirements engineering. The domains with a high degree of informality are called Informal Structured Domains (ISD). The increase in development time arising due to the ambiguity in the requirements engineering phase can be attributed to the tacit knowledge, and is not desirable. Thus, it is critical to convert this tacit knowledge into some sort of an explicit knowledge, which highlights the purpose of knowledge management in requirements engineering [12]. The requirement engineers should streamline their process towards elicitation and the features of the software being developed. This domain knowledge, which mainly comprises of the tacit knowledge has to be conveyed by the clients and users using the natural language to the requirement engineers. The two way knowledge transfer between the requirement engineers and domain specialists is subject to the interpretation, perspectives and pre-requisites of the receivers. Knowledge management is thus used for identification and capture of the tacit knowledge which helps in its formalization, which in turn, reduces the ambiguity and complexity of these requirements. A suitable knowledge management strategy can be applied during the development process [12].

6

Solution

The elicitation phase of requirements engineering cannot be skipped as this is the first step in capturing the functional and behavioral requirements. This is achieved by conducting one on one interviews, surveys, distributing questionnaires, and receiving feedback from the potential users of the application. The next step is to analyze the elicited requirements by categorizing them into functional and non-functional requirements. To cite an example, when a user specifies that he wants an application to order food without any delays, this implies that the functional requirement that should be considered is the food ordering feature and the non-functional requirement corresponding to it is the responsiveness. At the end of this knowledge transfer, we thus prepare a list of the most common requirements. The basic functionalities like view menu, send order, receive order, send notification, receive notification, pay bill and provide feedback and performance requirements like accessibility, efficiency, and responsiveness are obtained.

10

K.S. Ashritha et al.

However, we do not obtain the exhaustive list of the requirements. Also, the exact mapping between the corresponding functional and non-functional requirements cannot be figured at this stage. Hence, the application of activity theory justifies the purpose [5]. For the proper understanding of all the human aspects involved with the application, we divide the analysis into three major steps. (i) The first step is to map the elicited functional requirements to the activities of the system and construct the activity theory diagrams for the same. (ii) The next step is to determine the various components of the activity theorem diagrams. (iii) The final step of the analysis would be analyzing each activity in greater detail by decomposing it into levels of activity, actions and operations. We have to bear in mind that tackling requirements for operations and actions of each activity, results in the complete requirement analysis of that particular activity.

7

Implementation

The restaurant automation android application supports three types of users namely-customers, waiters and the kitchen staff. Automation can be achieved with the collaboration of all the three users. Each of these users are required to register and login with their credentials for gaining access to the application. At the time of registration, the user should specify if he/she is a customer, waiter or a chef. The user interface provided is subject to the type of the authenticated user. The application also provides a guide to help the users with the usage of the application. The different aspects of the software can be handled by clicking on various buttons in touch enabled android-based mobile phones and tablets. A database was made use of for handling the collected information about the users. The database used for holding the authentication details of users, information regarding the availability of the tables, storage of the menu, bill payment details and user feedback and rating was hosted on an on-line server. The android application thus, requires the use of Internet facility for connecting to the server. For example let us suppose that a user logs in as a customer. Now, he should first pick a table of his choice from the available options. He can choose a table only if at-least one of the tables is free. The availability information of the tables is stored in the table database on the server side. The open-source software tools used for the development include Android Studio, PHP, and MySQL.

8

Case Study

This paper deals with the Restaurant Automation android application, which aims to achieve automation in restaurants. We need to carry out the procedure specified in the solution. The main actors (users) involved in the application are the customer, the chef and the waiter. The main activities of the customer

Activity Theory Based Approach for Requirements Analysis

11

would be to select a table of his choice, order food items by choosing from a menu provided on the user interface, to send a notification to the chef which contains a list of the ordered items and finally to pay the bill after the bill is generated. The main activities of the chef would be to receive a notification from the customer and to send a notification to the waiter when the food is ready to be delivered. Similarly, the main activities of the waiter would be to receive a notification from the chef and also to uncheck a table after the customer leaves the restaurant. The three types of subjects considered are customer, chef and waiter. Their outcomes and objects vary, while the focal mediator is the android application which remains constant. The community consists of customer, chef, waiter, system administrator and restaurant manager who share the same object of achieving automation. While the subjects interact with the mediator to achieve the object, the rules govern the interactions of the system. From the definition, the non-functional requirements impose constraints on the system to achieve better performance. This portrays that the rules governing the operations are analogous to the non-functional requirements corresponding to that activity. We first analyze the activities taking customer as the subject. In our case study, there are four activities for customer namely View menu, Send Order, Pay bill and Provide feed- back, as shown in Fig. 3, Fig. 4, Fig. 5 and Fig. 6 respectively. The non-functional requirements that can be deduced from the rules in Fig. 3 are secure connectivity, privacy and accessibility, and from Fig. 4 are accessibility, disaster recovery, fault tolerance, efficiency, responsiveness and back-up. Those that can be inferred from Fig. 5 include accuracy, efficiency, security and fault tolerance whereas those obtained from Fig. 6 are accessibility and interactive user interface (Table 1).

Fig. 3. View Menu

Fig. 4. Send Order

Next we take into account the activities of the chef as the subject. In our case study there are two activities for chef namely Receive Order and Send notification as shown in Fig. 7 and Fig. 8 respectively. The non-functional requirements that can be deduced from Figs. 7 and 8 are accessibility and interoperability. The chef waits for the order from a customer. Once the food is ready to be served he sends a notification to the waiter (Table 2).

12

K.S. Ashritha et al.

Fig. 5. Pay Bill

Fig. 6. Providing Feedback

Table 1. Activity Decomposition for Customer Activity

Actions

Operations

View menu

Signup Login Select table

Customer enters the required details and clicks on SIGNUP/LOGIN and makes a table choice

Send order

Select items Specify quantity

Customer chooses food items, specifies quantities and clicks on SEND ORDER

Pay bill

Request for total amount Send order

Customer can order more after viewing amount or click on DONE

Provide feedback Decides to provide feedback

Fig. 7. Receiving order

Customer enters feedback and rates the restaurant and clicks on SUBMIT or discard feedbacks by clicking on CANCEL

Fig. 8. Sending notification

Finally we take into account the activities of the waiter as the subject. In our case study there are two activities for waiter namely Receive notification and Clear tables as shown in Figs. 9 and 10. The non-functional requirements corresponding to the rules in Fig. 9 include accessibility, fast communication and interoperability. Also, those that can be inferred from Fig. 10 are accessibility, disaster recovery, fault tolerance, efficiency, responsive, back-up and interactiveness (Table 3).

Activity Theory Based Approach for Requirements Analysis

13

Table 2. Activity Decomposition for Chef Activity

Actions

Operations

Receive order

Customer checks menu Click send order

Order food Order sent to database

Send notification click on SEND ORDER Order sent Notification sent

Fig. 9. Receiving notification

Fig. 10. Clear table

Using the above approach we successfully modularized our case study application into a number of small modules each of which is dealt with individually. By spending ample time for the requirements elicitation and analysis of each of these modules we have ensured that all the human aspects of the software are covered. In this way, we make sure that we capture all the organizational requirements pertaining to a single module and repeating this process for all the modules gives us the organizational requirements that are specific to the activity. Therefore, by applying an activity theory based approach for the Restaurant Automation android application, we can ensure that almost all the requirements are dealt with and hence we can reach up to the expectations of the user. Table 3. Activity Decomposition for Waiter Activity

Actions

Operations

Receive notification Customer checks menu Clicks on SEND ORDER Order received by chef Notification sent Clear table

Customer pays bill Checks used tables

Clicks on PAY BILL Clears tables in database

14

9

K.S. Ashritha et al.

Conclusion

In this paper, we have made use of the theoretical concepts of activity theory, in order to apply them for the analysis of requirements for android applications, which take into account the organizational requirements that deal with social and human aspects of requirements engineering process. Thus, we have suggested the use of activity theory as an alternative for the requirements analysis process of android applications in the paper. This paper has successfully captured the organizational requirements for the Restaurant Automation android application and has thus proven the practicality of the theory in terms of requirements analysis.

10

Future Work

Although activity theory is shown to be one of the best techniques in capturing requirements for android applications, it has its own set of limitations. One of the main limitations is that a thorough understanding of the functioning of activity systems and their activities pertaining to the specific problem or situation at hand is required. The ability to visualize the dynamic interaction that takes place in the systems is also necessary and this may be acquired only with intense research. Another difficulty faced by most of the research scholars is the segregation of the system into different levels namely activity, actions and operations. These can be considered for future work in this approach for requirements analysis. However, the benefits of application of the theory to various areas of software engineering certainly over-weigh its limitations.

References 1. Boehm, B.W.: Software engineering. IEEE Trans. Comput. 25(12), 1226–1241 (1976) 2. Nuseibeh, B., Easterbrook, S.: Requirements engineering: a roadmap. In: Proceedings of the Conference on the Future of Software Engineering, pp. 35–46, Limerick, Ireland, 04–11 June 2000 3. Kujala, S., Kauppinen, M., Lehtola, L., Kojo, T.: The role of user involvement in requirements quality and project success. In: Proceedings of the 2005 13th IEEE International Conference on Requirements Engineering (RE05) 4. Uden, L., Valderas, P., Pastor, O.: An activity theory based model to analyse web application requirements, Staffordshire University, The Octagon, Beaconside, Stafford, ST18 OAD, UK, vol. 13(2), pp. 1–24 (2008) 5. Engestrm, Y.: Expansive learning at work: towards an activity theoretical reconceptualization. J. Educ. Work 14(1), 133–156 (2001) 6. Engestrm, Y.: Development studies of work as a testbed of activity theory. In: Chaiklin, C., Lane, J. (eds.) Understanding Practice: Perspectives on Activity and Context, pp. 64–103. Cambridge University Press, Cambridge (1993) 7. Korpela, M., Mursu, A., Soriyan, H.A.: Information systems development as an activity. Comput. Support. Coop. Work 11, 111–128 (2002)

Activity Theory Based Approach for Requirements Analysis

15

8. Martins, L.E.G.: Activity theory as a feasible model for requirements elicitation processes. Sci. Interdisc. Stud. Comput. Sci. 18(1), 33–40 (2007) 9. Uden, L.: Activity theory for designing mobile learning. Int. J. Mobile Learn. Organ. 1(1), 81–102 (2007) 10. Uden, L., Willis, N.: Designing user interface using activity theory. In: Proceedings of the 34th Hawaii International Conference on System Science (2001) 11. Ahmad, N.A.N., Akhbariee, N.I., Hafizud-deen, M.: Requirements analysis of android application using activity theory: a case study. International Conference of Information and Communication Technology (ICoICT), pp. 145–149, pp. 20–22, March 2013 12. Snchez, K.O., Osollo, J.R.: A strategy to requirements engineering based on knowledge management. Mexican International Conference on Computer Science (2013) 13. Leontev, A.N.: Activity, Consciousness and Personality. Prentice-Hall, Englewood Cliffs (1978) 14. Kamaruddin, K.A., Yusop, N.S.M., Ali, M.A.M.: Using activity theory in analyzing requirements for mobile phone application. In: 2011 5th Malaysian Conference in Software Engineering (MySEC). IEEE (2011)

Machine Consciousness of Mind Model CAM Zhongzhi Shi1 ✉ , Gang Ma2, and Jianqing Li3 (

1

)

The Key Laboratory of Intelligent Information Processing, Institute of Computing Technology, Chinese Academy of Sciences, No. 6, Kexueyuan Nanlu, Zhongguancun, Beijing, China [email protected] 2 University of Chinese Academy of Sciences, Beijing, China [email protected] 3 Shandong University, No.27, Shan Da Nan Road, Jinan City, Shandong Province, China [email protected]

Abstract. Mind model consciousness and memory (CAM) is a biologically inspired mind model, which is a general framework for human-level intelligent systems. Consciousness module plays very important role in mind model CAM. This paper presents architecture of machine consciousness containing awareness, attention, motivation, metacognition, introspective learning and global work‐ space modules. The proposed machine consciousness will be applied in cyborg intelligence for action planning. Keywords: Intelligence science · Machine consciousness · Mind model CAM · Cyborg intelligence

1

Introduction

The theoretical and technical foundations of intelligence are derived from the study on brain-inspired computing. The study on intelligence is named as intelligence sciences, which is an interdisciplinary subject that dedicates to joint research on basic theory and technology of intelligence by brain science, cognitive science, artificial intelligence and others [16]. Intelligence Sciences is very important as it reveals the basic theory about the intelligence and gives explanations on the intelligent activities according to the intelligence theories. Furthermore, the intelligence science is the director that guides the development of brain-inspired computers and intelligent robots. One of the main issues in intelligence science is mind modeling which tries to find ways to model the human mental activities, such as perception, learning, memory, thinking, consciousness etc. The mind refers to the aspects of intellect and consciousness involves the combination of the brain’s conscious and unconscious cognitive processes including thought, perception, memory, emotion, will and imagination etc. The mind modeling is the process of building cognitive architectures which specify the underlying infrastructures for intelligent systems. It is also the foundation for creation and under‐ standing of synthetic agents that support the same capabilities as humans. We have proposed a new mind model named Consciousness And Memory Model (CAM) for general framework of human-level intelligent systems [17] © Springer International Publishing AG 2017 L. Uden et al. (Eds.): KMO 2017, CCIS 731, pp. 16–26, 2017. DOI: 10.1007/978-3-319-62698-7_2

Machine Consciousness of Mind Model CAM

17

The outstanding characteristic of CAM is mainly based on consciousness and memory, which is different with previous cognitive models, such as Soar, ACT-R and so on. The origin and essence of consciousness is one of the most important scientific problems. From the view with intelligence science, consciousness is the experience integration with external world, one’s own body and mental process. Farber and Churchland discussed the consciousness concept from three levels in the article “Consciousness and the Neurosciences: Philosophical and Theoretical Issues” [10], that is, sense of awareness, high-level ability and consciousness state. In 2012, Baars and Edelman explained their natural view about consciousness in the article [3], and listed the 17 characteristics of the state of consciousness. So far a lot of conscious‐ ness theories are proposed, such as Global Workspace Theory (GWT) [2], reductionism [7], theory of neuronal group selection [9], quantum theory [6, 15], information inte‐ gration theory [22]. In this paper we propose an architecture of machine consciousness in CAM from the view of intelligent system engineering. Machine consciousness refers to attempts by those who design and analyse infor‐ mational machines to apply their methods to various ways of understanding conscious‐ ness and to examine the possible role of consciousness in informational machines [1]. Franklin and Baars proposed a cognitive model LIDA with the functions of conscious‐ ness as identified by Baars’ Global Workspace Theory [11]. Sensorimotor learning helps an agent properly interact with its environment using past experiences in LIDA [8]. The next section describes the CAM model and its relationship to the machine consciousness. Section 3 contains an overview of machine consciousness. The execution of machine consciousness is described in Sect. 4 in detail. Section 5 introduces the simulation of a specific cyborg maze search. Finally, the conclusions and future works are given.

2

The CAM Model

In the mind activities, memory and consciousness play an most important role. Memory stores various important information and knowledge; consciousness make human having the concept of self, according to the needs, preferences based goals, and do all kinds of cognitive activity according to memory information. Therefore, the main emphasis on mind model CAM are memory functions and consciousness functions [18]. Figure 1 shows the architecture of the mind model CAM. CAM includes five main parts which are memory, consciousness, high level cogni‐ tive functions, perception and motor action. Each part plays a different role in cognitive process. Memory is responsible for storing cognitive information. Consciousness is a central controller which controls the interaction among memory components. Further‐ more, consciousness reads and writes the information stored in memory to perform cognitive activities. Consciousness cooperated with memory to give the basic cognitive function supports to high level cognitive functions. High level cognitive functions perform high level cognitive activities such as video understanding, action planning, problem solving etc. These high level cognitive activities are the foundations realizing the cognitive based intelligent applications, such as automatic surveillance, game

18

Z. Shi et al.

Fig. 1. Architecture of mind model CAM

playing, robotic house keeper etc. A brief introductions to each part of CAM are given in the following. Memory is one of the basic human brain functions. It stores the common knowledge about the world, reflects the things took place in the past and keeps the learned behaviors that help human to interact with the environment. Because of the existence of memory, human can reason logically through common knowledge, make present reactions on the basis of the past, and evolve from the past experiences which lead the reflections of environment deeper and more comprehensive. The information stored in Memory can be categorized into difference classes, for example common knowledge based informa‐ tion, past experience information, state belief information, and observation based infor‐ mation. In CAM, three memory components are proposed to store different types of information. These components are Long term memory, short term memory and working memory.

3

Architecture of Machine Consciousness in CAM

Figure 2 shows the machine consciousness in CAM. It consists of global workspace, awareness, attention, motivation, metacognition, introspective learning modules. 3.1 Awareness Module Awareness module begins with the input of external stimuli, and the primary features of the sensing system are activated. The output signal is sent to the sensory memory, where a higher level of functional detectors are used for more abstract entities, such as objects, categories, actions, events, etc. The resulting perception is sent to the workspace, where local connections short episodic memory and declarative memory will mark thread. These local associations are combined with perception to produce the current

Machine Consciousness of Mind Model CAM

19

Visual

Metacognition Module

Motivation Module

Working Memory

Semantic Memory

Action Plan

Smell

Global Working Space

Attention Module

Touch

Awareness Module

Auditory Environment

Introspection Learning Module

Episodic Memory Procedural Memory

Fig. 2. Architecture of machine consciousness in CAM

situation model, which is used to represent the understanding of the current events that are happening. In CAM, awareness is basically a perception combination of detect sensation. Agents work effectively in complex environments, and a subset of these combinations must be selected as the value of perception. 3.2 Attention Module Detection of new events is an important feature of any signal classification method. Because we are not able to train all the data that may be encountered in the machine learning system, it becomes very important to distinguish known and unknown object information in the test. Novelty detection is a very challenging task, which can be found in a complex, dynamic environment of the novel, interesting events. Novelty detection is an essential requirement for a good classification or recognition system, because sometimes the information contained in the test data is not known when the training model information is included. The novelty of awareness is related to cognition, and the novelty of cognition is related to knowledge. Based on a fixed set of training samples from a fixed number of categories, novelty detection is a dual decision task for each test sample to determine whether it belongs to a known classification or not. 3.3 Global Workspace Module The global workspace module is in the working memory area, in which different systems can perform their activities. Global means that the symbols in this memory are distrib‐ uted and passed through a large number of processors. Of course, each processor may have a number of local variables and run. But it is very sensitive to the symbol of the overall situation, and the information can be made in a timely manner. When faced with new and different things that are different from the habit, our senses will produce an orienting reaction. At the same time, all kinds of intelligent processors will display their

20

Z. Shi et al.

new things on the cognitive analysis scheme by way of cooperation or competition in the global workspace, until you get the best results. It is in this process that we have a sense of new things. The global workspace can be seen as a blackboard system of infor‐ mation sharing, through the use of the blackboard, each processor tries to spread the information of the global situation, the joint establishment of the problem solving approach. The internal structure of the work area is constituted by a variety of input buffer and three main modules: current scenario mode, register and the queue of awareness contents. The current scenario model is a structure that stores the current internal and external events that represent the reality. The construction encoder is responsible for creating the structure of the elements in each seed model using the work area. Register in the auxiliary space, in here construction encoder can construct possible structure, and then transfer them to the scenario mode. The queue of the consciousness content stores the contents of the continuous broadcast, which makes the CAM model understand and operate the concepts related to time. The competition of the global working space selects the most outstanding, the most relevant, the most important and the most urgent affair, their content becomes the content of consciousness. Then, the contents of the consciousness are broadcasted to the whole space, and the action selection phase is initiated. 3.4 Motivation Module Motivation could be represented as a 3-tuples {N, G, I}, where N means needs, G is goal, I means the motivation intensity. A motivation is activated by motivational rules which structure has following format: R = (P,D,Strength(P|D))

where, P indicates the conditions of rule activation; D is a set of actions for the moti‐ vation; Strength(P|D) is a value within interval [0,1]. At present CAM is going to apply to animal robot which is a brain-computer inte‐ gration system. All behaviors of brain-computer integration stem from a fixed and finite number of needs. According to characteristics and requirements of brain-computer inte‐ gration there are 3 types of needs, that is perception, adaptation and cooperation: (1) Perception needs: Acquire environment information through vision, audition, touch, taste, smell. (2) Adaptation needs: Adapt environment condition and optimize impaction of action. (3) Cooperation needs: Promise to reward a cooperation action between brain and machine. 3.5 Metacognition Module In mind model CAM, metacognition provides the cognition and monitoring of thinking activity and learning activity of the agent, which the core is knowledge about cognition and control of cognition. Metacognition module has the function of metacognitive

Machine Consciousness of Mind Model CAM

21

knowledge, metacognitive self-regulation control and metacognitive experience. Meta‐ cognitive knowledge includes knowledge about the subject, the knowledge of the task, and the knowledge of the strategy. Metacognitive experience refers to the experience of their own cognitive process. In cognitive process, through the metacognitive self-regu‐ lation control, select the appropriate strategy to realize the use of strategy, the compar‐ ison of process and goal, the adjustment of the strategy and so on. 3.6 Introspective Learning Module Knowledge base construction of introspective learning module uses ontology tech‐ nology based on the general introspective learning model. The classification problem of failure is an important problem in introspective learning. The classification of failure is the basis of the diagnostic task, and it provides important clues to explain the failure and to construct the correct learning objectives. Two important factors of failure classifica‐ tion should be considered, one is the granularity of failure classification, the other is the relationship between failure classification, failure explanation and introspective learning goals. Ontology based knowledge base is the combination of ontology based knowledge representation and expert system knowledge base, which has the advantages of concep‐ tual, formal, semantic, and sharing. By using the method of ontology based knowledge base to solve the failure classification problem of introspective learning, the failure classification will be clearer and retrieval process more effective.

4

Execution of Machine Consciousness

In mind model CAM the execution of machine consciousness can be divided 3 phases: awareness, motivation and action plan. Awareness phase is the process of attaining perception or understanding of the environment by organizing and interpreting sensory information. Using the incoming percept and the residual contents of working memory, as cues, local associations are automatically retrieved from transient episodic memory and from declarative memory. Motivation phase focuses on learners’ beliefs, expecta‐ tions, and needs for order and understanding. According to the impact factors of moti‐ vation, such as proportional activation, opportunism, contiguity of action, persistence, interruption, combination of preference we construct a motivation subsystem. Action plan will compose a group of actions through action selection, planning to reach the end goal. 4.1 Awareness Awareness is the state or ability to perceive, to feel, or to be conscious of events, objects or sensory patterns. In this level of consciousness, sense data can be confirmed by an observer without necessarily implying understanding. More broadly, it is the state or quality of being aware of something. In biological psychology, awareness is defined as a human’s or an animal’s perception and cognitive reaction to a condition or event.

22

Z. Shi et al.

Awareness is commonly viewed as an information path connecting to the environ‐ ment and understanding outside situations. We have proposed a conditional random fields based feature binding computational model [19], coding and combining features model [20], deep convolutional generative stochastic model [12] and so on. All these methods try to find what these objects are and where locate in from environment scenario. 4.2 Motivation Motivation is defined by psychologists as an internal process that activates, guides, and maintains behavior over time. Mook [14] defined motivation as “the cause of action” briefly. Maslow proposed hierarchy of needs which was one of first unified motivation theories [13]. Since it introduced to the public, the Maslow’s hierarchy of needs theory has been made a significant impact to the every life aspect in people’s life. Maslow actually was a humanistic psychologist who believed in the human potential that human can struggle to reach the success and look for the creativity in order to reach the highest wisdom and also the logic think. Bach has proposed a framework for an extensible motivational system of cognitive agents, based on research in psychology [4]. They developed a version of the model which has been successfully evaluated against human performance in problem solving games [5]. In CAM the structure of motivation module is shown in Fig. 3, which consists of 7 components: environment, internal context, motivation, motivation base, goal, action selection and action composition. Their main functions are explained as follows: (1) Environment provides the external information through sensory devices or other agents. (2) Internal context represents the homeostatic internal state of the agent and evolves according to the effects of actions. (3) Motivation is an abstraction corresponding to tendency to behave in particular ways according to environmental information. Motivations set goals for the agent in order to satisfy internal context. (4) Motivation base contains a set of motivations and motivation knowledge with defined format. (5) Goal is a desired result for a person or an organization. It used to define a sequence of actions to reach specific goals. A goal list consists of a number of goals which can be described formally:

Gt = {Gt1 , Gt2 , … , Gtn } (6) Actions selection is used to perform motivated action that can satisfy one or several motivations. (7) Action composition is the process of constructing a complex composite action from atomic actions to achieve a specific task.

Machine Consciousness of Mind Model CAM

23

Fig. 3. The structure of motivation module

5

Experiment

An actual simulation application in the cyborg rat-robot maze search in Fig. 4 will be provided here in order to significantly demonstrate the feasibility and validity of machine consciousness [21]. The following will mainly represent the actual design of the ratrobot agent with machine consciousness. The Cyborg sensory inputs (here mainly visual information) constantly trigger the awareness module, and convert those environment information into the unified internal motivation signal which are transferred to action plan module. Then the action plan module will select proper actions to response the environment. The task of the rat-robot agent with machine consciousness is to start moving at the maze entrance, and finally reach the maze exit denoted as a red flag depended on all guideposts in Fig. 3. In order to fulfil the maze search task, the rat-robot agent should implement all the three basic modules, , , . In the rat-robot maze search experiment, the rat-robot agent is designed to have 4 basic motivation behaviors moving on, moving back, turning left and turning right in the maze. We construct a sub-MNIST dataset extracting 4 types of handwritten digits with flag’0’ denoting moving on,’1’ moving back,’2’ turning left and’3’ turning right from the orig‐ inal MNIST dataset. When rat-robot agent moves on the path, its sensory inputs constantly drive aware‐ ness module and generate the motivation signal. In the experiment, there are four moti‐ vation signals, moving on, moving back, turning left and turning right. Which means the agent can response four types of action plans.

24

Z. Shi et al.

Fig. 4. Cyborg intelligence for action planning in rat-robot maze search

6

Conclusions and Future Work

Mind model CAM is a biologically inspired cognitive model, which is a general frame‐ work for human-level intelligent systems. Consciousness subsystem plays very impor‐ tant role in mind model CAM. This paper presented the architecture of machine consciousness, which contains awareness, attention, motivation, metacognition, intro‐ spective learning and global workspace. Most of above modules have been designed and implemented. The proposed machine consciousness has been applied in cyborg intelligence for action planning. In final, the simulation of cyborg rat-robot maze search was demonstrated. We are going to develop other modules of machine consciousness in CAM. The global workspace theory will guide action selections and chose the most outstanding, the most relevant, the most important and the most urgent affair for execution. Explore

Machine Consciousness of Mind Model CAM

25

the global workspace theory and connecting with attention. Metacognition is also inter‐ ested in topic, which performs self-regulation control, selects the appropriate strategy to realize. We try to apply machine consciousness in robots, automatic car, unmanned plane in the future. Acknowledgements. This work is supported by the National Program on Key Basic Research Project (973) (No. 2013CB329502), National Natural Science Foundation of China (No. 61035003, 61202212), National Science and Technology Support Program (2012BA107B02).

References 1. Aleksander, I.: Machine consciousness. In: Velmans, M., Schneider, S. (eds.) The Blackwell Companion to Consciousness, pp. 87–98 (2007). ISBN:9781405160001 2. Baars, B.J.: A Cognitive Theory of Consciousness. Cambridge University Press, New York (1988). Available from University Microfilm, Ann Arbor, MI, USA 3. Baars, B.J., Edelman, D.E.: Consciousness, biology and quantum hypotheses. Phys. Life Rev. 9, 285–294 (2012) 4. Bach, J.: Principles of Synthetic Intelligence–An Architecture of Motivated Cognition. Oxford University Press, Oxford (2009) 5. Bach, J.: Modeling motivation in MicroPsi 2. Artif. Gen. Intell. 2015, 3–13 (2015) 6. Chalmers, D.J.: The conscious mind: In: Search of a Fundamental Theory. Oxford University Press, New York (1996) 7. Crick, F.: The Astonishing Hypothesis. Scribner, New York (1994) 8. Dong, D., Franklin, S.: Modeling sensorimotor learning in LIDA using a dynamic learning rate. Biol. Inspired Cogn. Archit. 14, 1–9 (2015). doi:10.1016/j.bica.2015.09.005 9. Edelman, G.M.: Biochemistry and the sciences of recognition. J. Biol. Chem. 279, 7361–7369 (2004) 10. Farber, I.B., Churchland, P.S.: Consciousness and the neurosciences: philosophical and theoretical issues. In: Gazzaniga, M.S. (ed.) The Cognitive Neurosciences, pp. 1295–1306. MIT Press, Cambridge (1995) 11. Franklin, S., Madl, T., D’Mello, S., Snaider, J.: LIDA: a systems-level architecture for cognition, emotion, and learning. IEEE Trans. Auton. Mental Dev. 6(1), 19–41 (2014) 12. Ma, G., Yang, X., Zhang, B., Qi, B., Shi, Z.: An environment visual awareness approach in cognitive model ABGP. In: 27th IEEE International Conference on Tools with Artificial Intelligence, pp. 744–751 (2015) 13. Maslow, A.H.: Motivation and Personality. Addison-Wesley, Boston (1954). 1970, 1987 14. Mook, D.G.: Motivation: The Organization of Action. W.W. Norton and Company Inc, New York (1987) 15. Penrose, R.: The Emperor’s New Mind. Oxford University Press, Oxford (1989) 16. Shi, Z.: Intelligence Science. World Scientific Publishing Co., Singapore (2012) 17. Shi, Z.: CAM is a General Framework of Brain-inspired Computing. Invited speaker. World Congress of Robotics, Shenyang, China (2015) 18. Shi, Z.: Mind Computation. World Scientific Publishing Co., Singapore (2017) 19. Shi, Z., Yue, J., Ma, G.: Visual Awareness in Mind Model CAM. Cognitive 2014, Venice, Italy, pp. 262–268 (2014) 20. Shi, Z., Yue, J., Ma, G., Yang, X.: CCF-based awareness in agent model ABGP. In: Dillon, T. (ed.) IFIP AI 2015. IAICT, vol. 465, pp. 98–107. Springer, Cham (2015). doi: 10.1007/978-3-319-25261-2_9

26

Z. Shi et al.

21. Shi, Z., Ma, G., Wang, S., Li, J.: Brain-machine collaboration for cyborg intelligence. In: Shi, Z., Vadera, S., Li, G. (eds.) IIP 2016. IAICT, vol. 486, pp. 256–266. Springer, Cham (2016). doi:10.1007/978-3-319-48390-0_26 22. Tononi, G.: Consciousness as integrated information: a provisional manifesto. Biol. Bull. 215, 216–242 (2008)

The Knowledge Communication Conceptual Model in Malaysian Public Sector Rohaizan Daud ✉ , Nor Zairah Ab Rahim, Roslina Ibrahim, and Suraya Ya’acob (

)

Advanced Informatics School, Universiti Teknologi Malaysia, Kuala Lumpur, Malaysia [email protected], {nzairah,iroslina,suraya.yaacob}@utm.my

Abstract. Planning for Information and Communication Technology (ICT) future in the context of Malaysian public service is in the hand of group of IT experts and decision makers. Both parties are responsible for making decisions and plan the future ICT direction of the organization. Therefore it is important to comprise common understanding between Information Technology (IT) experts and decision makers so it will lead to a better and comprehensive decision making. However, there are still lacks of studies to describe the parties’ potential to under‐ take the situation from the perspective of the knowledge integration of among IT experts and decision makers especially in term of communication. Therefore, it is an essential to develop further understanding on the knowledge communication idea. It is done by integrating the concept of both knowledge sharing and knowl‐ edge transfer. From the literature analysis, the study found knowledge sharing framework and knowledge transfer framework might influence to the knowledge communication model. Therefore further study is recommended to validate the knowledge communication conceptual model in the real organization setting. Keywords: Knowledge communication · Knowledge sharing framework · Knowledge transfer framework · Organization

1

Introduction

Knowledge Management (KM) has widely covered the importance of knowledge in the organization. One of the important elements within the KM is about the mediator of knowledge in organization. This has become more essential since the activity between experts and decision makers such as decision making, sense making, problem-solving and strategizing are using face-to-face communication as a knowledge mediator. By having an effective way of communication, it can help to diffuse knowledge within the organization, prevent redundancies and create new knowledge by exchanging existing knowledge [1]. Therefore, due to the current trend of digitalization, the organization should dynamically utilize the web services technology for better way of communication especially between experts and decision makers. In order to do that, this study intends to develop knowledge communication conceptual model that can practically apply into Malaysian public sector.

© Springer International Publishing AG 2017 L. Uden et al. (Eds.): KMO 2017, CCIS 731, pp. 27–38, 2017. DOI: 10.1007/978-3-319-62698-7_3

28

R. Daud et al.

In this current digital society, knowledge has become one of the main resources in all organizations. Wise decisions were prepared based on knowledge gained from reli‐ able and competent sources. Knowledge also considered as resource to organization and a competitive advantage. Thus, the success and outcomes of an organization is closely related to strategic capabilities in decision making. In the context of Malaysian public sector, the rising of public expectations towards service deliverables are in the respond to dynamics population and technology trending. In the Eleventh Malaysia Plan (RMK-11) [2], the government will become more citizen-centric and focus on enhancing productivity of the public service through a whole-of-government approach supported by a lean and agile structure, competent talent, effective delivery of projects, and efficient services of the local authority. Therefore the parties responsible for determining the direction the organization should have skills and strategic thinking in order to produce more high-performance results. In term of Information and Communication Technology (ICT), since many applications or projects will be developed soon in Malaysia and one main standard is to ensure that it is world’s class quality, an exploration on how the development of Malaysian public sector ICT projects is one of interesting area to look into [3]. According to [4] the fact that many ICT projects have failed, it has encouraged many researchers to investigate the problems of poor performance. As of that, many studies have been conducted to find out reasons and resolve it from the perspective of ICT expertise and technology. However there are still less parties to undertake the situa‐ tion from the perspective of the integration of ICT knowledge among experts and deci‐ sion-makers mainly for communication such as meetings and discussions to any decision related to ICT. Since less attention is given in knowledge communication area and not much in existing academic publications therefore objectives of this paper are to explain: i. Importance of knowledge communication; and ii. Identify relevant frameworks from other area to develop a conceptual knowledge communication model. In this paper, past studies on several frameworks in same domain were reviewed in order to identify most relevant frameworks to develop a knowledge communication conceptual model. A knowledge communication conceptual model then will be devel‐ oped accordingly. The following sections are organized as follow; Sect. 1 introduces briefly about this study and Sect. 2 explains on literature of knowledge communication. Then Sect. 3 discusses on methodology applied in this study. Section 4 reports the findings and proposes a knowledge communication conceptual model. Finally, Sect. 5 concludes and provides recommended scope for future research.

2

Research Background

The main elements of humanity in any organization is the decision maker that can consist of one or a group of individuals who responsible for making wise decisions. This is also need to be accompanied by non-human elements such as facts, procedures and docu‐ mentation before blended to form useful for the assessment of knowledge in decision making. While decision makers typically have the authority to make strategic decisions,

The Knowledge Communication Conceptual Model

29

they often lack the expertise required to make an informed decision on complex issue [5]. Most of good and wise decisions that need to be made will usually make compulsory for decision makers to delegate the decision preparations to experts for technologically opinions in a more reliable manner. While in literature review, the relationship between IT experts and decision makers are widely addressed in ICT Project Management especially in the stage of project initiation [6]. Furthermore, [7] has mentioned the business analysis is crucial during project initiation. During business analysis, IT experts need to engage with decision makers to help with the project scopes and definition of the expected project outcomes (deliverable), project acceptance criteria and business requirements. One of the essential elements during business analysis is the communication pattern between IT experts, decision makers, stakeholders and IT team is crucial to clarify the knowledge for project successful. Therefore, a proper knowledge communication model is compulsory for organizations to produce best product in their field. 2.1 Overview on Knowledge Communication Knowledge communication is more than communicating information or feeling because it requires in conveying the whole thing such as situation, background and basic hypoth‐ esis. More importantly it requires the statement of personal opinion and experiences. It does not only differ in terms of what is communicated but also how one communicates with another. High-quality communication between decision makers and experts can be possible if interaction of experts can adapt their content and communication style to the needs of decision makers; and if only the decision makers fully briefs experts their requests and give clear and regular feedback. Communication is a process of acquiring all information then interprets and circulates it to who might need it while knowledge is understanding of or information by experience or study. Knowledge Communication (KC) is one of the divisions in Knowledge Management (KM). Although the concept of knowledge communication is still new as knowledge management but the effective‐ ness in managing aspect of an organization is undeniable. According to [8] knowledge communication can be defined as (deliberate) activity of interactively conveying and co-constructing insights, assessments, experiences, or skills through verbal and non-verbal means. Knowledge communication has taken place when an insight, experience or skill has been successfully reconstructed by an individual because of the communicative actions of another. The exchange of know-how, knowwhy, know-what, and know-who through face-to-face or media-based interaction. Although the phenomenon of this knowledge communication has been going on for quite some time in many organization but it is still new in the world of academic research learning. Knowledge communication is more than just communicating information or emotions such as facts, figure, hopes, commitment and others because it requires expressing perspective, basic hypothesis, personal opinions, experiences and back‐ ground of situation. Thus, in order for the management to decide on reliable and wise decisions, they need to have clear understanding on the particular issues. So the experts need to clarify their knowledge in a proper approach for others to understand the relevant issues.

30

R. Daud et al.

2.2 Knowledge Communication in Malaysian Public Sector By considering the motivation and initial steps of [9] study, the research is to continue in greater depth perspective of the decision makers in the public sector environment in Malaysia. It is also found that the relevance of this field by direct observation in the case of the public sector in Malaysia where IT experts and decision makers have differing requirements, guidelines for a comprehensive overview. IT experts and decision makers are members of a group who responsible for making decisions and planning future direction of the organization and in this context is Malaysian public sector; hence the failure of ICT projects must not only be imposed and settled on ICT and technical side. On the other hand decision makers should be proactive in identifying issues ranging from whether the decision making level for the planning, implementation and moni‐ toring of ICT projects. As the Malaysian government has released most of the allocations for ICT projects it is expected that output/outcome must be equal to money invested. With people’s different background, scope and diversity of the various specialties of ideas it is difficult to understand as a whole when the integration is carried out between various domains of knowledge. Therefore when IT experts and decision makers discuss on particular issue without proper guidelines they will construct an overall picture of the issues based on their own understanding. The uncertainty and confusion in the establishment of a comprehensive overview will lead to actions that deviated in the phase of analysis and subsequent synthesis. As a result, to resolve certain issues (system of interest), the achievement of an overview of the standard and focused is important to readjust the mind thought of all parties involved so that an understanding can be accom‐ plished. It will also act as a blueprint which will be the main driver to the understanding, correct understanding and decision-making processes with higher integrity [9–13]. Next section discusses on relevant frameworks to apply in developing knowledge communi‐ cation conceptual model. 2.3 Web Services as Mode of Knowledge Communication The meaning of web services is an automated interaction between sender and receiver for connecting business processes [14]. People around the world use personal computers, tablets and smart phones to do many things from browsing recipes, socializing and more. In 2017 web services are foresee as a one of most build up technology by providing better voice commands, cloud adoption data centre and gadget-technology integration therefore it is an essential for knowledge communication to use web services. Moreover web services also capable to reach new or current user, operating efficiently and worth to apply. Therefore decision making process will be easier and more smooth for parties whom involved. In addition to the impact of web services also, knowledge communi‐ cation between IT experts and decision makers can be held anywhere and anytime over the internet.

The Knowledge Communication Conceptual Model

3

31

Methodology

This section outlines the search strategy and search criteria adopted for this study. A total of 5 research databases were searched with journals or articles from 2007 until the present (2016). The databases were web of science, emerald, Science Direct, IEEE explore digital library, google scholar. The keywords used are “knowledge” AND “sharing” AND “framework” for knowledge sharing framework while keywords for knowledge transfer model are “knowledge” AND “transfer” AND “model” or “frame‐ work” anywhere in the articles. Then, a comprehensive examination was made and at this point any articles or journals were excluded if the organization or communication were insufficiently described by the authors. Relevant publication titles were also selected during this period of time. Finally 5 relevant knowledge sharing frameworks and 5 relevant knowledge transfer frameworks were chosen. Fundamentally, knowledge sharing framework is attempt to collect all facts and data into one before it supplies a more complete approach to understand the event of knowl‐ edge sharing between IT experts and decision makers. On the other hand, knowledge transfer framework is an interactive, dynamic and multidirectional process. Therefore, it will provide a solid basis for gathering proof and facts which will enable to confirm, invalidate or revise each of the process before a new thought of most appropriate selec‐ tions might happen. Knowledge sharing usually happened when a person is interested to help other people develop a new action potential. Knowledge sharing is able to increase intellectual and organizational resources and competitiveness, change organi‐ zational direction to better ones and reduce cost. As to that it is important for any organ‐ ization to apply knowledge sharing among employees because it can help to increase performance and facilitate new information/knowledge. From the search before, 5 knowledge sharing frameworks by [15–19] are being reviewed and compared. The first comparison is the knowledge sharing framework by [15]. According to [15] literature analysis’ there are four major factors influence knowledge sharing between individual in organizations: the nature of knowledge, motivation to share, opportunities to share and the culture of work environments. These identified factors have major influence on knowledge sharing process and relate to each other. Secondly, is knowledge sharing framework by [16] which were carried out based on existing knowledge sharing literature and recap the related current issues and future research needs. Five significant areas and relationship between each area were classified by [16] namely organizational context, interpersonal and team characteristics, cultural characteristics, individual char‐ acteristics and motivational factor. Third knowledge sharing framework was from [17] whom identified that internal marketing (management support) and organization culture (trial and improvement) influence employee knowledge sharing attitude and apparent behavioural control. The factors were categorized in this framework includes: internal marketing, organizational culture, perceive behaviour control and attitude. Then, there are six factors have been identified by [18] that is: attitude, perception, motivation, biographical characteristics, learning and personality. Finally, the framework proposed by [19] or known as ShaRInK specified four main factors form individual perception of knowledge sharing. The four (4) factors are sharer,

32

R. Daud et al.

institution, relations and knowledge. Based on discussion about knowledge sharing frameworks, the commonality factors between these 5 frameworks were chosen to develop a knowledge communication conceptual model. Therefore it is reasonable to choose ShaRInk framework because it’s the most recent, 12 interrelationships with all the factors is deemed to knowledge communication in Malaysian public sector, it is also an enhancement from [16] and definitely it would be interesting to study in our research context. The relevant factors of five selected frameworks are as in Table 1 and the framework is in Fig. 1. Table 1. Factors of five selected knowledge sharing frameworks Factors in KS/ Authors Motivation Culture characteristics Knowledge Individual & team characteristics Personality Perceptions Organizational characteristics

Ipe (2003) / /

Wang and Noe (2010) / /

Aslani et al. (2012) / X

Chen and Cheng (2012) X /

Schauer et al. (2015) / /

/ X

X /

X X

X X

/ /

X / X

/ / /

/ / X

/ / /

/ / /

Legend: / - YES; X - NO

Fig. 1. ShaRInk framework by Schauer et al. (2015)

The Knowledge Communication Conceptual Model

33

Knowledge transfer on the other hand is defined as conveyance of knowledge from one place, person or ownership to another. According to [20] knowledge transfer involves communicating with others what one knows or consulting to learn what they know. There are three types of knowledge transfer processes which are a linear process, a cyclical process and a dynamic multidirectional process. Although, knowledge transfer models/frameworks are slightly different, have much resemblance. Basically the idea of knowledge transfer is communication or partnership between two main components; source (sender) that shares knowledge and receiver who obtain the knowledge [21]. For the reviews and comparison, each of the five knowledge transfer frameworks/models are briefly summarised below. The framework proposed by [22], combines five factors: leadership, problemsolving behaviour, positive capacity, support structure and types of knowledge. Then knowledge transfer framework by [23] carried out a model of knowledge transfer in various cross-border scenarios. The three factors of cross-border knowledge transfer model are types of knowledge, nature of cultural pattern and cognitive style in organi‐ zation. Third is a knowledge transfer model by [24]. The idea of knowledge transfer model by [24] is mainly built upon two main elements, source and receiver or known as ‘an act of communication’. Four factors have been introduced in the model: knowl‐ edge relevance, knowledge distribution, acquisition and absorptive capability. The advantage of this model is the inclusive of type of knowledge which are tacit and explicit based on [25]. Ward et al. (2010) have developed a framework illustrates five factors or components and connected through a multidirectional process. The five factors are: problem identi‐ fication, utilization, involvement, context analysis and knowledge research. Last but not least is the framework by [26] is based on knowledge transfer process across project in project-based organizations (PBO) which holds a great impact to the success of

Table 2. Factors of five selected knowledge transfer frameworks Factors/ Authors Knowledge type Transfer mechanism Transfer success Problem identification Context analysis Absorptive capability Leadership

Goh (2002) /

Bhagat et al. (2002) /

Ward et al. (2010) /

Liyanage et al. (2009) /

Van Waveren et al. (2014) /

/

/

/

/

/

X

X

/

/

/

/

X

/

/

X

X

X

/

X

X

/

/

X

/

X

/

X

X

X

X

Legend: / - YES; X - NO

34

R. Daud et al.

organizations. Consequently this framework is to reunite theoretically the knowledge transfer process. The factors in this framework namely: knowledge type, transfer mech‐ anism and transfer success. For future research it is suggested to validate the relationship amongst factors. From the discussion on knowledge transfer framework, the factors are compiled as in Table 2. As identified above, knowledge type and transfer mechanism play a huge roles in developing knowledge transfer framework. The knowledge transfer framework is slightly same between all five frameworks. The needs factors for knowledge commu‐ nication conceptual model have make obligatory for knowledge transfer framework by [24] was chosen. Therefore, the end of this chapter a proposed conceptual knowledge communication model collaborate two frameworks which are knowledge sharing frame‐ work by [19] and knowledge transfer framework by [24]. The illustrated framework is shown in Fig. 2. Based on these figures the next section proposes a knowledge commu‐ nication conceptual model and brief description about proposed conceptual model.

Fig. 2. Knowledge transfer framework by Liyanage (2009)

The Knowledge Communication Conceptual Model

4

35

Findings and Discussion

Based on the identified factors from Tables 1 and 2, the proposed knowledge commu‐ nication conceptual model combines both of knowledge sharing framework and knowl‐ edge transfer model. The proposed conceptual model is shown as in Fig. 3. In order to expand appreciation of the factors in proposed conceptual model, each is briefly discussed in the following subsection.

SHARER

(individual characteristics, motivational & perceptions)

RELATIONSHIP

(Interpersonal & Team Characteristics)

Explicit to tacit (internalization)

Tacit to explicit (externalization)

e.g : learn from a report

e.g : answer questions, dialogue

INSTITUTION

Modes of knowledge sharing/transfer Tacit to tacit

(cultural, characteristics, organizational context)

Explicit to explicit

(socialization)

(combination)

e.g : discussions

e.g : email a report

KNOWLEDGE (Type of knowledge)

Fig. 3. Knowledge communication conceptual model

4.1 Proposed Knowledge Communication Conceptual Model This proposed conceptual model combines of knowledge sharing framework and knowl‐ edge transfer model. There are four main categories which are sharer, relationship, institution and knowledge. The first factor is concerning the sharer itself which consists of three items. The items are individual characteristics, motivational and perceptions on knowledge to be shared. Second factor focuses on institutions that act as a united entity which includes cultural characteristics and organizational context as the items. For rela‐ tionship category, it concentrates as the association between the sharer and other-sharer. The only item in this factor is interpersonal and team characteristic. Last but not least, the knowledge itself presented independently in which type of knowledge is the item. The modes of knowledge transfer from the source and receiver are divided by four types [25]. The four modes are explicit to tacit (learning from a report), tacit to explicit (small dialogue session), tacit to tacit (team meetings) and explicit to explicit (email a report). The whole proposed model theoretically offers many insights to the process of

36

R. Daud et al.

knowledge communication in an organization. Complete background, context and expe‐ rience of relevant issues or problems can be shared amongst team of decision maker and IT experts. With that, common understanding between IT experts and decision makers will lead to a better and wise decision making to organization. The whole proposed model theoretically offers many insights to the process of knowledge communication in an organization. Common understanding between IT experts and decision makers will lead to a better and wise decision making to organi‐ zation.

5

Conclusion and Future Works

As a conclusion, the objectives of this paper were to explain the importance of knowl‐ edge communication and to develop a conceptual knowledge communication model. From analysis of the literature has revealed the importance of IT experts and decision makers in relying on each other through evolving knowledge communication when attempting in making reliable and excellent decision for the organizations. A knowledge communication conceptual model was proposed by addressing components and factors found in knowledge sharing framework and knowledge transfer framework. The type of frameworks was chosen according to [1], which indicated that knowledge commu‐ nication is to overcome the existing weakness of knowledge management like knowl‐ edge sharing and knowledge transfer respectively. Therefore, the practical implication of the proposed conceptual model may help respective authorities to take effective measures to improve knowledge communication situation especially for ICT projects in Malaysian public sector. In addition to that, from the conceptual model, this study realizes the significant of tacit knowledge in knowledge communication. Hence, by utilizing web services technology this study intend to externalize the tacit into an explicit knowledge. This is because by having an explicit knowledge the issues of any organi‐ zation are able to easily stored, shared and transferred. However, since the conceptual model has yet been tested, an initial study is much recommended. An initial study is to discover factors influences knowledge communi‐ cation in the organizations. A comparison between theoretical and actuality is an approach to improve the above proposed conceptual model. Then, verification from the practitioners in industries is highly suggested. Finally as for theoretical implication, this study is hopefully can allow the researchers and practitioners to understand the factors influencing the knowledge communication process between IT experts and decision makers. Acknowledgment. The research is financially supported by Public Service Department of Malaysia and Universiti Teknologi Malaysia under Research University Grant (RUG), Vot No: 14H16.

The Knowledge Communication Conceptual Model

37

References 1. Reinhardt, R., Stattkus, B.: Fostering knowledge communication: concept and implementation. J. Univers. Comput. Sci. 8(5), 536–545 (2002) 2. P. M. D. Economic planning unit. Eleventh Malaysia Plan (2016–2020) (2016) 3. Meng, C.C., Samah, B.A., Omar, S.Z.: A review paper: critical factors affecting the development of ICT projects in Malaysia. Asian Soc. Sci. 9(4), 42–50 (2017) 4. Patanakul, P.: Managing large-scale IS/IT projects in the public sector: problems and causes leading to poor performance. J. High Technol. Manag. Res. 25, 21–35 (2014) 5. Watson, C.M.: Don’t blame the engineers. MIT Sloan Manage. Rev. 4 (2004). Winter 2004 6. Knipfer, K., Mayr, E., Zahn, C., Schwan, S., Hesse, F.W.: Computer support for knowledge communication in science exhibitions: novel perspectives from research on collaborative learning. Educ. Res. Rev. 4(3), 196–209 (2009) 7. Korban, S.: Project manager and business analyst in tandem for success. BA Times (2014) 8. Eppler, M.J.: Knowledge communication problems between experts and decision makers: an overview and classification. Electron. J. Knowl. Manag. 5(3), 291–300 (2007) 9. Mengis, J., Eppler, M.J.: Integrating knowledge through communication: an analysis of expert-decision maker interactions. Universtu of Lugano (2007) 10. Lamont, M.: Toward a comparative sociology of valuation and evaluation. Annu. Rev. Sociol. 38(21), 201–221 (2012) 11. Eppler, M.J.: 10 years of knowledge-communication.org results, insights, perspectives (2012) 12. Ya’acob, S., Mohamad Ali, N., Mat Nayan, N.: Understanding big picture and its challenges: experts and decision makers perspectives. In: Zaman, H.B., Robinson, P., Olivier, P., Shih, T.K., Velastin, S. (eds.) IVIC 2013. LNCS, vol. 8237, pp. 311–322. Springer, Cham (2013). doi:10.1007/978-3-319-02958-0_29 13. Mengis, J.: Integrating knowledge through communication - the case of experts and decision makers. In: OLKC 2007, vol. 44, no. 0, pp. 699–720 (2007) 14. Benslimane, D., Dustbar, S., Sheth, A.: Services mashups: the new generations of web applications. IEEE Internet Comput. 12, 13–15 (2008) 15. Ipe, M.: Knowledge sharing in organizations: a conceptual framework. Hum. Resour. Dev. Rev. 2(4), 337–359 (2003) 16. Noe, R.A., Wang, S.: Knowledge sharing: a review and directions for future research. Hum. Resour. Manag. Rev. 20(2), 115–131 (2010) 17. Chen, W., Cheng, H.: Management factors affecting the knowledge sharing attitude of hotel service personnel. Int. J. Hosp. Manag. 31, 468–476 (2012) 18. Aslani, F., Mousakhani, M., Aslani, A.: Knowledge sharing: a survey, assessment and directions for future research: individual behavior perspective. Int. J. Soc. Behav. Educ. Econ. Bus. Ind. Eng. 6(8), 2025–2029 (2012) 19. Schauer, A., Vasconcelos, A.C., Sen, B.: The ShaRInk framework: a holistic perspective on key categories of influences shaping individual perceptions of knowledge sharing. J. Knowl. Manag. 4(19), 770–790 (2015) 20. Van Den Hooff, B., De Ridder, J.A.: Knowledge sharing in context: the influence of organizational commitment, communication climate and CMC use on knowledge sharing (2005) 21. Carlile, P.R.: Transferring, translating, and transforming: an integrative framework for managing knowledge across boundaries. Organ. Sci. 15(5), 555–568 (2004) 22. Goh, S.C.: Managing effective knowledge transfer: an integrative framework and some practice implications. J. Knowl. Manag. 6(1), 23 (2002)

38

R. Daud et al.

23. Bhagat, S.R., Ben Kedia, L., Harveston, D.P., Triandis, C.H.: Cultural variations in the crossborder transfer of organizational knowledge: an integrative framework. Acad. Manag. Rev. 27(2), 204–221 (2002) 24. Liyanage, C., Elhag, T., Ballal, T., Li, Q.: Knowledge communication and translation – a knowledge transfer model. J. Knowl. Manag. 13(3), 1–23 (2009) 25. Nonaka, I.: A dynamic theory of organizational knowledge creation. Organ. Sci. 5(1), 14–37 (1994) 26. Van Waveren, C.C., Oerlemans, L.A.G., Pretorius, M.W.: Knowledge transfer in projectbased organizations. A conceptual model for investigating knowledge type, transfer mechanisms and transfer success. In: IEEE, pp. 1176–1181 (2014)

Applying Process Virtualization Theory in E-HR Acceptance Research: Testing and Modifying an Experiment C. Rosa Yeh ✉ and Shin-Yau Hsiao (

)

Graduate Institute of International Human Resource Development, National Taiwan Normal University, 162 Sect. 1 Heping East Road, Taipei, Taiwan, Republic of China [email protected]

Abstract. Some HR processes are more easily accepted when they go online, why? The Process Virtualization Theory provides some viable explanation. This article presents the development, testing and modification of an experiment to be used in studies predicting virtualizability of HR processes. The experimental procedure was developed along with an instrument that contains measurement of Process Virtualization requirements and other criterion variables. Two e-HR process mock-up were developed for the testing purpose. Data was collected from 230 business majors from six different colleges located in northern Taiwan. Students were randomly divided into two groups in a computer lab setting. Each group experienced a different e-HR process mock-up. T-test result shows that the procedure and the instrument was able to find a significant difference in relation‐ ship requirements and monitoring capability between the two e-HR processes. However, the two groups do not show a significant difference on the criterion variable behavior intention. Keywords: E-HR · Process virtualization · Behavioral intention · Experimental method

1

Introduction

As HR practices are being replaced with their technological counterparts, processes that used to be conducted physically are now being converted into electronic ones. This is known as the process of virtualization [1]. The Process Virtualization Theory seeks to find out which processes are more suitable to be conducted using technology, and which are not. It is a good idea to find out whether a certain function or process is suitable for virtualization or not before diving deep into purchasing or developing the technology [1]. E-HR systems introduce a lot of benefits for the HR professionals, the employees, and the organization itself. However, although E-HR is gaining popularity among organizations, there are still a lot of problems associated with the actual willingness to implement and use the system. Implementing an online HR system is a complex process that requires firms to manage both significant changes for the employees as well as the technical aspects of implementation. The initial costs of implementing an e-HR system © Springer International Publishing AG 2017 L. Uden et al. (Eds.): KMO 2017, CCIS 731, pp. 39–48, 2017. DOI: 10.1007/978-3-319-62698-7_4

40

C.R. Yeh and S.-Y. Hsiao

is a large burden on the organization, involving over tens of thousands of US dollars. Costs associated with implementing an E-HR system may include computer and internet to access the system for every employee, hardware costs for servers, software costs for application programs, training, and maintenance of the system [2]. This study intended to develop an experimental method to empirically test the Process Virtualization Theory in predicting the use of E-HR systems. It is important to explore the issue of virtualizing HR processes. For organizations, it would be a better idea to invest in e-HR processes that are more easily virtualized. Without understanding the requirements for process virtualization, organizations may invest in costly tech‐ nology implementation that employees never use.

2

Theoretical and Literature Background

2.1 E-HR (Electronic Human Resource) The term “E-HR” was first used in the 90 s when “e-commerce” (or electronic commerce) was introduced to the business world. Like e-commerce, E-HR referred to conducting HR functions over the internet [2]. The “e-enabled” versions of HR has been referred as many different terms such as HRIS, E-HR, B2E, ESS, web enabled ESS, HR portal (Hawking et al. 2004). E-HR systems contribute to major changes in the way in which organizations operate. The technology makes many of the administrative processes available for the general employee. E-HR also frees up the HR personnel from administrative burden and allows them to take a more strategic role. However, the EHR technology itself is not the only determinant in making HR work more strategic. The user of the technology, mainly HR personnel also needs to adapt themselves to the technology in order to take a more strategic role [3]. Therefore E-HR adoption still remains to be a critical issue today. 2.2 Process Virtualization Theory Overby [1] stated that some processes are more amenable to being conducted virtually than others. This means that some processes are more suitable to be conducted elec‐ tronically. In the current society where everything is becoming virtualized, it is vital to know which processes are fit to be virtualized; hence the development of Overby’s Process Virtualization Theory. The theory includes four main constructs; they are Sensory Requirements, Relationship Requirements, Synchronism Requirements, and Identification and Control Requirements. Overby [1] argued that the processes become less virtualized when each of the requirement increases, as Shown in Fig. 1. In other words, the processes would be more amenable to being virtualized if these requirements are low.

Applying Process Virtualization Theory in E-HR

41

Fig. 1. Theoretical model of Process Virtualization Theory. (Adapted from overby, 2008).

Sensory Requirements. Sensory Requirements, defined as “the need for process participants to be able to enjoy a full sensory experience of the process and the other process participants and objects. Sensory experiences may include tasting, seeing, hearing, smelling, and touching other process participants or objects, as well as the overall sensation that participants feel when engaging in a process, e.g., excitement, vulnerability, etc.” [1, p.280]. The lack of physical interaction makes it difficult for a participant in a virtual process to establish a sensory connection to objects and/or other people. Thus, Sensory Requirements are posited to have a negative relation to process virtualizability (P1). Relationship Requirements. Relationship Requirements, defined as “the need for process participants to interact with one another in a social or professional context. Such interaction often leads to knowledge acquisition, trust, and friendship development” [1, p. 280]. Relationship Requirements are also thought to have a negative relation to process virtualizability (P2). Synchronism Requirements. Synchronism Requirements, defined as “the degree to which the activities that make up a process need to occur quickly with minimal delay” [1, p. 281]. In which Overby [1] used grocery shopping as an example, if a process needs to be conducted in a synchronous manner, such as getting fresh produces and perishable goods right after purchase, it will benefit from the physical context and resist virtuali‐ zation. Therefore, Synchronism Requirements appear to have negative relation with process virtualizability (P3). Identification and Control Requirements. Identification and Control Requirements are defined as “the degree to which the process requires unique identification of process participants and the ability to exert control over/influence their behavior” [1, p. 282]. Overby [1] believes that virtual processes are prone to identity spoofing because partic‐ ipants cannot physically identify others. Therefore, it is difficult to detect who actually engage in the interactions when it is important to identify the other process participants. Thus, processes high on the Identification and Control Requirements are believed to resist virtualization (P4).

42

C.R. Yeh and S.-Y. Hsiao

2.3 Behavioral Intention to Use Technology One of the very popular theories involving the use of technology is the Technology Acceptance Model (TAM) developed by Fred Davis in 1986 [4], which was considered as one of the most influential and commonly cited theories for explaining an individual’s acceptance of information systems [5]. Although TAM has generally been used to explain how Perceived Usefulness (PU) and Perceived Ease of Use (PEU) predict initial acceptance to adopt an information system, it has also been deployed for predicting users’ intention to use an information system [6]. The theoretical basis of Davis’s TAM is Fishbein and Ajzen’s Theory of Reasoned Action (TRA). According to Fishbein and Ajzen [7, p. 380], “since much human behavior is under volitional control, most behav‐ iors can be accurately predicted from an appropriate measure of the individual’s inten‐ tion to perform the behavior in question”. Therefore, it is safe to assume that intention can predict actual use. Many factors are believed to affect an individual’s Behavioral Intention towards using a particular technology. Therefore, many scholars have sought to extend TAM by adding external factors. One of the more cited extension of TAM is Venkatesh et al.’s Unified Theory of Acceptance and Use of Technology (UTAUT) model [8] which inte‐ grated many factors from previous models including TAM and TRA. Some examples in addition to Davis’s PU and PEU are factors such as Subjective Norm (would the people around me want me to use this technology?), Complexity (how much effort it would take for the user to learn the new technology?), and Compatibility (does the technology fit the job as intended?). Although Computer Self-efficacy and Attitude toward Technology were both said to have indirect and moderating effects in the study of Venkatesh et al. [8], the study failed to prove any significant result, however, the two factors were shown to have direct effect on intention in other studies [9].

3

Research Method

3.1 Research Design Different HR processes have been suspected to account for different degrees of process virtualizability. The study chose two HR processes that show opposite trend in degrees of virtualization as measurement context: (1) change of personal profile information and (2) performance appraisal. The chosen processes shall only act as a skeleton for future explorations of other processes (i.e. the adoption of the measurement items to a specific process). The measures are tested on two “extreme examples”, to see if they work out well in at least two different and distinct situations. Similarly, the research controls the variation in sophistication of IT technology by conducting the research in an E-HR Lab where two “extreme examples” of IT are used to test virtualization requirements of different processes. The “change of personal profile information” HR process was believed to be low in the process virtualization requirements because minimal social interaction is involved. During the general process, the applicant only had to request for a form to change his/her personal profile, fill it out and turn it in. Therefore, interaction with human being

Applying Process Virtualization Theory in E-HR

43

or object is assumed to be minimal. On the other hand, the “performance appraisal” HR process was believed to have a much higher process virtualization requirement because normally the performance appraisal process has to be done with a lot of forms and human interactions. The forms have to go back and forth between different departments, managers, and HR professionals. The process may also require face-to-face meetings between applicants and managers/HR professionals. The experiment was conducted using one of the most sophisticated E-HR products, PeopleSoft. Due to the complexity of the PeopleSoft software, an imitation of the soft‐ ware was used on the subjects to provide control over the experiment and reduce time spent on training the subjects. Using an imitated “mock-up” version of PeopleSoft, all the participants were able to go through the same procedure without wandering off to other functions of the software. This also minimized the difference in familiarity with the software. The participants were still able to click and interact with the designated functions for the experiment. The idea was to make the experience as uniformed as possible, providing sufficient control over the process of the experiment. The entire process took an average time of approximately 40 min. A flow chart on the experimental procedure is shown in Fig. 2. 3.2 Sample and Data Collection The sample for this research was targeted at college to graduate level students with business major. This is due to the fact that they are the soon-to-be potential workforce. Also, business major students may have a better idea of organizational processes. Convenience sampling was used to find sample students. The researcher contacted teachers of business major classes and asked for permission to come to their classes to solicit participation. At the end, the researcher paid a total of five visits to different college campuses in northern Taiwan, set up E-HR labs in their computer classrooms, and conducted the experiments. Some participants were invited to the researchers’ eHR lab for experiment. A total of 230 samples were collected from 6 different schools. 217 out of 230 samples were valid for analysis in the end, which was about 94%. The subjects ranged from college level to master level students. 3.3 Measurement Process Virtualization Requirements. A scale for measuring the process virtualiza‐ tion requirements from Overby and Konsynski [10] was used as a starting point for developing measurement under HR context. The original measurement scale did not have enough items, and some of the items did not fit the HR context, therefore, some items for measuring the process virtualization requirements had to be developed. For the Sensory Requirements, the original concept from the theory was to measure the degree to which the participants need to enjoy a full sensory experience of the process and the other process participants and objects. Another measurement scale by Schmitt, Zarantonello, and Brakus [11] was supplemented in the development of measurement items. The Relationship Requirements aim to measure the need for process participants

44

C.R. Yeh and S.-Y. Hsiao

Student Subjects

Group (A) HR process with high Process virtualization requirements (Performance appraisal)

Group (B) HR process with low Process virtualization requirements (Change of personal profile information)

Part A Questionnaire (510min) Attitude toward Technology Computer Self-efficacy

Group (A) Process Briefing (5min)

Group (B) Process Briefing (5min)

Part B Questionnaire (5-10min) Process Virtualization requirement measurements

Group (A) (5min) Tutorial video

Group (A) (10-15min) Hands on experience with PeopleSoft

Group (A) (5min) Tutorial video

Group (B) (10-15min) Hands on experience with PeopleSoft

Part C Questionnaire (5-10min) IT Capability measures Behavioral intention measures Fig. 2. Experimental Procedure

to interact with one another in a social or professional context. Items from the scale of Lin, Lin, and Laffey [12] was used to complete the scale for Relationship Requirements.

Applying Process Virtualization Theory in E-HR

45

Synchronism Requirements refer to the need for a process to occur quickly with minimal delay. Items were adapted from Culiberg and Rojšek [13] and Jun, Yang, and Kim [14] to develop a scale to measure Synchronism Requirements under HR context. The goal of the Identification and Control Requirements is to measure the need for a process to require unique identification of process participants and the ability to exert control over or influence their behavior. Items from Ho and Lee [15] were adapted to complete the measurement scale for Identification and Control Requirements under HR context. The measurement was conducted using a Likert type scale rating of 1 to 5, 1 would mean strongly disagree with the statement and 5 would be strongly agree with the state‐ ment. A higher score would mean a higher process virtualization requirement. Scores were aggregated under each of the process requirements. During the scale validation process, Sensory and Relationship Requirements merged into one factor during exploratory factor analysis, and thereafter was analyzed as one construct. Behavioral Intention. Behavioral Intention attempted to measure an individual’s overall intention to use online HR technology. A scale from Hu, Chau, Sheng, and Tam [16] was adapted in development of the scale to measure Behavioral Intention. The measurement was conducted using a Likert type scale rating of 1 to 5, 1 would mean strongly disagree with the statement and 5 would be strongly agree with the statement. A higher score would mean a higher Behavioral Intention to use online HR technology. Control Variables. Attitude toward Technology, Computer Self-efficacy, Represen‐ tation Capability, Monitoring Capability have been known to affect Behavioral Intention to use technology; they are therefore included in the study as controls [8, 17].

4

Findings

The participants received two types of treatments, one with assumed low process virtu‐ alization requirements; change of personal profile information. The other one was performance appraisal, which was assumed to have higher process virtualization require‐ ments. The comparisons of mean scores on all the research variables between groups “change of personal profile information” and “performance appraisal” are shown in the table below. According to t-test result shown in Table 1, the experiment and the measurement scale was able to identify different Relationship Requirements with different HR processes. It was also able to identify different Monitoring Capability of E-HR software between different processes. In Relationship Requirements, it was assumed that performance appraisal would have higher requirements. The performance appraisal process requires a lot of communication between the supervisor, the employee, and the HR. With so much interaction requirements, people would want to interact in person; this was confirmed with the experimental results. The Synchronism and the Identifica‐ tion Requirements were both not significant in this study. Since the two selected processes do not involve physical transactions or carry a sense of urgency, the

46

C.R. Yeh and S.-Y. Hsiao

participants may see both processes with equally moderate Synchronism Requirements. On the other hand, both processes require users to identify themselves and to enter userspecific information, therefore it is reasonable for the participants to see both processes with equally high level of Identification Requirements. The significant difference of Monitoring Capability may implicate that the participants were more worried about privacy issues with the change of personal profile process. Therefore, they feel less confident with their privacy in the change of personal profile process, resulting in a perception of lower Monitoring Capability in the change of personal profile process in PeopleSoft software. Significant differences on Behavioral Intention was not found. Table 1. T-test between different treatments Variable

Treatment (Mean value) Change of profile Performance information appraisal Attitude toward technology 3.74 3.78 Computer self-efficacy 3.54 3.49 Relationship requirements 3.71 3.90 Synchronism requirements 3.69 3.72 Identification requirements 4.15 4.06 Representation capability 3.72 3.71 Monitoring capability 3.28 3.57 Behavioral intention 3.51 3.61

Mean difference

t-value

−.034 .046 −.192* −.025 .086 .015 −.288*** −.097

.676 .575 .013* .764 .300 .848 .001* .195

Note. *p < 0.05, **p < 0.01, ***p < 0.001

5

Discussion and Conclusions

The Process Virtualization Theory included four process virtualization requirements, sensory, relationship, synchronism, and identification requirements, however when used under HR context, Sensory and Relationship Requirements merged into one. The study was a brand new experiment requiring the development of new scales, instruments, and experimental procedures to test the theories. It was also an informative test of the Oracle PeopleSoft E-HR software. For the Process Virtualization Theory, the developed experi‐ ment was able to detect differences in Relationship and Sensory Requirements between different HR processes. In empirical test of the Process Virtualization Theory, it is important to note that out of the four process virtualization requirements, Sensory and Relationship Requirements merged into one during exploratory factor analysis when it was tested under HR context. It is vital to know that when used for interaction with people, instead of products, Sensory Requirements are less relevant. The Process Virtualization Theory did not show signif‐ icant results for the Net generation under the HR context. Another reason that the Process Virtualization Theory did not affect Behavior Intention may have to do with the lack of a pre-test and a post-test of behavior intention to use E-HR in the experiment procedure. The assessment of Behavioral Intention at the end of the procedure might have been the result of a direct reaction toward the E-HR software capability which mimics the

Applying Process Virtualization Theory in E-HR

47

PeopleSoft E-HR system. To conduct more timely and thus more accurate assessment of the effect of Process Virtualization Requirements, another assessment of Behavioral Intention should have been added after an HR process was introduced and before the hands-on experience on PeopleSoft. The particular research also developed a way of carrying out theoretical evaluation tests by introducing “mock-up” software for maximum experiment control. It proved to be very effective in letting the participants experience the E-HR technology in a controlled manner to contain unwanted effects. All in all, this study set a start in how EHR software experiments can be conducted. The experimental procedures appeared to be effective in testing the capability of E-HR software.

6

Limitations and Future Research Suggestions

Due to resources and funding issues, the study could not examine the process virtuali‐ zation requirements for each HR process one by one. Only two extreme examples were selected to test the virtualization requirements. The measurement scales and experi‐ mental procedures used in this particular study are all newly developed. This makes the study lean towards the exploratory side. A standardized and efficient experimental research procedure for testing HR virtu‐ alization requirements and the capability of E-HR software has been developed and tested in this particular research. The use of “mock-up” software also proved to be useful in controlling the experiment. Future researches may follow the pattern of revised experimental procedures with the addition of pre-test/post-test Behavioral Intention measures. However, because the current measurements have only been tested on students of Net generation, this set of measurements should best be validated again before testing on other HR processes. Moreover, future research should bring the experi‐ ment to actual industry settings instead of schools in order to make maximum contribution. Acknowledgement. This research was supported by the Ministry of Science and Technology of Taiwan, ROC. (NSC 102-2410-H-003-102 -).

References 1. Overby, E.: Process virtualization theory and the impact of information Technology. Organ. Sci. 19(2), 277–291 (2008) 2. Lengnick-Hall, M., Moritz, S.: The impact of e-HR on the human resource management function. J. Labor Res. 24(3), 365–379 (2003) 3. Marler, J.H., Fisher, S.L.: An evidence-based review of E-HRM and strategic human resource management. Hum. Res. Manage. Rev. 23(1), 18–36 (2013) 4. Davis, F.D.: A technology acceptance model for empirically testing new end-user information systems: Theory and results. Doctoral dissertation, Sloan School of Management, Massachusetts Institute of Technology (1986) 5. Lee, Y., Kozar, K.A., Larsen, K.R.: The technology acceptance model: Past, present, and future. Commun. Assoc. Inf. Syst. 12(1), 50 (2003)

48

C.R. Yeh and S.-Y. Hsiao

6. Wangpipatwong, S., Chutimaskul, W., Papasratorn, B.: Understanding citizen’s continuance intention to use e-government website: A composite view of technology acceptance model and computer self-efficacy. Electron. J. e-Gov. 6(1), 55–64 (2008) 7. Fishbein, M., Ajzen, I.: Belief, Attitude, Intention and Behavior: An Introduction to Theory and Research. Addison-Wesley, Reading (1975) 8. Venkatesh, V., Morris, M.G., Davis, G.B., Davis, F.D.: User acceptance of information technology: Toward a unified view. MIS Q. 27(3), 425–478 (2003) 9. Vijayasarathy, L.R.: Predicting consumer intentions to use on-line shopping: the case for an augmented technology acceptance model. Inf. Manage. 41(6), 747–762 (2004) 10. Overby, E., Konsynski, B.: Task-technology fit and process virtualization theory: An integrated model and empirical test. Emory Public Law Research Paper (10–96) (2010) 11. Schmitt, B., Zarantonello, L., Brakus, J.: Brand experience: what is it? How is it measured? Does it affect loyalty? J. Mark. 73(3), 52–68 (2009) 12. Lin, Y., Lin, G.Y., Laffey, J.M.: Building a social and motivational framework for understanding satisfaction in online learning. J. Educ. Comput. Res. 38(1), 1–27 (2008) 13. Culiberg, B., Rojšek, I.: Identifying service quality dimensions as antecedents to customer satisfaction in retail banking. Econ. Bus. Rev. 12(3), 151–166 (2011) 14. Jun, M., Yang, Z., Kim, D.: Customers’ perceptions of online retailing service quality and their satisfaction. Int. J. Qual. Reliab. Manage. 21(8), 817–840 (2004) 15. Ho, C.I., Lee, Y.L.: The development of an e-travel service quality scale. Tour. Manag. 28(6), 1434–1449 (2007) 16. Hu, P.J., Chau, P.Y., Sheng, O.R.L., Tam, K.Y.: Examining the technology acceptance model using physician acceptance of telemedicine technology. J. Manage. Inf. Syst. 16(2), 91–112 (1999) 17. Sun, H., Zhang, P.: The role of moderating factors in user technology acceptance. Int. J. Hum. Comput. Stud. 64(2), 53–78 (2006)

Knowledge Sharing

Virtual Teams Stress Experiment Proposal: Investigating the Effect of Cohesion, Challenge, and Hindrance on Knowledge Sharing, Satisfaction, and Performance Andree E. Widjaja ✉ (

)

Department of Information Systems, Pelita Harapan University, Tangerang, Indonesia [email protected]

Abstract. This current paper will present a research proposal with regard to the effect of stress on virtual teams. The proposed research will specifically concen‐ trate on the virtual teams’ stress as double constructs (positive stress as the chal‐ lenge and negative stress as the hindrance), cohesion as social support to mitigate the stress, and how these three factors will affect the virtual teams’ knowledge sharing, satisfaction, and performances. A study which incorporates an experi‐ mental research methodology will be specifically proposed to test the aforemen‐ tioned relationships. Related literature reviews on stress, hypotheses develop‐ ment, as well as experimental research design will be mentioned in the paper. Keywords: Virtual Teams · Stress · Challenge · Hindrance · Knowledge sharing · Experimental research design

1

Introduction

In recent years, Virtual Teams (VT) has been widely incorporated by many organizations as an effective way to gain competitive advantage over its competitors. VT itself can be broadly defined as the teams in which its members are interacting, working, and collab‐ orating together using communication and information technology. Working in VT indeed has many advantages, such as the teams can collaborate together any time (flex‐ ibility in time), in any different geographical regions (flexibility in proximity), and use various means of technology (e.g., video conferencing, collaboration software, and many others). The context of VT is also varied, for instance, the use of VT in the organ‐ izations or businesses, projects, as well as in higher education. Although a VT indeed possesses numerous advantages, it is not without problems [1]. Various VT problems have been widely discussed in numerous literature, ranging from trust [2, 3], communication [4], cohesion [5], conflicts [6], to many other “social factors” [1]. Nevertheless, to the best of our knowledge, one salient social factor which has rarely been discussed in the VT related literature is concerning VT stress. Actually, prior studies mentioned a stress related technology so called “Techno Stress” and “Cyber Stress”. This specific type of stress has been widely investigated in the Information Systems as well as Management related literature [7–10]. Such terms were introduced since working with the information technology would tend to intensify work, hence it © Springer International Publishing AG 2017 L. Uden et al. (Eds.): KMO 2017, CCIS 731, pp. 51–63, 2017. DOI: 10.1007/978-3-319-62698-7_5

52

A.E. Widjaja

could create numerous stressful as well as unhealthy working conditions [11]. As the Internet has made online collaborations possible (in form of the VT), thus the stress related technology nowadays would have become much more complex than ever before. For that reason, we would argue that further investigation in this matter is important and necessary, particularly within VT context. Meanwhile, researchers have conflicting opinions concerning stress [12], thereby causing the variety definitions of stress [13]. The conceptualizations of stress are also varied, depending on how the stress is actually treated. Furthermore, there were some inconsistencies found in the prior stress related works and findings [14]. For instance, Cavanaugh [15] asserted that not all stress are negative. Cavanaugh’s [15] study distin‐ guished two kinds of stress, they are: positive stress which is defined as a challenge, and negative stress which is defined as the hindrance. Positive stress can result in a compet‐ itive edge and force positive change [16]. The cause of stress, which later is called as the stressor would be appraised by the individual whether as the challenge or the hindrance. The type of stressors is therefore closely associated with the challenge or the hindrance. For example, some stressors such as time constraint and tasks load are considered as the challenge. Meanwhile, stressors such as resource inadequacy and roletasks clarity are referred as the hindrance. 1.1 Objective The main objective of this current paper is to propose a VT stress experiment research proposal by applying stress as double constructs as suggested by Cavanaugh [15]. Meanwhile, social supports have also play important role to buffer the stress. One example of social supports pertinent in VT is a team’s cohesion. Even though the “rela‐ tionship” factor such as cohesion has been found in many social supports as well as the stress related literature, still little is known of how cohesion will actually affect challenge and hindrance perceived stress within the VT context. Meanwhile, knowledge sharing is also crucial affecting the VT effectiveness. Other factors such as satisfaction and performance will also be the significant contributors concerning whether the VT can work effectively or not. However, it is also still unclear how the challenge and hindrance would actually influence the VT knowledge sharing, satisfaction, and performance. This proposed study will therefore concentrate on the two research questions: First, “how the challenge and hindrance will influence the effectiveness of virtual teams, measured by knowledge sharing, satisfaction, and performance?” And the second, “how social support (cohesion) will affect the challenge and hindrance perceived stress?”An experimental study on VT stress which will manipulate both stressors (challenge and hindrance) and how these stressors affect VT’s knowledge sharing, satisfaction, and performance will become the primary contribution of this current paper. The second contribution will be regarding the role of social support as a team’s cohesion acting as a buffer of the perceived stress in VT. This current paper will be organized by firstly introducing the theoretical framework. Then secondly, hypotheses development will be discussed and be based on several factors mentioned in the VT related literature. The last sections will discuss the proposed experimental procedures, proposed data analysis, discussions, and future works.

Virtual Teams Stress Experiment Proposal

2

53

Theoretical Framework

2.1 Understanding Stress Stress is inevitable in one’s life [17]. Some stressful events might have large impact on individuals, while some stressful events are trivial. Some of these are controllable, and some are not [17]. Scholars have agreed about the subjectivity nature of stress since the stress is perceived by each individual. Stress can be defined as “when the perceived pressure exceeds your perceived ability to cope” [18]. Individual may perceive stress differently; one can feel stress, yet others may not. In other words, one person may feel the situation as stressful, while others as enjoyable [19]. There are many factors that can cause stress. Educators realized that collaboration can cause stress [20]. Working environment can induce a job stress in which prolonged exposure to stressors incurred at work can closely linked to physical illness, psycho‐ logical dysfunction such as anxiety and depression, job dissatisfaction, as well as the job performance [21]. Uncertainty factors such as lack of rules, procedures, and lack of information can cause stress. Meanwhile, task load and time pressures can also cause stress [22]. Stress can affect teams in many other ways as well. Prior studies showed the effect of stress on teams performance [23, 24]. For instance, Driskell, Salas, and Johnston [22] examined social behavior in teams decision making. Based on the examples from the industry, they suggested that stress can reduce the group focus which is necessary for maintaining proper crew coordination and situational awareness [22]. 2.2 Stress Model Researchers have divergent opinions concerning stress [12], hence introducing numerous definitions of stress in the literature [13]. However, the process of stress is best distinguished by two components, they are: stressor and strains. Stressors are the demands, constraints, opportunities, or challenges which are unique to an individual and that may or may not lead to strain [25]. In other words, the stressors are the stimuli that evoke the stress process, whereas the strains are the outcomes in this process [14]. If a stimulus is perceived by an individual as a stressor and is not effectively coped with, the strain (negative emotional, physiological, and or behavioral response) will be resulted. Stressor and strains concept is consistent with Stokes and Kite [13] review of stress and evolution in which they suggested the two traditional models of psychological stress, they are: stimulus-based and response-based. Stimulus based stress approach assumes certain conditions to be stressful (e.g. work load, heat and cold, time pressure, etc.). This is similar as stressor in the previous stress process aforementioned. Many researchers have selected this approach as the exogenous variables and done various experiments. Meanwhile, response based approach stress or strain is defined by the pattern of response (behavioral, cognitive, affective) resulted from exposure to a given stressor [12, 13]. The response variables can be considered endogenous or coming from within individual.

54

A.E. Widjaja

Stress as Double Constructs. It is a general assumption that stress is always being referred negatively [22]. However, many researchers have also argued that stress is not always negative (bad), implying that there is a positive (good) stress. Early literature suggests that stress can be separated into two categories: Eustress (good stress) and distress (bad stress). Eustress is defined as “stress that creates challenge and feelings of fulfillment or achievement and it was found to be a positive motivating force”. Conversely, distress is “a stress that is not accompanied by challenge or feelings of fulfillment or achievement” [26]. Perceived stress can be recognized through a self-report stress and it can be associ‐ ated with both positive and negative outcome. Drawing from Lazarus’s [27] study, selfreported work stress associated with some stressors may result in negative outcome, whereas self-reported work stress associated with other stressors may result in positive outcomes [15]. Hence in Cavanaugh’s [15] study, the two types of stress were intro‐ duced, positive stress referred as the challenge and negative stress as the hindrance. By distinguishing challenge and hindrance, they argued that it would increase our under‐ standing on self-reported work stress [15]. Challenge is defined as “work-related demands or circumstances that, although potentially stressful, have associated potential gains for individuals.” Meanwhile, hindrance is defined as “work-related demands or circumstances that tend to constrain or interfere with an individual’s work achievement and that do not tend to be associated with potential gains for the individual” [15]. The distinction of challenge and hindrance should be based on type of demand rather than level of demand [26]. In other words, there are some types of stressors that will determine challenge, while other type of stressors will determine hindrance. For instance, stressors appraised as challenge may include workload, job/role demands, job complexity, job scope, high responsibility, and time pressures [14, 15, 27, 28]. Challenging job demands or work circumstances produce productive positive feeling, though it may be stressful. Meanwhile, stressors (stimuli that place demands on individuals) are appraised as hindrances such as resource inadequacy, role ambiguity, role and interpersonal conflict, role dissensus, role overload, role interference, role strain, role clarity, supervisor-related stress, hassles, red tape, organizational politics, and concerns about job security [14, 15, 27]. Prior studies showed that Challenge will be associated with positive work outcomes, whereas Hindrance with the negative work outcomes. Cavanaugh’s [15] study results shown that challenge related self-reported stress was related to job satisfaction. Conversely, hindrance related self-reported stress was negatively related to job satis‐ faction and positively related to job search as well as voluntary turnover. Consistent with Cavanaugh’s [15] research, Lepine [14] also confirmed the hindrance stressors had a negative direct effect on performance, whereas challenge stressors had a positive direct effect on performance.

Virtual Teams Stress Experiment Proposal

55

2.3 Social Supports Social supports play an important role and considered as a key resource in the stress [19]. Social support is defined as “the availability helping relationships and the quality of those relationships” [29]. Source of social supports can come from many people such as co-workers, supervisors, family and friends, and other people at work. The kind of supports are varied from non-tangible to tangible [30]. House [31] argued that social supports involves the exchange of resources at least two persons with the purpose of helping the person who receives the supports. It may involve providing the empathy, care, love, and trust (emotional support); aid in time, money, energy (instrumental support); information relevant to self-evaluation (appraisal support), and advice information and suggestion (informational support). However, the social supports are actually perceived by the people who receive those supports. According to Solberg’s [32] study, perceived social supports moderated the relationship between stress and distress, so that those who have higher perception of social support would have lower distress rating. 2.4 Hypotheses Developments Time Constraint. Time manipulation is a stressor that can cause stress [12]. Time stressor is considered as a challenge which can improve both performance and satis‐ faction [14, 15]. The shorter time constraint can induce more stress since VT may perceive that they may have limited time to finish their task. This limitation of time can be perceived as challenge. For instance, by quickly finishing their task VT member may be forced to exert their knowledge, efforts, and abilities in such limited time. In this proposed study, we will manipulate the time constraint as stressor. Thus, we expect that: H1: Shorter time constraint will increase challenge perceived stress. Tasks Level. Tasks level or job overload can also cause challenge or positive stress [14, 15]. With so many tasks, VT may perceive the tasks as challenging. VT may divide the tasks and put more efforts necessary in order to comply with the goals. In doing so, by fulfilling such “difficult” challenge, VT may enhance their satisfaction, thus increasing their performance at the end. We will manipulate the tasks level as stressor. Hence, we will propose the following hypothesis: H2: More difficult tasks will increase challenge perceived stress. Teams Size. One of the examples of hindrance given in the literature is resource inad‐ equacy. In this study, we will choose teams’ size to represent resource inadequacy. VT should have adequate resource in order to perform their job better. Thus, lack of members may cause negative stress since VT will not have enough assistance, knowledge, or supports from their members to do the tasks. Based on this, we will propose: H3: Smaller teams size will increase hindrance perceived stress.

56

A.E. Widjaja

Role Task Clarity. Role task clarity is given as one example of hindrance or negative stress. In this study, we will manipulate the role task clarity as stressor. VT should be given clear objectives and job roles in the beginning of the task. Lack of role task clarity will cause VT experiencing negative stress. Moreover, unclear job roles and responsi‐ bility may cause another problem as some jobs or roles overlaps or perhaps can cause many confusions. Thus, we expect that: H4: Unclear role and task will increase hindrance perceived stress. Cohesion. Cohesion is an important factor in VT [5]. Teams will work better when there is cohesion [33]. Cohesion can induce commitment which is built through suppor‐ tive activities, and responding or recognizing the work of others. In VT, cohesion is considered as social supports which can buffer the stress effect. We argue that cohesion can suppress both challenge and hindrance perceived stress. Without cohesion, VT may experience more stress. In hindrance stressful environments (negative stressors – smaller teams size, unclear role-tasks), cohesion is likely to decrease the hindrance perceived stress. By having a good relationship among VT members, unclear role-task and inad‐ equate teams member resources may be better handled. In addition, we argued that the stronger cohesion may also decrease challenge perceived stress. The VT who is working closely together with an established relationship will help the teams to cope with stressful environments. Hence, we expect that: H5a: Under shorter time constraint and difficult task, higher cohesion will decrease challenge perceived stress; H5b: Under smaller team size and unclear role-tasks, higher cohesion will decrease hindrance perceived stress. Knowledge Sharing. Within VT environment, knowledge sharing may include members’ interactions via e-mails, telephone, instant messaging, text messaging, elec‐ tronic bulletin, boards and discussion forums, dedicated web pages, and etc. [34]. By sharing the knowledge, it would contribute to the VT effectiveness. Knowledge sharing is indispensable in VT. In VT, knowledge sharing is based more on the benefit than the cost. We would argue that when individuals work as a team, they would share the same goal in which that goal should be achieved by working and contributing together. Under challenge stress condition, the teams will try to contribute and share their knowledge because they may feel of being challenged to contribute more. On the other hand, hindrance stress condition, such as team’s size and unclear role-task, the members will be reluctant to share their knowledge because they simply don’t have adequate resource or perhaps they may be confused on sharing what kind of knowledge. Satisfaction and Performance. Teams’ performance is generally assessed through efficiency measures (e.g. decision time), effectiveness measures (e.g. decision quality), satisfaction measures (e.g. satisfaction with decision process and outcome), participa‐ tion and consensus [35]. Teams will not and cannot be effective if their members are not satisfied with the way the teams functioning [36]. Therefore, the more satisfaction the teams have, the better their performance will be. Under challenge stress condition, the teams will have better satisfaction and performance [14] because, though stressful, the teams is challenged to working at their best. It will lead to better satisfaction and team’s performance at the end. On the other hand, under hindrance stress condition, the

Virtual Teams Stress Experiment Proposal

57

satisfaction and performance will decrease since they have lack of resource as well as uncertainty regarding their tasks and roles. Based on those arguments, we will propose several hypotheses: H6a: Knowledge sharing is higher under shorter time constraint and difficult task; H6b: Virtual teams performance is higher under shorter time constraint and difficult task; H6c: Virtual teams satisfaction is higher under shorter time constraint and difficult task; H7a: Knowledge sharing is lower under smaller teams size and unclear role-tasks; H7b: Virtual teams performance is lower under smaller teams size and unclear role-tasks; H7c: Virtual teams satisfaction is lower under smaller teams size and unclear role-tasks; H8: Higher VT satisfaction will increase VT Performance; H9: Higher knowledge sharing will increase VT performance. Challenge will have more positive effect in knowledge sharing, satisfaction, and performance than hindrance. Therefore, we expect that: H10a: Knowledge sharing is higher under shorter time constraint and difficult task than in under smaller teams size and unclear role tasks; H10b: Virtual teams performance is higher under shorter time constraint and difficult task than in under smaller teams size and unclear role tasks;H10c: Virtual teams satisfaction is higher under shorter time constraint and diffi‐ cult task than in under smaller teams size and unclear role tasks. Based on the literature review and hypotheses development mentioned previously, the following figure describes the research model of this proposed study (Figs. 1 and 2).

Fig. 1. Stressors experiment in VT research model

58

A.E. Widjaja

Fig. 2. Hypotheses in the research model

3

Research Methodology

3.1 Experimental Design This study will use 2 × 2× 2× 2 full factorial design. There will be four variables manipulated to two different levels, high and low. Due to the complexities of the study and high possibility of the scarcity of the participants, there will be no independent control group assigned (Table 1).

Virtual Teams Stress Experiment Proposal

59

Table 1. Full factorial experimental design Teams Size Hindrance Challenge

Difficult

Not Clear

Easy

Task

Small

Time

Big Role-Task Clarity Clear Not Clear

Clear

Short

1

2

3

4

Long

5

6

7

8

Short

9

10

11

12

Long

13

14

15

16

Ethical Approval. Prior conducting the experiment, this study will follow ethical guidelines standards for using human subjects in a social experimental research. A specific proposal with regard to the ethical considerations of this study will be proposed to Internal Review Board (IRB) member of the University. After getting the approval given from the IRB members, the experiment process will be commenced. Tasks Descriptions. The participants in this proposed study will be University students who are taking Information Systems and Management related courses. The task for this experiment will be a software development case study. The case study will be mainly about the problems of Management Information Systems (MIS) in a fictional company setting. The VT should come out with their own solutions in order to overcome such problems, for instance by developing some new software applications. Since we will use problem solving case study, hence knowledge sharing will be crucial to collect the required knowledge in developing software applications. For example, one student may expert in doing systems design, thus he or she will share the knowledge in designing a good system. Others, who are excellent in programming, they may share the advice regarding the feasibilities in transforming the systems designed into the programming codes. The case study will be given to each VT in the beginning of the task. Each VT will have to form the teams consisting of 3 and 8 members. We consider the teams with 3 members as a small team, whereas the one with 8 members is considered as a big team. The quantity difference of the number of members between the small and big teams is 5 members. Hence, we would argue that this categorization of big and small teams is appropriate as the difference is quite apparent. The grouping will be randomly assigned by the researchers. It will be expected to have at least 32 teams with equal numbers of 3 and 8 members. In addition, it will also be expected that at least three different Univer‐ sities will be participated in this study, in which each university will be better if they are geographically separated.

60

A.E. Widjaja

To ensure the teams will interact virtually, the maximum of two VT members who belong to the same University will be grouped together, while the rest should come from different Universities, in whom they haven’t met each other. A web-based collaboration system will be used to let the teams collaborate with their members. Aside from using the systems, the VT members will be allowed to use various technologies such as messaging, teleconferencing, e-mail, or any computer mediated communication. In accordance with the ethical principle, all participants will be told that they are going to participate in a VT experiment. It is mandatory that each participant should agree on the informed consent given prior their participation in the experiment. The following are our proposed operationalization of the experimental design: • Time limit: – Short – 12 weeks – Long – 16 weeks • Task Level: – Difficult Hard case study with various issues and problems There are different and various requirements to overcome the problems Comprehensive and detailed documentations Progress report and once every two weeks should do online presentation in front of the instructors – Easy Simple case study with minor problems Simple requirements Adequate documentations • Teams size: – Small: 3 people – Big: 8 people • Role and Task Clarity – Clear: Specific guidance, directions, and objective will be provided. E.g., users’ requirements and designs Role is clear, e.g. who will become project leader, programmer, tester, with specific jobs responsibility – Not Clear: General guidance and objective – students have to think by themselves No Roles assigned Experimental Procedures. Experimental procedures in this study are the following: • • • • •

Preparing the tasks Teams formation (randomly chosen) Plotting the teams formation to experimental groups Demographic questionnaires will be administered Case study assigned

Virtual Teams Stress Experiment Proposal

61

• There will be three measurements (T1, T2, and T3) for the variables. The variables will include cohesion, challenge, hindrance, satisfaction, performance, and knowl‐ edge sharing. • For short time group, T1, T2, and T3 will be measured after week 3rd, 7th, 12th. While for long time group, T1, T2, and T3 will be measured after week 5th, 10th, and 16th. • Software delivery, Documentation submitted – project finished • Grading process Measurement Items. According to Matteson and Ivanchevich [37], self-reported measures of perceived stress would be better adapted for assessment method due to its easiness and administrations. The self-reported work stress, both challenge and hindrance will be adapted from Cavanaugh, Boswell, Roehling and Boudreau [15]. Cohesion questionnaire items will be adapted from Lin, Standing and Liu [38]. Knowl‐ edge sharing questionnaire items will be adapted from Bock, Zmud, Kim and Lee [39]. Satisfaction questionnaire items will be adapted from Lin, Standing and Liu [38] and Chidambaram [40]. Performance questionnaire items will be adapted from Lurey and Raisinghani [36] and Amy [41]. All questionnaire items will be administered using 7 points Likert scale and modified accordingly to fit this study. The software application and documentation will be graded by several instructors independently. General grading criteria will be based primarily on the functionality as well as the design of the software, for example whether the software is functioning properly or not, documentations quality, and other relevant or related measurements.

4

Proposed Data Analysis

After all data from experiment is gathered, the data needs to be analyzed accordingly. We will use SPSS software to analyze our data. Several statistical analyses will be performed, such as descriptive analysis, group comparison of means (ANOVA), cluster analysis, discriminant analysis, and multiple regression analysis. All hypotheses will be tested (whether the hypotheses will be supported or not) based on the given statistical analysis results. The rule of thumbs with regard to the statistical test significant criteria will be employed accordingly.

5

Discussions and Future Works

Working with technology, especially in VT might not only create many problems, but also might induce some stresses. To the best of our knowledge, the issue of the team members’ stress within VT context has rarely been found in the existing literature. In this paper, we have proposed an experimental research design and several hypotheses in which we have carefully developed based on the stress and VT related literature to answer the two important research questions: First, “how the challenge and hindrance will influence the effectiveness of virtual teams, measured by knowledge sharing, satis‐ faction, and performance?”And the second, “how social support (cohesion) will affect the challenge and hindrance perceived stress?” Future work is strongly suggested to

62

A.E. Widjaja

incorporate our experimental research design to gain more insights on the real effect of team members’ stress on VT. It is hoped that by having better understanding of the importance of stress and its effect on the VT, we are able to better design the appropriate tasks, procedures, and supporting technologies which in turns, can further significantly improve the overall VT’s knowledge sharing, effectiveness, and performance.

References 1. Levi, D.: Group Dynamics for Teams. Sage Publication, Thousand Oaks (2007) 2. Javernpaa, S.L., Leidner, D.E.: Communication and trust in global virtual teams. Organ. Sci. 10, 791–815 (1999) 3. Iacono, C.S., Weisband, S.: Developing trust in virtual teams. In: Proceedings of the 30th Hawaii International Conference on System Sciences, pp. 412–420. IEEE Computer Society Press (1997) 4. Bergiel, B.J., Bergiel, E.B., Balsmeier, P.W.: Nature of virtual teams: a summary of their advantages and disadvantages. Ind. Manag. Data Syst. 31, 99–110 (2008) 5. Salisbury, W.D., Carte, T.A., Chidambaram, L.: Cohesion in virtual teams: validating the perceived cohesion scale in a distributed setting. ACM SIGMIS Database 37, 147–155 (2006) 6. Kankanhalli, A., Tan, B.C.Y., Wei, K.-K.: Conflict and performance in global virtual teams. J. Manag. Inf. Syst. 23, 237–274 (2006) 7. Gardner, W., Scermerhorn, J.: Computer networks and the changing nature of managerial work. Public Product. Rev. 11, 85–89 (1988) 8. Duxbury, L.E., Higgins, C.A., Irving, R.H.: Attitudes of managers and employees to telecommuting. Infor 25, 273–285 (1987) 9. Crouch, M.-L., Montecino, V.: Teaching in Communities Online Conference (1997) 10. Ayyagari, R., Grover, V., Purvis, R.: Technostress: technological antecedents and implications. Manag. Inf. Syst. Q. 35, 831–858 (2011) 11. Carayon, P., Haims, M.C., Hoonakker, P.L.T., Swanson, N.G.: Teamswork and musculosketal health in the context of work organization interventions in office and computer work. Theor. Issues Ergon. Sci. 7, 36–69 (2006) 12. Staal, M.A.: Stress, Cognition and Human Performance: A Literature Review and Conceptual Framework. Ames Research Center, Moffett Field (2004) 13. Stokes, A.F., Kite, K.: On grasping a nettle and becoming emotional. In: Hancock, D. (ed.) Stress, Workload, and Fatigue. L. Erlbaum, Mahwah (2001) 14. Lepine, J.A., Podsakoff, N.P., Lepine, M.A.: A meta-analytic test of the challenge stressorhindrance stressor framework: An explanation for inconsistent relationship among stressor and performance. Acad. Manag. J. 48, 764–775 (2005) 15. Cavanaugh, M.A., Boswell, W.R., Roehling, M.V., Boudreau, J.W.: An empirical examination of self-reported work stress among U.S. managers. J. Appl. Psychol. 85, 65–74 (2000) 16. Marino, S.: The stress epidemic. Industry Week 246 (1997) 17. Cheng, C., Cheung, M.W.L.: Cognitive processes underlying coping flexibility: differentiation and integration. J. Pers. 73, 860–886 (2005) 18. Palmer, S., Cooper, C.: Creating a Balance: Managing Stress. British Library, London (2003) 19. Zwicker, A., DeLongis, A.: Gender, stress, and coping. In: Chrisler, J.C., McCreary, D.R. (eds.) Handbook of Gender Research in Psychology. Springer, Vancouver (2010) 20. Palloff, R., Pratt, K.: Building Learning Communities in Cyberspace. Josey-Bass, San Francisco (1999)

Virtual Teams Stress Experiment Proposal

63

21. Jimmieson, N.L., McKimmie, B.M., Hannam, R.L., Gallagher, J.: An investigation of the stress-buffering effects of social support in the occupational stress process as a function of teams identification. Group Dyn. Theory Res. Pract. 14, 350–367 (2010) 22. Driskell, J.E., Salas, E.: Does stress lead to a loss of teams perspective? Group Dyn. Theory Res. Pract. 3, 291–302 (1999) 23. Karau, S.J., Kelly, J.R.: The effects of time scarcity and time abundance on group performance quality and interaction process. J. Exp. Soc. Psychol. 28, 542–571 (1992) 24. Cannon-Bowers, J.A., Tannenbaum, S.I., Salas, E., Volpe, C.E.: Defining competencies and establishing teams training requirements. In: Guzzo, R.A., Salas, E. (eds.) Teams Effectiveness and Decision Making in Organizations, pp. 333–380. Jossey-Bass, San Francisco (1995) 25. Glazer, S.: Cross cultural issues in stress and burnout. In: Handbook of Stress and Burnout in Health Care, pp. 79–93. Nova Science Publisher, Huntington (2008) 26. Selye, H.: History and present status of the stress concept. In: Goldberger, L., Breznitz, S. (eds.) Handbook of Stress, pp. 7–17. Free Press, New York (1982) 27. Lazarus, R., Folkman, S.: Stress, Appraisal, and Coping. Springer, New York (1984) 28. McCauley, C.D., Ruderman, M.N., Ohlott, P.J., Morrow, J.E.: Assesing the developmental components of managerial jobs. J. Appl. Psychol. 79, 3–14 (1994) 29. Leavy, R.L.: Social support and psychological disorder: a review. J. Community Psychol. 11, 3–21 (1983) 30. Viswesvaran, C., Sanchez, J.I., Fisher, J.: The role of social support in the process of work stress: a meta analysis. J. Vocat. Behav. 54, 314–334 (1998) 31. House, J.S.: Work Stress and Social Support. Addison-Wesley, Reading (1981) 32. Solberg, V.S.: Examination of self-efficacy, social support, and stress as predictors of psychological and physical distress among Hispanic college students. Hispanic J. Behav. Sci. 19, 182–201 (1997) 33. Yoo, Y., Alavi, M.: Media and group cohesion: relative influence on social presence, task participation, and group consesus. MIS Q. 25, 371–390 (2001) 34. Rosen, B., Furst, S., Blackburn, R.: Overcoming barriers to knowledge sharing in virtual teams. Org. Dyn. 36, 259–273 (2007) 35. Paul, S., Seetharaman, P., Samarah, I., Mykytyn, P.P.: Impact of heterogeneity and collaborative conflict management style on the performance of synchronous global virtual teamss. Inf. Manag. 41, 303–321 (2003) 36. Lurey, J.S., Raisinghani, M.S.: An empirical study of best practices in virtual teamss. Inf. Manag. 38, 523–544 (2001) 37. Matteson, M.T., Ivanchevich, J.M.: Controlling Work Stress: Effective Human Resource and Management Strategies. Jossey-Bass Publishers, San Francisco (1987) 38. Lin, C., Standing, C., Liu, Y.-C.: A model to develop effective virtual teams. Decis. Support Syst. 45, 1031–1045 (2008) 39. Bock, W.G., Zmud, W.R., Kim, G.Y., Lee, N.J.: Behavioral intention formation in knowledge sharing: examining the toles of extrinsic motivators, social-psychological forces, and organizational climate. MIS Q. 29, 87–112 (2005) 40. Chidambaram, L.: Relational development in computer-supported groups. MIS Q. 20, 143– 165 (1996) 41. Amy, E.: Psychological safety and learning behavior in work teams. Adm. Sci. Q. 44, 350– 383 (1999)

Knowledge Sharing on YouTube Eric Kin Wai Lau ✉ (

)

Department of Marketing, City University of Hong Kong, Kowloon Tong, Hong Kong [email protected]

Abstract. The purpose of the present study is to focus on knowledge sharing by posting video clips in YouTube and its effectiveness. YouTube was chosen as the knowledge transfer channel as it is a popular social media network that enables free interactions among registered users about the particular video clip available. Different video contents were manipulatively created, uploaded and their popu‐ larity was tested on YouTube. Video counts are based on YouTube statistics. Using real examples collected on YouTube, several factors were identified as the reasons for the popularity of the video clips; i.e., topic selected, keywords used and urgency of the topic. This paper helps people to develop their video clips on YouTube and to create successful video marketing campaigns on YouTube. Keywords: Social media · YouTube · Web 2.0 · User participation

1

Introduction

YouTube was launched in 2005 and became a new social medium which hosted video content. YouTube was founded by Chad Hurley, Steve Chen and Jawed Karim. The name “YouTube” means to “broadcast yourself” and it became an important video sharing platform on the Internet. Registered users can share their views about video clips freely and share the particular video clip in other social media platforms, such as Facebook. It is a kind of social activity and interpersonal interaction [6]. Using real examples collected from YouTube, Dynel [4] pointed out that forms of interaction and levels of communication on YouTube and traditional television are different. YouTube users are relatively active in the communication process, while television viewers are passive and simply the recipients of information being broadcast on TV channels. Active participation is common on YouTube and YouTube users can interact with other users. People’s communication roles on YouTube (i.e., “vlogging YouTubers”) can be as senders and recipients at the same time.

2

Previous Empirical Studies on YouTube

People are linked with social media networks (SMN). They can now interact with others voluntarily and freely. Smith [11] described the growth of social access as a “revolution” with the help of Web 2.0 and user-driven IT. He also pointed out that voluntarily sharing opinions in social media has become mainstream and creates a lot of marketing

© Springer International Publishing AG 2017 L. Uden et al. (Eds.): KMO 2017, CCIS 731, pp. 64–71, 2017. DOI: 10.1007/978-3-319-62698-7_6

Knowledge Sharing on YouTube

65

opportunities. YouTube creates a community of its active users, who are known as “YouTubers”. Originally, video was just one-way communication. However, social media networks enable interactions among users and have turned video into a two-way communication channel. People can engage on YouTube by liking, disliking and commenting. Engagement with social networking sites (SNS) engenders “social connectedness”. A sense of social support is important from the concept of SNS. Alloway and Alloway [2] studied the impact of SNS on working memory and attentional control. They examined 284 university students using online tests and found significant differences between active and passive SNS engagement on working memory skills and attentional control. Interestingly, they found that higher self-reported levels of connect‐ edness were significantly associated with Facebook use (i.e., limited to existing personal friend circles), but not with Twitter or YouTube use (i.e., people who are publicly accessible). They concluded that users are not able to enjoy a sense of belonging from Twitter and YouTube. Oh and Syn [8] tested 10 motivation factors that motivate people to participate in social media. These influential factors were enjoyment, self-efficacy, learning, personal gain, altruism, empathy, community interest, social engagement, reputation, and reci‐ procity. In a sample of 1056 social media users from five different social media sites (i.e., Facebook, Twitter, Delicious, YouTube, and Flickr), they found that learning is the most important motivator and social engagement is the second most important moti‐ vator for people sharing information on social media. In addition, they also found that female users are more highly motivated by personal gain, community interest, and social engagement than male users. Haridakis and Hanson [6] tested communication motives and users’ backgrounds that predict the viewing and sharing of YouTube video content. They examined a sample of 427 American university students and tested for (a) social and psychological antece‐ dents (i.e., locus of control, innovativeness, sensation seeking, interpersonal interaction, and social activities), (b) motives for using YouTube, (c) affinity with YouTube, and (d) amount of time spent watching YouTube videos and sharing them with each other. Multiple regression analysis was conducted and found that socially active users tended to share YouTube videos for purposes of social interaction, co-viewing, and convenient entertainment. In addition, they also found a gender effect in the use of YouTube; i.e., socially active males used YouTube more than females. Ahn and Shun [1] proposed that empathy drives people’s participation in visual media and video games. They examined a sample of 300 Korean adults and found that time spent with video media (both television and film) was positively correlated with empathy and social connectedness. However, time spent on gaming media was nega‐ tively associated with empathy. Shepherd [10] suggested a new trend of self-promotion/self-marketing with the help of social media. It allows an individual to develop his/her own brand presence and engage with other people. YouTube enables an individual’s social media marketing campaign in which compelling video content can create awareness and attention from other people. As a result, a lot of key opinion leaders are created by YouTube channels. Dehghani, Niaki, Ramezani and Sali [3] tested four dimensions with regard to YouTube advertising (i.e., entertainment, informativeness, customization, and irritation) relating

66

E.K.W. Lau

to advertising value, brand awareness and purchase intention. Using a sample of 323 European students, they found that perceived utility of entertainment, informativeness, and customization of YouTube advertising are positively associated with advertising value. As expected, the perceived utility of irritation of YouTube advertising is nega‐ tively associated with advertising value. They also identified that advertising value through YouTube is positively associated with brand awareness. In addition, brand awareness through YouTube is positively associated with purchase intention. In addition, YouTube is a social media platform that allows nonverbal communica‐ tion [7]. More so than verbal cues, facial expressions, gestures, eye contact, posture, and tone of voice can convey an idea in YouTube video clips. Lewinski investigated the campaigns of large European banks on YouTube and found that the lack of facial emotions and the presence of facial non-emotions affected the popularity of those selected video campaigns on YouTube [7]. Facial expressions of happiness, sadness, and surprise were significantly associated with the number of video views.

3

Research Question

How do people feel, behave and think on YouTube and why? Imagine people can now produce and upload their video clips on YouTube and let others view, share and comment, for free. Hundreds, thousands, or millions of views can result. It becomes the mainstream media and another way to exchange knowledge. Importantly, a lot of users either receive the video link directly from their social circle or from searching for video clips they want to watch using a keyword search in YouTube. However, video content is the key. The important research question is what constitutes correct and good video content that can create awareness on YouTube?

4

The Case Study

The popularity of video uploading on YouTube is the focus of the present study. A YouTube channel was launched in Sept 2016 to examine the effectiveness of different video topics (i.e., number of hits) and examine how users engaged (i.e., shared and commented). The following information technology was used in the video production: • Camera: Canon 6D • Editing software: CyberLink PowerDirector 15 • Computer: Dell Desktop Computer with Intel i7-4790 CPU, 8 GB RAM, 256 MB SSD Table 1 provides a sample of video clips uploaded to the YouTube channel (the complete list is in Appendix I). After uploading to the YouTube channel, they were then announced and promoted on Facebook and made available on YouTube by using keywords in a search engine. We needed to quantitatively measure users’ engagement on YouTube and their behaviours. Video counts are based on the statistics provided by YouTube (Fig. 1). In total, the YouTube channel triggered much engagement; some

Knowledge Sharing on YouTube

67

users even left video commentaries to express their feelings on the posts which received about 27,299 views within two months. Table 1. The sample list of video clips created and tested in the study The video topic Japan Tokyo fish market consumer behaviour Japan Tokyo sensoji Japan Tokyo tour part 1 Japan Tokyo tour part 2 I can’t live without GPS in Japan My favourite restaurant in Tokyo Webchat Mooncake Uber BuyBuyBuy 2A3A Saturday Pricing Shopping Cheating stories Fishmarket No more bank statements Government and property developers Donald v Hillary Losing money

Category Tour Tour Tour Tour Tour Tour Personal sharing Personal sharing Personal sharing Personal sharing Buying blog Personal sharing Teaching – marketing Buying blog News Tour Teaching – marketing News News Buying blog

Uploaded date 21 Sept 2016 21 Sept 2016 21 Sept 2016 21 Sept 2016 21 Sept 2016 21 Sept 2016 21 Sept 2016 21 Sept 2016 21 Sept 2016 21 Sept 2016 21 Sept 2016 21 Sept 2016 21 Sept 2016 22 Sept 2016 22 Sept 2016 24 Sept 2016 26 Sept 2016 26 Sept 2016 27 Sept 2016 27 Sept 2016

From September 2016 to November 2016, 105 video clips were produced and uploaded on the YouTube channel. Table 2 provides basic statistics of these 105 video clips with regard to the average number of views, average number of user engagements (i.e., likes, unlikes, shares, comments), average video duration in seconds, average viewing duration in seconds, and average video clip view percentage. The study targeted Hong Kong Chinese YouTube users and all video clips were created in Cantonese with Chinese topics and descriptions. Two professors from a media school in Hong Kong independently analysed these 105 video clips and assigned each to one of ten categories based on video content, video topic, and description (Table 3). These 105 video clips were analysed using YouTube statistics. The most popular video type viewed was news (average views were 838 and 7084 views of a particular video). On the contrary, YouTube users were not interested in video clips about personal tour sharing, with an average of only 22 views of eight video clips.

68

E.K.W. Lau

Fig. 1. Basic statistics provided by YouTube analytics (in Chinese)

Table 2. Basic statistics summary Video clip category

Buying blog Fengshui Ghost story Investment Love affair News Personal sharing Pet Teaching Tour

Number

Average views

9

324.78

Average Average engagements video duration (seconds) 3.33 317.33

Average view duration (seconds) 136.56

Average view percentage

16 8 8 4 18 20

137.69 101.13 194.38 194.00 838.06 123.05

3.75 3 2.13 4.75 9.33 2.05

467.19 572.25 567.75 465.50 526.44 487.75

244.81 321.00 292.50 270.75 225.06 214.15

54.63 57.63 52.13 58.50 44.83 46.20

3 11 8

25.00 112.36 22.00

1.00 2.82 0.25

27.00 605.55 405.00

17.67 705.55 88.75

67.33 44.36 37.63

50.56

All descriptive information collected about the video clips, user engagements, and views were inputted and analysed using SPSS. Basic descriptive statistics, ANOVA, and correlations were conducted. A one-way analysis variance (i.e., ANOVA) was conducted to test the differences between the YouTube users’ engagement variables (i.e., number of views, total number of likes, unlikes, shares, and comments and view

Knowledge Sharing on YouTube

69

Table 3. Video classification categories Category Buying blog Fengshui Ghost story Investment Love affair News Personal sharing Pet Teaching Tour

Description Sharing the buying experience Sharing Fengshui case studies and theories Sharing of personal ghost stories Sharing of personal investment strategies Sharing of personal views on love affairs Sharing of personal views on recent news topic Sharing of daily life events A video posted about a pet An educational video Sharing of a tour

percentage of particular video clip). It was found that the video topic did not affect the number of views (F(9, 95) = 1.360, p > 0.05), total number of likes, unlikes, shares and comments (F(9, 95) = 1.776, p > 0.05). However, it was found that the video topic affected the view percentage of particular video clips (F(9,95) = 3.071, p < 0.05; Table 4). Table 4. One-way ANOVA Construct Number of views

Number of likes, unlikes, shares and comments View percentage

Between groups Within groups Total Between groups Within groups Total Between groups Within groups Total

Sum of Squares 7779300.808 60363322.18 68142622.99

df 9 95 104

Mean Square 864366.756 635403.391

F 1.360

Significance 0.217

780.336 4636.711 5417.048

9 95 104

86.704 48.807

1.776

0.083

4308.405 14810.509 19118.914

9 95 104

478.712 155.900

3.071

0.003

A correlation analysis was conducted on all variables to explore the relationship between variables related to YouTube users’ engagement. The bivariate correlation procedure was subject to a one-tailed test of statistical significance at two difference levels: highly significant (p < 0.001) and significant (p < 0.05). The Pearson correlations are shown in Table 5. The Pearson results are reported below. For the strength of rela‐ tionships between variables, the guidelines suggested by Rowntree [9] were followed:

70

E.K.W. Lau Table 5. Pearson correlation analysis (significance)

Construct Number of views Number of likes, unlikes, shares and comments View percentage

1 1.000 0.963** (0.000) −0.095 (0.335)

2

3

1.000 −0.074 (0.455)

1.000

Note: ** Correlation is significant at 0.01 level (1-tailed)

Correlation coefficient (r) 0.0–0.2 0.2–0.4 0.4–0.7 0.7–0.9 0.9–1.0

Strength of relationships Very weak, negligible Weak, low Moderate Strong, high, marked Very strong, very high

As shown in Table 4, it was found that the number of views has a positive and relatively strong association with the number of likes, unlikes, shares and comments (r = 0.963, p < 0.001). Interestingly, there were not many significant relationships between the number of views and view percentage (r = −0.095, p > 0.001). In addition, there were no relationships between the number of likes, unlikes, shares and comments and view percentage (r = −0.074, p > 0.001).

5

Implications and Conclusion

YouTube is an open channel for everyone and its visual contents are worth a look. Producing video clips on YouTube is important nowadays as a way to engage with others and gain social support. Creating video content is the first step. Everyone can script and shoot video clips. We investigated case studies of different video topics posted on YouTube and recorded their popularity. Obviously, the study found that the number of views correlated with users’ engagements (number of likes, unlikes, shares and comments). Finding a good story is more important than video production on YouTube. YouTube ranks uploaded video clips by viewership and users’ engagements. Video producers as content providers should consider creating video content that can capture people’s attention so as to engage with them. Interestingly, the study found that the video topic did not affect the number of views and user engagement. However, it affected the percentage of video views. In addition, the length of video clips posted on YouTube also affected the percentage of video views. The general principle is that short videos work better than long videos on YouTube. The essence of popular video on YouTube is that a better storyboard makes a vlog concise but comprehensive. These results help people interested in vlogging to have a more grounded understanding of how YouTube can be effectively used; i.e., selecting the right length of video to post on YouTube and making the video relevant and meaningful to the YouTube users. A more thorough investigation of video context, especially the influence of human emotions as non-linguistic cues

Knowledge Sharing on YouTube

71

(Alloway and Alloway 2012) and other multimedia elements (such as music as suggested by Gregory and Gooding [5]) placed in the video, merits future work. Also, it is worth investigating the reasons for users’ comments and their engagements (both regular and active engagement with the social media network). Social media is a hotly debated research topic. However, there are very few empirical studies on the effectiveness of the use of the YouTube video. This paper provides evidence to support the concept. The study focuses on participatory action research in social media. It uses a collaborative inquiry approach. The investigation may help video producers to manage YouTube users’ engagements more effectively in their social media strategies. YouTube, as one of the most popular social media networks, also provides a rich research environment for anthropologists, social psychologists, marketers, politi‐ cians, communication scientists, etc.

References 1. Ahn, D., Shin, D.: Observers versus agents: Divergent associations of video versus game use with empathy and social connectedness. Inf. Technol. People 29(3), 474–495 (2016) 2. Alloway, T.P., Alloway, R.G.: The impact of engagement with social networking sites (SNSs) on cognitive skills. Comput. Hum. Behav. 28(5), 1748–1754 (2012) 3. Dehghani, M., Niaki, M.K., Ramezani, I., Sali, R.: Evaluating the influence of YouTube advertising for attraction of young customers. Comput. Hum. Behav. 59, 165–172 (2016) 4. Dynel, M.: Participation framework underlying YouTube interaction. J. Pragmat. 73, 37–52 (2014) 5. Gregory, D., Gooding, L.G.: Viewers’ perceptions of a YouTube music therapy session video. J. Music Ther. 50(3), 176–197 (2013) 6. Haridakis, P., Hanson, G.: Social interaction and co-viewing with YouTube: Blending mass communication reception and social connection. J. Broadcast. Electron. Media 53(2), 317– 335 (2009) 7. Lewinski, P.: Don’t look blank, happy, or sad: Patterns of facial expressions of speakers in banks’ YouTube videos predict video’s popularity over time. J. Neurosci. Psychol. Econ. 8(4), 241–2498 (2015) 8. Oh, S., Syn, S.Y.: Motivations for sharing information and social support in social media: A comparative analysis of Facebook, Twitter, Delicious, YouTube, and Flickr. J. Assoc. Inf. Sci. Technol. 66(10), 2045 (2015) 9. Rowntree, D.: Statistics Without Tears. Penguin Books, New York (1981) 10. Shepherd, I.D.H.: From cattle and coke to Charlie: meeting the challenge of self marketing and personal branding. J. Mark. Manage. 21, 589–606 (2005) 11. Smith, T.: The social media revolution. Int. J. Mark. Res. 51(4), 559 (2009)

Internal Knowledge Sharing Motivation in Startup Organizations Jouni A. Laitinen ✉ and Dai Senoo (

)

Department of Industrial Engineering and Management, Tokyo Institute of Technology, Tokyo, Japan {Laitinen.j.aa,Senoo.d.aa}@m.titech.ac.jp

Abstract. Knowledge sharing is an integral part for increasing the innovation capability of organizations. For small organizations, this is of particular impor‐ tance, as they require a high innovation capability in order to support organiza‐ tional growth. To support internal knowledge sharing, incentives encouraging sharing can be introduced. However, there is no clear understanding of how incentives should be used to support knowledge sharing in startups. Hence this article aims to answer the question: how can incentives be used to support knowl‐ edge sharing in startup organizations? To answer the research question, 10 semi-structured interviews with were carried out in Hong Kong and Japan. Nine of the interviews were carried out with founders and one was carried out with a lawyer who specializes in incentive crea‐ tion. The interviews were then analyzed using cross-case analysis methodology. Results of the analysis show that the need for incentives depends on how clear the founder’s vision for the company is, how the founder views the employees and how motivated by the topic of work the employees are. The results give insight into whether incentives are needed to encourage knowledge sharing and when they should be used. The results also give future research directions into how incentive usage evolves over time as the organization starts to grow. Keywords: Knowledge sharing · Incentives · Collaboration · Startups

1

Introduction

All organizations need to pay attention to how they compete for their position in the relevant ecosystem. For this to happen, newer companies should find a unique niche or an innovation to fill the needs of the customers. Increasing the communication and collaboration between individuals, who don’t normally interact, is important for increasing the creation of new innovations [1]. One way to support innovation within in organizations is to support knowledge sharing [2]. For knowledge sharing to happen, many companies choose to implement incentives to encourage employees to take ownership of their work. These incentives include stock options and other incentive plans. Stock options are particularly used when there are optimistic expectations about the company performance [3].

© Springer International Publishing AG 2017 L. Uden et al. (Eds.): KMO 2017, CCIS 731, pp. 72–83, 2017. DOI: 10.1007/978-3-319-62698-7_7

Internal Knowledge Sharing Motivation in Startup Organizations

73

While there have been studies on knowledge sharing in small and medium enterprises [4], this research focuses on startups. We define a startup, as a company early in the stages of developing a product or a service for a market that is unserved by others or served by inferior products with the goal of achieving rapid growth. That is, this research focuses on companies on the birth and growth stages of the Business Life Cycle as defined in [5]. This difference in foundations of the organization lead to differences in how the organizations operate and grow [6]. Hence, it is of high importance to under‐ stand the foundations clearly (Fig. 1).

Birth

Growth

Maturity

Revival

Decline

Fig. 1. Business life cycle (adapted from [5])

The effects of incentives on employee motivation to collaborate in early stage startups are yet to be fully understood. Therefore, this article aims to answer the question of how can incentives be used to support knowledge sharing in startup organizations. To answer this question, the relationship between early stage startup founders and their relationship to incentives was studied through 10 semi-structure interviews carried out in Japan and Hong Kong. The results indicate that the clear founder vision improves increases employee motivation to share knowledge and that employees are mostly intrinsically motivated to engage in knowledge sharing. These results will influence how incentives should be used once the organization starts to grow bigger and become more mature. The rest of the article is structured as follows: first incentives and knowledge sharing are reviewed and then the interview data is presented. Then, the interviews are analyzed and discussed in detail. Finally, limitations of the study and future research directions are discussed.

2

Literature Review

2.1 Incentives in Organizational Context There are two major perspectives towards how incentives influence behavior: psycho‐ logical and economical. While the two can be seen as opposite due to their fundamentally different views on what motivates individuals, they should be understood as comple‐ mentary. As such, understanding when to use which perspective becomes important. From the economic perspective, individuals consider the utility that can be derived from carrying out a given task [7]. For organizations, the most frequently used theory on incentive usage is based on Agency Theory [8], which considers situations, where tasks are delegated by one individual to another. When the task does not align with the self-interest of the individual carrying it out, the so-called agent, the delegator, also known as principal, can introduce incentives to carry out the task. The frequency and length of interaction can be considered when calculating the potential utility derived

74

J.A. Laitinen and D. Senoo

from the action through the inclusion of discounted utility [9]. This theory can also be used to extend the focus of analysis to multiple subjects. Psychological theories are based on laboratory experiments about motivation and how incentives influence motivation to carry out given tasks [10]. Individual’s motiva‐ tion to carry out a given task can be divided into two different classes: intrinsic moti‐ vation and extrinsic motivation. Intrinsically motivated tasks are carried out without any external incentives due to the individual’s want to carry out the task [10]. However, if there are external motivators, such as bonuses or rewards for completing a task, then it is said that the individual is extrinsically motivated [10]. A considerable body of research exists on the types of incentives and what effect they have on individuals [11]. When the psychological foundations are expanded to cover dyadic relationships in an organizational context, Fiske’s work on Relational Models [12] gives insight into what rules govern those interactions. The dyadic relationships carry given set of rules and guidelines, which when broken will create friction in the interaction and cause a change in the relation type or even cause a backlash from the other person. The different types of relationships range from a family-like relation, where individuals carry out tasks for close individuals without considering the cost of the task, to a market-type relationship, where cost-benefit analysis of each action is carried out in regards to the interaction individuals. These relational models give an insight into the governing rules of the relationship. In the context of knowledge sharing, both perspectives have been used to carry out research. For example, researchers such as Nan [13] and Cabrera and Cabrera [14] have used pure economic perspectives to theorize knowledge sharing and individual behavior. For the psychological approach, Fiske’s theory was used by Lin, Wu and Lu [15] to explore factors affecting knowledge sharing behavior. When comparing the two perspec‐ tives in the context of knowledge sharing in startups organizations, of particularly note‐ worthy is their predictions on how individuals would interact and what is their motiva‐ tion to engage in such a behavior. From a pure economic perspective, the motivation would be to maximize utility. This will lead to increase knowledge sharing if the indi‐ viduals engaged in activities can derive a bigger utility from collaborating than they can from hoarding knowledge. It can also lead to the lack of certain types of individuals in the startup context. From a psychological perspective, the motivation to engage in knowledge sharing would be more dependent on the individual, the person’s relation‐ ships with others and how they perceive the task. 2.2 Knowledge Sharing Knowledge sharing is an integral part of knowledge management [2]. Knowledge sharing is the action to share and make knowledge available throughout the organization with the goal of increasing knowledge utilization [14]. Knowledge sharing takes place frequently when there is collaboration between individuals. For the scope of this research, we are only concerned with knowledge sharing that happens within the organ‐ ization. It has been shown that increased knowledge sharing is linked with increased innovation capability [16]. As engagement in knowledge sharing cannot be forced [17], it is material to understand why individuals engage in the sharing activities and how to encourage more people start sharing.

Internal Knowledge Sharing Motivation in Startup Organizations

75

Motivation to engage in knowledge sharing is either based on a want to engage in the activity, i.e. intrinsic motivation towards the task, or there is some external factor influencing motivation, i.e. extrinsic motivation [18]. Influencing the individual’s intrinsic motivation to engage in a given task is challenging [19] but not impossible [20]. However, changes in intrinsic motivation take a fairly long time to happen [21]. Thus, designing extrinsic incentives to support engagement in knowledge sharing should be emphasized. Existing research on incentive design to encourage knowledge sharing has received some attention (e.g. [22, 23]). However, the results from empirical studies have failed to derive a consensus on the effectiveness of incentives to encourage knowledge sharing, as positive [24], negative [25] and neutral [2] results have been derived. As such the discussion on how to design incentives to support knowledge sharing is still on-going. From the literature review we can see that there are two different perspectives, which start from very different foundations. Both have been used to understand how incentives should be used in KM setting. However, the differences in the motivation to engage in knowledge sharing can have significant effect on the organization in the long-term. As such, more research into why the organization members engage in knowledge sharing is needed.

3

Interviews

To better understand what motivates individuals in early stage startups to engage in knowledge sharing and what types of incentives should be used, a case oriented research approach was selected [26]. For this, interviews were carried out in Japan and Hong Kong. The two countries were selected as both countries have a growing startup ecosystem. The companies selected to take part in the research process were selected based on recommendations from local ecosystem leaders. The semi-structured inter‐ views took place between January and November 2016. In total, there were 10 inter‐ views, 5 in Japan and 5 in Hong Kong. 9 of the interviews were carried out with the early stage startups. The language of the interviews was English in both countries. In Japan, if the interviewee wasn’t sure how to correctly make a statement in English, the interviewer asked for clarification in Japanese, which was then later translated into English to confirm the meaning of the original statement. The last interview was carried out with a lawyer, who specializes in creating incen‐ tive systems for startups. This was done in order to understand how do outside actors, i.e. individuals, who will bring more maturity to the organizational structures in later stages of the life cycle, perceive how incentives influence the behavior of individuals in the organization. Summary of the interviews can be seen in the table below (Table 1). The average lengths range between 20 min and 45 min. The interviews were recorded and the interviewer took notes at the same time. After the interview, the recordings were transcribed and combined with the field notes.

76

J.A. Laitinen and D. Senoo Table 1. Overview of the interviewees

Identifier JP1 JP2 JP3 JP4 JP5 HK1 HK2 HK3 HK4 HK5

Country JP JP JP JP JP HK HK HK HK HK

Type Founder Founder Founder Founder Lawyer Founder Founder Founder Founder Founder

Background Finance Sales Sales Finance Lawyer Finance Consulting Manufacturing Finance Serial entrepreneur

Number of founders 1 2 3 3 NA 3 2 2 3 3

To better understand the common patters between all of the different interviews, a cross-case analysis methodology was chosen for the analysis phase. The combined records from the interviews were then analyzed based [27, 28], who described the anal‐ ysis process for cross-case analysis. This was done by creating pairs of 2 × 2 matrixes with selected dimension to highlight differences and similarities across the cases. Based on the literature review, dimension pairs are (1) founder vision clarity x employee moti‐ vation to share knowledge and (2) founder perception for the need for incentives x collaboration motivation of employee. For the creation of the matrixes, all transcribed data was analyzed and the relevant information was entered into the relevant dimension of the 2 × 2 matrix. After this the matrix was analyzed from common patterns and a descriptive name was given to each cell.

4

Results of the Interviews

4.1 Founder Vision Clarity X Employee Motivation to Share Knowledge The first analysis of particular interest is the cross-analysis of founder vision clarity and employee motivation to share knowledge. This was brought up with the founders, i.e. 9 out the 10 interviews covered this topic (excluding the lawyer). In the results, three patterns can be seen: high clarity x low motivation, high clarity x high motivation, low clarity x high motivation. The patterns are shown in Table 2 below. First, founder with a clear vision & employee with high motivation to share knowl‐ edge. This pattern aligns both company and employee interests leading to higher moti‐ vation to carry out the task, which was shown in statement given by HK2, who stated in regards to employee motivation that they are motivated by: “… [B]eing able to work relaxed in a bigger scale with more impact…” The interviewee later added: “We really appreciate each other personally. So, that’s really cheesy but I guess it’s really the driving force [in collaboration].” As knowledge sharing is one form of collaboration, this is indicative of a high motivation to share knowledge with other. Similar statements were also made by JP2, who emphasized personal vision also for employees by stating that “[S]o, why are you working [here]? What is your purpose in life?” The founder supported

Internal Knowledge Sharing Motivation in Startup Organizations

77

this goal of having all of his employees find their “personal vision” by bringing in external advisors that would not only give advice on the company but would also give mentoring to the employees. This indicates the founder’s want to increase motivation to collaborate and share knowledge between organization members. The second pattern that was found across the interviews is when the founder doesn’t have a clear vision for the company but the employees are highly motivated to share knowledge. This pattern was the clearest in the case of founder from Hong Kong, HK3. The interviewee first described changes in their product development strategy by stating: “…[W]e made a lot of mistakes… [W]e didn’t validate the market…This time I know who my market is and I’m going to build a product for that market.” These statements are indicative of a founder, who has a vision but no clear understanding of how to achieve that vision or if the vision is realistic. After describing their product development plan for the next stage, the founder described their employees’ motivation as follows: “If you are a mechanical engineer then there is one job you can do and that’s HVAC [heating, ventilation and air conditioning]. Nobody wants to do that. So, you tell them that ‘hey there is a product development job,’ I think people would do it for free.” This indicates that the employees, who join the company have high motivation to collaborate and share knowledge with the other members of the company. The founder’s lack of a clear vision combined with the high employee motivation results in an organization that is eager to engage in new actions but lacks direction. The third patter found within the interviews is high founder vision clarity but the employees’ motivation is low. This pattern was observed in the statements of a Japanese founder, JP1, who stated that: “we’re not like a typical startup, where you get together every day and [..] be like ‘ooooh, here’s our dream.’” The founder later also indicated that he controls what the employees are doing and why they are doing it. This was indicated when the founder stated that: “each day they come in they set their own goals and when they finish they fill out [in daily a task sheet] what they have accomplished that day and what’s remaining.” These are indicative of a founder, who has a clear vision of where he wants the company to go. From the employee side, the employees want to achieve the founder’s vision but it needs to be done the way the founder wants, as was shown when the founder talked about troubles in collaboration: “…[S]ometimes things are difficult. For example, I would want the person to do [task] but he wouldn’t want to do it that way, you know [so I had to fire the person in question].” This indicates that the founder views employees more as a resource than as an asset.

Table 2. Founder vision clarity x employee motivation to share knowledge Founder vision clarity High Employees are a way to achieve vision Low Low

Want to achieve the vision together Want to work on an interesting topic High

Employee motivation to share knowledge

78

J.A. Laitinen and D. Senoo

4.2 Founder Perception for the Need for Incentives X Collaboration Motivation of the Employees Second cross-case analysis matrix deals with how founders understand the need for incentives and what motivates employees to collaborate. All of the 10 interviewees were asked about this question. This 2 × 2 matrix is shown below in Table 3. Table 3. Need for incentives X collaboration motivation Founder’s perception for the need for incentives No

Yes

Employees do the job Employees motivated they are hired to do by company vision, want to achieve something Employee’s low compensation supplemented by giving out stock Extrinsic Intrinsic Employee collaboration motivation

Out of the 10 interviews carried out, 9 interviewees gave statements that align with intrinsic motivation being the main motivator for sharing knowledge. For example, A Hong-Kong-based founder, HK3, stated that his motivation to start the company was to: “…[B]uild things for myself” and continued on to state on the reasons why employees in his company collaborate because: “[I]f we [founder and employees] can make it happen then both are better off and everybody is happier.” Similar statements can also be found in the statements of one founder, JP2, interviewed in Japan, who stated that: “collaboration leads to the success of the company.” Shortly after he continued by stating that “it makes no sense […] not to share, generally at least.” This was also supported by the interviewed lawyer, JP5, who stated that “usually people are motivated anyway, even without stock options…” The outsider perspective on the need for incentives thus aligns with the majority of the founder interviews. According to the founders, the companies that gave out incentives mostly used stock to compensate for low salary. This was also confirmed by the lawyer. In contrast to these, one founder interviewed in Japan, JP1, stated on the topic for incentive usage that: “I haven’t given out any stock options yet but to anyone who makes the big contributions I would consider.” As the company in question is one of the older ones in the group, this is of particular interest. This statement clearly deviates from the rest of the interviews, as it indicates that the founder views the other members of the organization as extrinsically motivated and as such there is a need to have incentives, which can be achieved through sharing and collaborating.

Internal Knowledge Sharing Motivation in Startup Organizations

5

79

Analysis and Discussion

The interviews described in the results section describe two different aspects of organ‐ izational behavior at an early stage startup: first, how does the founder vision influence employee motivation and second, how are incentives and collaboration motivation linked. 5.1 Founder’s Vision Influence on Employee Motivation to Carry Out Tasks Based on the interviews, for founder vision’s influence on employee motivation, there are three different patterns: high clarity x low motivation, high clarity x high motivation, low clarity x high motivation. If the employee motivation is high, then the given task would get carried out even without any need for the founder to motivate the employee. However, some of the founder stated that giving context for the tasks improves employee motivation to carry them out. This was best shown in a statement given by a Hong Kong founder, HK1: “[working long on a prototype] will demotivate someone easily because you don’t see the end product anymore. So, we always try to tell them the big picture why we are doing.” Similar results have been previously derived [29]. Along the similar lines another Hong Kong founder, HK3, described the importance of founder vision for the company as: “You can’t have two people with strong ideas, that is the biggest thing … You can either have zero or one… one [is the best] because I think at zero you end with everyone having a say and you end up with something that no one really wants.” This is in line with [30], who also emphasized the role of market knowledge in top management. Hence, not only does founder vision increase motivation but it also increases the likelihood of success. This can be seen also in the case where the founder vision is low. If the employees like what they are doing, they won’t need the extra motivation that comes from the founder. However, a lack of vision will most likely lead to an increased mortality rate of the companies [31]. The final pattern, that was found in the first perspective, is the case where the founder has a strong vision for the company but the employee has low motivation to carry out tasks. In this case, the founder’s vision no longer provides any extra boost for the employee. As such, this case resembles the traditional principal-agent scenario described in [8], where incentives are needed from very early on. Therefore, the main way for the founder to motivate the employees is to create an incentive system, which aligns the employee’s goals with the company through providing large enough incentives. 5.2 Founder’s Perception for the Need for Incentives and Employee Motivation to Collaborate Switching perspective to founders’ understanding for the need for incentives. We can see the influences of founder’s vision on how they perceive the need for incentives. Based on the interviews, most founders see that there is no need to incentivize knowledge sharing within the organization at this stage, which would imply that the employees are intrinsically motivated to carry out the given tasks. Fiske’s [12] theory of relational models predicts that in small groups, in communal sharing, individuals would be

80

J.A. Laitinen and D. Senoo

intrinsically motivated to carry out tasks. This is shown in comments such as “we know we are more valuable to each other as a(sic.) unit” from JP3. Similar results have been previously derived in [15] and in [32] and it highlights the relational aspects of the organization. In contrast, if the founder sees the organization as being motivated more by extrinsic motivation, this will be reflected in how incentives are used and how the organizational culture works. This was indicated in the statements made by a Japanese founder, JP1, who declared that the organizational culture is based on a flat culture that supports communication. However, this was later contradicted by the same founder by stating that all employees need to fill in daily task-sheets in the morning and at the end of the day. In combination with the strategy of only giving out stock once an employee had made a “big contributions” displays that in the founder view the employees and founder are not on the same level. This is typical behavior that is displayed in a market-type relation [12] and it is in line with previous results derived [15]. The difference between these two patterns highlights the need to understand the organizational context and the need to understand the contextual factors of incentives. Similar findings have been previously described in [21, 33]. Based on the interviews carried out, most founders see that their employees are intrinsically motivated and as such incentives should focus more on the relational aspects. For example, job design can be used to positively increase individual’s motivation to engage in knowledge sharing [20]. One way that many founders incentivize employees is through giving stock to early stage members. Based on the interviews, giving out stock was used to compensate for low salary but it has also another effect. As valuation of stock at early stage companies is considerably difficult [34], the act of giving out stock signals the founder’s appreci‐ ation and want to share with the early employees. These types of socially “infused” incentives have been shown to increase the effort levels of the employees [35]. Overall it can be stated that the interviews show that early stage startups closely resemble what Fiske´s Relational Models Theory [12] predicts for small, closely-knit communities. That is, they function more based on a motivated to work towards a common goal, i.e. founder’s vision. This is supported by the small number of employees at the organization, which leads to more frequent interaction between the employees and the founder. As such, the contextual and relational factors of incentives play a very important role. Therefore, should incentives be used, they should emphasize the social aspects of the relationships in the organization. It should be noted that in addition to the three perspectives discovered for both crosscase analyses, there is one additional common pattern across all cases: low vision clarity x low motivation and high need for incentives x extrinsically motivated employee. Both of these patterns were missing from the interview records. This is most likely due to the fact that in the case of low vision clarity and low motivation employees would not join a startup or would quit very soon due to unfulfilling work, and employees, who only work for the company for the incentives, can derive a better pay in other, more estab‐ lished companies. This can be seen in statements from founders from both Hong Kong (HK4) and Japan (JP2).

Internal Knowledge Sharing Motivation in Startup Organizations

6

81

Conclusions, Limitations and Future Research

The research described in this article started with the aim of better understanding of how to motivate knowledge sharing in early startup organizations. This is of particular importance these days, as the number of startup companies is increasing. Previous research on these types of organizations shows that there is not just one type of organ‐ ization but multiple patterns that the organization can follow as it grows [6]. Addition‐ ally, previous research on incentive usage [36], their influence to increase innovation [37] and how employees of the organization behave when they are given stock options [3] have all been carried out in large organizations. Hence, there is a lack of under‐ standing how do incentives relate to smaller organizations. To get fill this gap in knowl‐ edge, interviews were carried out in Hong Kong and Japan with early stage startups. The interviews were themed around two main topics: First, how does the founder vision influence employee motivation and second, how founders see the connection between incentives and collaboration motivation of the employees. Based on the results of the analysis, it can be stated that incentives to share knowledge within the organization depend on how the founder vision clarity, in what role does the founder see the employees and employee motivation. If the founder has a clear vision for the company and views the employees as equal or the employees have intrinsic interest in the work, there doesn’t seem to be a need for incentives to be used to encourage internal knowledge sharing, as sharing will take place naturally. This highlights that startups more resemble families than companies due to their close-knit relations between the employees and founders. However, if the founder views employees as resources to achieve the company goal, there will be a need for incentives. For the creation of incen‐ tives to be successful there are guidelines that need to be followed [33]. While difficult [37], it can be done [36]. These results have multiple implications: first, it increases the importance for under‐ standing how incentives link to the culture of sharing. The right balance on incentives can help to encourage sharing in organizations, where sharing is limited or non-existent. However, the results also imply that organizations that collaborate without the use of incentives should take particular care when they use any types of incentives, be they yearly bonuses or any other incentives. A wrong configuration of incentives can cause damage to the sharing culture. Second, when the results are analyzed through the theories proposed by Fiske [12] they imply that as the companies mature the internal culture will develop towards a more generic culture and away from the culture that makes startups more efficient. What is of interest is that the results imply that this development towards a more generic culture can be postponed or cancelled if the motivations of the members are carefully aligned with fundamental values of the organization. The research presented in this articles also has limitations. While one the goals of carrying out multiple interviews in multiple countries was to increase the level of confi‐ dence in the results, there is a need to develop a statistical instrument to test the results derived here on a larger scale. This will remove uncertainty caused by any unconscious biases that the researchers might have. The development of the statistical instrument can also shed light into how the organizational culture develops over time.

82

J.A. Laitinen and D. Senoo

What these results imply about the life-cycle of the startup company is that the need for incentives to encourage internal knowledge sharing goes through a radical change as the organization grows. This also means that at the earlier stages of the life cycle the incentives should be aimed at supporting the relationships between the members of the organization. Due to this the research carried out here suggest suggests a couple of new research directions: first, what happens once the organizations starts to grow and receives funding from external resources such as venture capitalists? When the monetary aspects of stock and incentives become more salient, how does this change influence organiza‐ tional culture and collaboration. Second, how does the organizational culture develop over time when more employees join the company and how can the organization create processes, which can help maintain the culture of collaboration? This is of particular interest due to the close-knit relations that smaller startups have. Both of these directions can give insight into how to create more influential knowledge management programs in organizations regardless of its size.

References 1. Granovetter, M.S.: The strength of weak ties. Am. J. Sociol. 78(6), 1360–1380 (1973) 2. Lin, C.P.: To share or not to share: modeling tacit knowledge sharing, its mediators and antecedents. J. Bus. Ethics 70(4), 411–428 (2007) 3. Bergman, N.K., Jenter, D.: Employee sentiment and stock option compensation. J. Financ. Econ. 84(3), 667–712 (2007) 4. McAdam, R., Reid, R.: SME and large organisation perceptions of knowledge management: comparisons and contrasts. J. Knowl. Manage. 5(3), 41–231 (2001) 5. Miller, D., Friesen, P.H.: A longitudinal study of the corporate life cycle. Manage. Sci. 30(10), 1161–1183 (1984) 6. Delmar, F., Davidsson, P., Gartner, W.B.: Arriving at the high-growth firm. J. Bus. Ventur. 18(2), 189–216 (2003) 7. Fehr, E., Falk, A.: Psychological foundations of incentives. Eur. Econ. Rev. 46(4), 687–724 (2002) 8. Eisenhardt, K.M.: Agency theory: an assessment and review. Acad. Manag. Rev. 14(1), 57– 74 (1989) 9. Frederick, S., Loewenstein, G., O’donoghue, T.: Time discounting and time preference: a critical review. J. Econ. Lit. 40(2), 351–401 (2002) 10. Ryan, R.M., Deci, E.L.: Intrinsic and extrinsic motivations: classic definitions and new directions. Contemp. Educ. Psychol. 25(1), 54–67 (2000) 11. Cameron, J., Banko, K.M., Pierce, W.D.: Pervasive negative effects of rewards on intrinsic motivation: the myth continues. Behav. Analyst. 24(1), 1 (2001) 12. Fiske, A.P.: The four elementary forms of sociality: framework for a unified theory of social relations. Psychol. Rev. 99(4), 689 (1992) 13. Nan, N.: A principal-agent model for incentive design in knowledge sharing. J. Knowl. Manage. 12(3), 13–101 (2008) 14. Cabrera, A., Cabrera, E.F.: Knowledge-sharing dilemmas. Organ. stud. 23(5), 687–710 (2002) 15. Lin, T.C., Wu, S., Lu, C.T.: Exploring the affect factors of knowledge sharing behavior: the relations model theory perspective. Expert Syst. Appl. 39(1), 751–764 (2012) 16. Lin, H.F.: Knowledge sharing and firm innovation capability: an empirical study. Int. J. Manpower 28(3/4), 315–332 (2007)

Internal Knowledge Sharing Motivation in Startup Organizations

83

17. Gibbert, M., Krause, H.: Practice exchange in a best practice marketplace. In: Knowledge management case book. Siemens Best Practices, pp. 89–105 (2002) 18. Lin, H.F.: Effects of extrinsic and intrinsic motivation on employee knowledge sharing intentions. J. Inf. Sci. 33(2), 135–149 (2007) 19. Osterloh, M., Frey, B.S.: Motivation, knowledge transfer, and organizational forms. Organ. Sci. 11(5), 538–550 (2000) 20. Foss, N.J., Minbaeva, D.B., Pedersen, T., Reinholt, M.: Encouraging knowledge sharing among employees: how job design matters. Hum. Resour. Manage. 48(6), 871–893 (2009) 21. Bowles, S., Polania-Reyes, S.: Economic incentives and social preferences: substitutes or complements? J. Econ. Lit. 50(2), 368–425 (2012) 22. Bartol, K.M., Srivastava, A.: Encouraging knowledge sharing: the role of organizational reward systems. J. Leadersh. Organ. Stud. 9(1), 64–76 (2002) 23. Lee, D.J., Ahn, J.H.: Reward systems for intra-organizational knowledge sharing. Eur. J. Oper. Res. 180(2), 938–956 (2007) 24. Kubo, I., Saka, A.: An inquiry into the motivations of knowledge workers in the Japanese financial industry. J. Knowl. Manage. 6(3), 262–271 (2002) 25. Bock, G.W., Zmud, R.W., Kim, Y.G., Lee, J.N.: Behavioral intention formation in knowledge sharing: examining the roles of extrinsic motivators, social-psychological forces, and organizational climate. MIS Q. 1, 111–187 (2005) 26. Ragin, C.C.: The Comparative Method: Moving Beyond Qualitative and Quantitative Strategies. University of California Press, Oakland (2014) 27. Voss, C., Tsikriktsis, N., Frohlich, M.: Case research in operations management. Int. J. Oper. Prod. Manage. 22(2), 195–219 (2002) 28. Eisenhardt, K.M.: Building theories from case study research. Acad. Manag. Rev. 14(4), 532– 550 (1989) 29. Abbott, G.N., White, F.A., Charles, M.A.: Linking values and organizational commitment: a correlational and experimental investigation in two organizations. J. Occup. Organ. Psychol. 78(4), 531–551 (2005) 30. Li, T., Calantone, R.J.: The impact of market knowledge competence on new product advantage: conceptualization and empirical examination. J. Mark. 1, 13–29 (1998) 31. Castrogiovanni, G.J.: Pre-startup planning and the survival of new small businesses: theoretical linkages. J. Manage. 22(6), 801–822 (1996) 32. Boer, N.I., Berends, H., van Baalen, P.: Relational models for knowledge sharing behavior. Eur. Manage. J. 29(2), 85–97 (2011) 33. Laitinen, J.A., Senoo, D.: Creating incentives for knowledge sharing: combining context and signaling. In: Proceedings of the 11th International Knowledge Management in Organizations Conference on The changing face of Knowledge Management Impacting Society, p. 35. ACM, 25 July 2016 34. Miloud, T., Aspelund, A., Cabrol, M.: Startup valuation by venture capitalists: an empirical study. Venture Capital 14(2–3), 151–174 (2012) 35. Heyman, J., Ariely, D.: Effort for payment a tale of two markets. Psychol. Sci. 15(11), 787– 793 (2004) 36. Gibbs, M., Neckermann, S., Siemroth, C.: A field experiment in motivating employee ideas. Rev. Econ. Stat. 1 April 2014 37. Ederer, F., Manso, G.: Is pay for performance detrimental to innovation? Manage. Sci. 59(7), 1496–1513 (2013)

Knowledge Sharing in Constructability Management Marja Naaranoja(&) and Joni Vares Department of Construction Management, Vaasa University of Applied Sciences, Wolffintie 30, 65200 Vaasa, Finland [email protected]

Abstract. Knowledge sharing in construction is one of the key challenges for successful implementation of a construction project. Thus, it would be a good idea to improve the knowledge sharing practices. This paper focuses in constructability management in large infrastructure projects. Constructability pursues to find ways to construct effectively. The paper analyses how the knowledge is shared when using the constructability management principles. We proposes how to improve the theoretical model on the incentives on knowledge sharing. We also concluded this paper with some future directions and suggestions about improving constructability by using knowledge sharing. Keywords: Social networks management System




Facebook




Online learning




Learning

1 Introduction A building design and production process is also a knowledge management process. The actors of the process need to understand what the building to be developed is about, and how it should be designed and constructed to respond adequately to the needs of the parties. Knowledge management in the context of a building process contains understanding of the design, usability and constructability of the building; expectations of the client’s wishes, expectations and needs, user needs; and how to deal with the design and its visual impact to the build environment. This paper focuses on the role of constructability in the building process. Constructability means that the project optimally links the goals and the use of resources of the construction process and thus the decisions, which have a significant impact on the ease of construction, are made at the early stages of the project lifecycle (McGeorge et al. 2012). Constructability principles (Russel 1995) are practical principles that integrate the design and construction processes, in order to provide an effective and efficient solution to the needs of the client. Construction projects are often large and complex, thereby the number of professionals involved in projects lengthen the project life cycles and generate complex interfaces (compare Aets et al. 2016, Chou et al. 2012, Gasik 2011). This in turn influences the types and quantities of project-related information that are generated, making them more fragmented and more complex. Accordingly, development projects require substantial amounts of specific knowledge. Project management capabilities are © Springer International Publishing AG 2017 L. Uden et al. (Eds.): KMO 2017, CCIS 731, pp. 84–94, 2017. DOI: 10.1007/978-3-319-62698-7_8

Knowledge Sharing in Constructability Management

85

based on the resources in the project and the capabilities of the project network (Brady and Davies 2004; Davies and Brady 2016). According to Naaranoja and Uden (2007) the most common failure reasons were lack of decision making process, lack of time for planning, difficulty in updating construction regulations, users do not known what they need, lack of trust, lack of risk assessment, resistance to use of IT, and lack of change management. This paper focuses on capabilities related to risk assessment, decision making process and change management by using the concept constructability. Constructability concepts and principles are taken account during the different phase of the project: conceptual planning, design, procurement and construction, and the improvement of the construction processes generally, and making the process easier to manage, faster and more cost effective. Alawi et al. (2015) found that the degree of implementation of constructability concepts was less than 70% still though the principles have been developed already in 1990. The best possible constructability implementation starts at the very beginning of the project and lasts through the construction and installation, to the maintenance and to the daily operations. While the benefits of the constructability are recognized, there are often barriers to implement constructability optimization in the projects. (Gambatese et al. 2007) The design and construction processes in the life cycle of a construction project are many times separated in traditional construction business. The separation of design and construction has caused challenges as the designers are working many times on only on their own field, which might sometimes require costly and complicated sequences and methods to construct at site. It might also happen that the contractor does only what he is told to do, even in a manner that may be inefficient. The knowledge sharing should not only be understood as a sharing process between individuals like specific designer of the team but as a collective process where all involved actors cooperatively participate. The goal of this paper is to analyse the role of knowledge in using constructability principles in large infrastructure projects in an MNC company. The paper is based on 21 interviews. The analysis covered also many other issues but this paper focuses on the role of knowledge that either enables or is a barrier of the using of constructability principles. This article is divided into five sections. Section 1 provides an introduction to the background and motivation of the paper. Section 2 is the review of related literatures. In Sect. 3, the research method and interviews are introduced. Section 4 introduces the results of the analysis. In Sect. 5, we concluded the paper and provide future directions and suggestions.

2 Constructability and Knowledge Sharing This section introduces literature review related to the topic, namely: the knowledge sharing incentives in general; knowledge sharing in projects; constructability principles.

86

M. Naaranoja and J. Vares

Knowledge sharing incentives The incentives related to knowledge sharing can be divided to monetary and non-monetary incentives (e.g. Hau et al. 2013). Additionally, it has been shown that sometimes non-monetary incentives have a bigger effect than monetary incentives (Kube et al. 2012). Even the strength of explicit economic incentives varies with context (Fehr and Falk 2002). Hence, by allowing non-explicit incentives also to exist, the analysis of the collaboration motivation can become realistic Assumption of the

Fig. 1. The influence of incentives on knowledge sharing (Laitinen and Sanoo 2016)

employees being rational utility maximizing agents. (Laitinen and Sanoo 2016) (Fig. 1). According to Laitinen and Sanoo (2016) there is no consensus on if incentives actually have a positive or a negative effect on sharing levels. The incentives can be divided into a management monitored system level incentive, monetary incentive, and self determined sharing. Knowledge management in projects Knowledge is perceived by most project organisations as a vital organisational resource and source of competitiveness. It is being acknowledged that KM can bring about the much needed innovation and improved business performance in the project based industry (Carillo et al. 2000). Knowledge is defined as a dynamic human process of justifying personal belief towards the truth (Nonaka et al. 2000). It can also be defined as ‘know-why’, ‘know-how’ and ‘know-who’, or an intangible economic resource from which future resources will be derived (Clarke 2001). Knowledge is built from data, which is first processed into information (i.e. relevant associations and patterns). Information becomes knowledge when it enters the system and when it is validated (collectively or individually) as a valid, relevant and useful piece of knowledge to implement in the system (Blumentritt and Johnston 1999). Project-creators and project team-members work on a project for only a period and then move on. Individually they learn, but that knowledge/those lessons remain with them, and are not necessarily absorbed by the involved permanent organisation(s). There are many ways how this learning is pursued to be used e.g. in project

Knowledge Sharing in Constructability Management

87

Table 1. Drivers and barriers to KM in public private partnership projects (Aerts et al. 2016) Drivers

Barriers

Increased project complexity ⚬ New types and amounts of construction-related information ⚬ Fragmentation of knowledge retention bins ⚬ New forms of infrastructure procurement Construction industry culture ⚬ Slow absorption of new knowledge Market logic ⚬ Confidentiality of knowledge ⚬ Uniqueness of knowledge ⚬ Reliability of knowledge

Fostering innovation and continuous improvement and alliance formation ⚬ Long term commitments ⚬ Reduction of rework ⚬ Repeated interaction ⚬ Limited number of potential bidders Employee turnover Lack of processes and tools for knowledge transfer ⚬ Lack of organisation level commitment to knowledge management ⚬ Lack of individual level motivation for knowledge sharing

management office instructions, quality management systems and lessons learned databases but most of the learning is waste however. (Aerts et al. 2016) (Table 1). Barthorpe et al. (2000) indicate that: “Levels of innovation in the construction industry compared to other industries have been at best modest. The industry portrays a conservative and at times “laggardly” approach to new ideas, mainly due to its fragmented nature and lack of ability to invest time and money into innovation, research and development” (Barthorpe et al. 2000). Constructability Direct measuring of the constructability is impossible since it is always project specific and subjective. Every measurer defines the constructability in a different way, through the experiences and habits. If the constructability as a definition is simplified, it can mean that the goals of the project are accomplished in a way, which is effective and fluent. Construction experience and knowledge are keys to the good constructability. Knowledge and understanding of the construction process, the needed information, for example regarding the materials, labour and necessary equipment to build, and the limitations of and constraints on construction work are essential part of the success. (Gambatese et al. 2007) Addressing of the constructability to the projects can be undertaken informally or through a formalized process. The nature and format of the constructability process is often dependent of the type of project. Addressing the constructability early in the project enables the opportunities to impact to cost and quality of the project. Constructability programs might come at a cost to the project, but the cost, however, is often outweighed by the possible benefits, such as reduced construction costs, less-rework, shorter construction schedules, improved construction site safety and other positive project outcomes. (Gambatese et al. 2007)

88

M. Naaranoja and J. Vares

A workable concept of constructability needs to be recognized that there are several factors in the project environment, which have an impact on the design and construction processes, and the link between design and construction and maintenance of the power plant or building. (McGeorge et al. 2012) The constructability management during different phases of the project contains following viewpoints: Project phase: conceptual planning 1. Constructability program are made an integral part of the project execution plans 2. Early project planning actively involves construction knowledge and experience in order to create an effective project team 3. Early construction involvement is considered by using alternative contracting strategy 4. The project master schedule is construction sensitive 5. Basic design approaches focuses to use major construction methods Project phase: design and procurement 6. 7. 8. 9. 10. 11.

Site layouts promote efficient construction, maintenance and operation Procurement and design schedules are construction driven Designs are configured to enable efficient construction Design Elements are standardized Construction efficiency is considered in detailed design development Module/pre assembly designs are prepared to facilitate fabrications, installation and transportation 12. Designs promote construction accessibility of personnel, equipment and material 13. Design facilitate construction under adverse weather conditions Project phase: field operations 14. Constructability is enhanced when innovative construction methods are utilized. (CII 2007) Qualitative benefits are either a result from strategic or key execution decisions or based on functional analysis. Both above mentioned categories lead typically to the reduction in engineering, construction cost and schedule duration. Strategic decision are having the largest impact on design and construction costs and on the project schedule. Reduction in engineering can be gained through the use of standardized products and design details. Construction costs can be reduced by using labour in a more efficient way through the prefabrication, preassembly an efficient use of construction materials. Several key factors need to be considered in order to get the needed impact on the construction costs and schedule. Factors include contracting strategy, methods and techniques of construction and sequencing. Measurement of these factors can be obtained by evaluating the impact of these changes between the standard practices. (Russel et al. 1994) (Fig. 2). The implementation of value engineering includes six steps, information, functional analysis, speculation, evaluation, planning/ proposal and implementation/follow-up. Value engineering can be done in two ways, either proactively or reactively. The proactive approach includes starting to collect ideas at the beginning of the design, and multiple design options are considered and most cost effective will be chosen.

Knowledge Sharing in Constructability Management

89

Fig. 2. Cost and schedule savings when using constructability principles (Russel et al. 1994)

The reactive approach collects cost effective alternatives through the design reviews, it is performed after the design of specific component is complete. The primary objective of value engineering is reduce the total life cycle cost of a facility, whereas constructability focuses on the optimization of the entire construction process. Normally VE is performed during the design phase of the project, and effective constructability program ideally begins during the conceptual planning phase of the project and lasts through the construction (Russel et al. 1994) (Table 2). Table 2. Comparison of value engineering and constructability (Russel et al. 1994) Criteria Focus

Value engineering Overall reduction of life-cycle cost

Implementation

A brainstorming session where life cycle cost alternatives are considered for systems components while maintaining design function Usually performed during design phase. In many cases performed as a reactive process to reduce the cost after design has been completed

Timing

Constructability Optimize construction process in terms of construction cost, schedule, safety and quality An integral part of project management and scheduling allowing construction knowledge and experience to be integrated into project planning and design On-going from conceptual planning through construction and start-up

90

M. Naaranoja and J. Vares

Construction optimization and achieving of the lowest life cycle cost can be developed at the same time. VE recognizes the increased benefit from early implementation, however information available during the design and planning phase of the project is normally limited. The implementation of constructability can be as a precursor to VE, by providing information though construction input and lessons learned, that VE could be more effective. (Russel et al. 1994) Total Quality Management, Constructability and Value Engineering are not mutually exclusive. Constructability and Value Engineering are complementary work processes that might be used as key elements in achieving the total quality (Russel et al. 1994) (Fig. 3).

Fig. 3. Constructability and Value Engineering related to the Total Quality Management (Russel et al. 1994) Table 3. Most common constructability implementation barriers (Gambatese et al. 2007) Rank 1 2 3 4 6 7 8 9 10 11 12 13 14 15 16 17 18

Description Complacency with the status quo Reluctance to invest additional money and effort in early project stages Limitations of lump sum competitive contracting Lack of construction experience in design organization Designer’s perception that “we do it” Construction input is requested too late be value of Belief that there are no proven benefits of constructability Owner’s lack of awareness/understanding on the concepts of constructability Misdirected design objectives and designer performance measures Owner’s perception that “we do it” Lack of genuine commitment to constructability Designer’s lack of awareness/understanding of the concepts of constructability Poor communication skills of constructors Lack of documentation and retrieval of “lessons learned” Lack of team-building or partnering Poor timeliness of constructor input The right people were/are not available

Knowledge Sharing in Constructability Management

91

Fragmented and split scope of the project and unclear roles of the project stakeholders are the significant barriers to implementation and developing of the constructability program successfully. Competitive stance or motivation of construction personnel, obstacles to innovation in design and demarcation of the project are the difficulties and barriers, which create to the further development and use of constructability concepts will not be easy to adapt (Gambatese et al. 2007) (Table 3).

3 Research Method and Results The relatively small amount of interviewees – namely 18 persons - has impact on the reliability of the research. However, since the interviewees shared most of the time the opinions the interview results indicate the real life situation in the company and there was no need to add more interviewees. The interviews worked either as project team members or project managers or had a role of a site manager or a section manager. Unfortunately the interviewees were only from one organisation and did not cover client representatives or other stakeholders. However, the interviewees covered variety of disciplines like mechanical, electrical and civil engineering. All the interviews were recorded during the session, and all the results from the interviews were anonymized. The interview focused on the current use and needed use of constructability principles. This paper analyses the knowledge management practice development need related to the use of constructability principles. The research is qualitative and the interview questions were analysed question by question.

3.1

Results

According to the interviews the unconscious and conscious constructability principles are taken into account mainly during the general planning of the project and as a part of project management process. The infrastructure projects in the company have a structured decision making process, but the time for planning is limited, there are difficulties to take into account the local environment so the construction limitations are often not clear, the clients might not be experts in the infrastructure field but the goals are clear for the project group, risk assessment is done as carefully as possible in the time frame, ICT systems are used, and change management procedures are well defined. The constructability was not systematically reviewed in the infrastructure projects under one single process, it was buried under other processes such as risk management, project planning and design review. The constructability process had challenges in non-standard or complex projects and markets. In other words, the use of constructability was not taken into account when the knowledge management of the project was challenging and the lack of specific knowledge prevented the use of constructability principal. The knowledge sharing did not take place due to the lack of knowledge. In other words, knowledge sharing did not happen when there was nothing to share or there was not enough time for sharing and thus the constructability level did not reach to goal.

92

M. Naaranoja and J. Vares

The case data revealed the willingness for knowledge sharing though the actual sharing did not take place due to lack of time to get knowledge of the complex project. The constructability usage level and quality was seen to be very much linked to the competences and experience of personnel at the company. Thus first step to improve knowledge sharing is to teach or consciously improve the process in order to take more time to learn in the beginning of the project or analyse the constructability. Improving the use of constructability principles requires not only improving the knowledge of constructability principles both in the system and individual levels but also selecting the experienced knowledgeable staff for complex projects. The results indicate that the knowledge sharing behavior in constructability management depends partly on the norms of sharing (like risk management, project planning and design review). The attitudes of sharing were weaker when the project was complex and the actual knowledge can be evaluated to be created too late if the personnel was unexperienced. The results also indicate that the project parties were willing to share knowledge but there was not enough time to study the project and learn to know what needed to be shared.

4 Discussion The results indicate that the norms of sharing in this case can be defined to be the constructability principles and that the norm was well understood in the organisation. Short term sharing was not always possible due the short of time to create knowledge that could be shared and thus the actual sharing behaviour was not realised. Thus we propose to add the knowledge creation into the influence of incentives (Fig. 4).

Fig. 4. The influence of incentives on knowledge sharing (modified Laitinen and Sanoo 2016)

Knowledge Sharing in Constructability Management

93

5 Conclusion and Future Works In this paper, we present a knowledge sharing challenge in constructability ensuring process. Knowledge sharing behaviour in constructability management depends partly on the norms of sharing (like risk management, project planning and design review) and on the situation like hurry but also the knowledge of the employee. The attitudes of sharing were weaker when the project was complex and the actual knowledge can be evaluated to be created too late if the personnel was unexperienced. Experienced persons found it easy to share knowledge even in case of hurry situation since they understood the situation and benefit of sharing. In the future, we should plan to improve the constructability process by using co-creation of value. This is supposed to give more detailed knowledge on how to share the knowledge in each stage.

References Aerts, G., Dooms, M., Haezendonck, E.: Knowledge transfers and project-based learning in large scale infrastructure development projects: an exploratory and comparative ex-post analysis. Int. J. Proj. Manage. 35, 224–240 (2016) Alalawi, M., Ali, M., Johnson, S., Han, S., Mohamed, Y.: Constructability: capabilities, implementation, and barriers (2015) Blumentritt, R., Johnston, R.: Towards a strategy for knowledge management. Technol. Anal. Strat. Manage. 11(3), 287–300 (1999) Brady, T., Davies, A.: Building project capabilities: From exploratory to exploitative learning. Organ. Stud. 25(9), 1601–1621 (2004) Carrillo, P.M., Anumba, C.J., Kamara, J.M.: Knowledge management strategy for construction: key IT and contextual issues. Proc. CIT 2000, 28–30 (2000) Chou, J.S.Yang, Yang, J.G.: Project management knowledge and effects on construction project outcomes: an empirical study. Proj. Manag. J. 43(5), 47–67 (2012) Clarke, T.: The knowledge economy. Education + Training 43(4/5), 189–196 (2001). doi:10. 1108/00400910110399184 O’Connor, J.T.: Barriers to constructability implementation. J. Perform. Const. Facil. 8(2). American Society of Civil Engineers (1994) Construction Industry Institute. Constructability Concepts File, Construction Industry Institute, Austin (1987) Davies, A., Brady, T.: Explicating the dynamics of project capabilities. Int. J. Proj. Manage. 34 (2), 314–327 (2016). doi:10.1016/j.ijproman.2015.04.006 Gambatese, J.A., Pocock, J.B., Dunton, P.S.: Constructability concepts and practice. American Society of Civil Engineers (2007) Gasik, S.: A model of project knowledge management. Proj. Manag. J. 42(3), 23–44 (2011) Laitinen, J., Senoo, D.: Understanding the signaling information of incentive programs. In: European Conference on Knowledge Management, Kidmore, pp. 938–944 (2015) Naaranoja, M., Uden, L.: Major problems in renovation projects in Finland. Build. Env. 42(2), 852–859 (2007)

94

M. Naaranoja and J. Vares

Nonaka, I., Toyama, R., Konno, N.: SECI, Ba and leadership: a unified model of dynamic knowledge creation. Long Range Plan. 33(1), 5–34 (2000) McGeorge, D., Zou, P.X.: Construction Management: New Directions. Wiley, New York (2012) O’Connor, J.T., Miller, S.T.: Barriers to constructability implementation. J. Perform. Constr. 8 (2), 110 (1994). http://dx.doi.org/10.1061/(ASCE)0887-3828 Russell, J.S., Swiggum, K.E., Shapiro, J.M., Alaydrus, A.F.: Constructability related to TQM, value engineering, and cost/benefits. J. Perform. Constr. Facil. 8(1), 31–45 (1994)

Knowledge Transfer and Learning

Skills Sets Towards Becoming Effective Data Scientists Wardah Zainal Abidin(&), Nur Amie Ismail, Nurazean Maarop, and Rose Alinda Alias Advanced Informatics School, Universiti Teknologi Malaysia, Kuala Lumpur Campus, Kuala Lumpur, Malaysia {wardah,nurazean.kl,alinda}@utm.my, [email protected]

Abstract. The tsunami of data and information brings with it challenges in decision making to organisations especially so to the government sector. Decision making is vital in ensuring effective service delivery to constituents. Nowadays, Big Data Analytics (BDA) tools and software are readily available but what is lacking are the skills and competency of the personnel to handle and manage these data. The Government of Malaysia requires its IT officers to assume a more important role to extract data and turn into valuable information which is beneficial to operations and planning. However initial findings revealed that these IT officers are lacking in data scientist skills. Therefore, there is a need to propose a guideline on the direction towards acquirement of these skills to become data scientists. This paper presents the findings conducted recently via experts’ views using the Delphi technique, regarding data scientist skills required by the Government IT officers. The findings revealed 46 in-service skill sets which are deemed mandatory for IT officers to have, of which the top seven being analysis, data visualisation, data modelling, decision making, ethics, communication and basic database knowledge and skills. This data is helpful towards building a Data Scientist Competency Development Roadmap for the next five years as a stop gap measure before data scientists graduates are churned from universities. Keywords: Data scientist skills
Data science
Data science curriculum
Competency development
Roadmap
Big data
Big data analytics
Tsunami of data

1 Introduction BDA or Big Data Analytics is a mechanism to improve Public Service delivery to citizens and businesses as well as to assist top management in organisations to make accurate and effective decision making [1]. In 2015, the very first Malaysian public sector BDA project was started and developed by external consultants. The most common issues of IT outsourcing are over-reliant on third party, cost to the Government, and lack of internal expertise [2]. In any BDA projects, data scientist plays an important role. The data scientist is an expert who is capable of extracting meaningful value from the data and also managing the whole lifecycle of data [3]. Currently in © Springer International Publishing AG 2017 L. Uden et al. (Eds.): KMO 2017, CCIS 731, pp. 97–106, 2017. DOI: 10.1007/978-3-319-62698-7_9

98

W.Z. Abidin et al.

Malaysia, there are only three hundred (300) data scientists whereas MDeC, Malaysian Digital Economic Corporation, has targeted to have 2,000 data scientists by 2020. This means, the country is currently facing shortages and needs to produce an approximate of 425 data scientists per year. To increase this number several programmes and initiatives were carried out such as conferences, trainings, and certifications. However, these are still insufficient. Several central government agencies (namely JPA, INTAN, and MAMPU) have launched the Capability Development Roadmap (CRD) for IT Services but the roadmap does not include nor is specific to the data scientist competency development. Therefore, the Malaysian public sector needs a guideline on the direction and skills required to become a data scientist specialist. The objective of this paper is therefore to investigate the data scientist competency profile from global best practices which are necessary skills and competencies needed to overcome this gap which is faced by the government of Malaysia, especially in implementing its BDA projects effectively. The second objective is to determine the most important data scientist skills required by Malaysian public sector IT officers. To wait for the local universities to churn graduates specialising in this area will take at least four or five years more and this will prove to be detrimental in achieving the goal towards a digital government. Hence this paper shall report on the findings of a study conducted recently using the Delphi technique via engaging local and global data science and BDA experts to identify the needed skills which existing IT officers in the government sector of Malaysia must have for them to have a good understanding and capability in managing BDA projects. It will start with a literature review revolving the area of data science and big data, followed by a description of the research methodology and finally a discussion on the findings before providing suggestions towards the design of a framework which consists of guidelines on the direction and skills required to become a data scientist.

2 Literature Review 2.1

The Review Process

After the process of identifying the key words is done, the search started by focusing on journals and books related to the topics. List of computerized databases used in this study are Association for Computing Machinery (ACM) Digital Library, IEEE, Science Direct, Springer Link, Wiley Online Library, ERIC, Gartner, and Google Scholar. The journals and books reviewed are within the period of 2011 to 2016. A total of 17 papers are investigated in detail to furnish the research with a baseline set of skills of a data scientist.

2.2

Data Science

Nowadays, with the vast amounts of data available in the world, companies across the industry are focusing on exploiting data for their competitive advantage. Hence, a realization to hire data scientists or equip current employees with data scientist skills is

Skills Sets Towards Becoming Effective Data Scientists

99

apparent. [4] found that data science is closely related to the data mining concept. According to [5], data mining is the process of discovering meaningful new correlations, patterns, and trend by shifting large amount of data stored in repositories. Data mining uses pattern recognition technologies as well as statistical and mathematical techniques.

2.3

Data Scientist

Data scientists are valuable because they can discern the option value of large datasets [6]. According to [7], a data scientist is a practitioner who collects raw data and use his/her skill, knowledge, and experience especially in analytics to churn valuable information. On the other hand [8] define data scientist as someone who has high capability in problem solving. Overall one can conclude that roles and responsibilities of a data scientist are data cleaning for predictive analysis, data mining for patterns and interacting with data dynamically, and providing additional resources to acquire all necessary tools to effectively do the job.

2.4

Competency Development

In an organization, profitability, efficiency, and activeness are the three main points which have been emphasized the most. Therefore, organisations need employees who are dedicated and highly skilled in achieving the goals of the organization. The managers within the human resources department in any organization, will plan and develop programs such as training to improve skills and competencies of their employees. This is usually assigned to and allocated some budget and resources. Several definitions have been explored as listed in Table 1. Table 1. Definition of competency development Paper [9] [10] [11]

2.5

Definitions Refers to those activities carried out by the organization and the employee to maintain or enhance the employees’ functional, learning, and career competencies Complex ability for self-organization, which makes it possible to respond to constantly changing complex environments with new behavioural strategies An underlying characteristic of an individual which is causally related to criterion-referenced effective and/or superior performance in a job or a situation

Data Scientists Skills

This project manages to locate and identify 50 different reports of research in articles, journals, and books on the said topic. Seventeen (17) papers from global best practices including Malaysian public sector are selected. A list of 41 data scientist skills from these papers are shown in Table 2 according to five (5) categories adapted from [12]. The categories are computer science, analytics, data management, decision management, and entrepreneurship.

100

W.Z. Abidin et al. Table 2. Accumulated skills for a data scientist

No.

Category/Skills

A

Category: Computer Science Basic Programming Hadoop Python R Programming Java Ruby MapReduce

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. B 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26.

C/C ++/C# Clojure Cloud computing Distributed System Mahout Pig Privacy and Security System architecture Category: Analytics Machine Learning Statistic Analysis Mathematics Natural Language Processing Algorithm SAS Simulation Artificial Intelligence Probability Matlab

Frequency in papers

No.

Category/Skills

C

Category: Data Management Basic Database

8

27.

6 4 4 4 2 2

28 29 30 31 32 33

1 1 1 1 1 1 1

34 35 36

Data Visualisation Data Modelling Data Mining SQL Hive Business Intelligence Data Manipulation Data Processing Data Warehousing

Frequency in papers

5 5 3 3 2 2 1 1 1 1

1 D 9 8 6 4 2

37 38 39

2

E

2 2 2

40 41

1 1

Category: Arts & Design Communication Decision Making Ethics

6 1 1

Category: Entrepreneurship Business Economic

6 1

Skills Sets Towards Becoming Effective Data Scientists

101

3 Research Methodology This overall research has five main phases, namely preliminary study, design, data collection and analysis, developing, and finalising. For the data collection, the Delphi technique was used by obtaining a consensus of expert opinions via a series of intensive questionnaires. This technique facilitates dialogue and interactions among the experts who cannot interact face-to-face. The Delphi technique is chosen because the subject matter is still nascent and is interpreted differently by the industry players. Hence by using Delphi a structured way of deriving at consensus can be achieved. In Delphi round 1, an open-ended questionnaire containing the 41 data scientist skills were put forth. The objective of this round is to examine the most important data scientist skills required by the Malaysian public sector IT officer in order to enhance their knowledge, skills, and capabilities in data science. Six (6) Likert scale (5 = Extremely important, 4 = Very important, 3 = Moderate important, 2 = Slightly important, 1 = Not important and 0 = I’m not Sure) was used. In Delphi round 2, an open-ended questionnaire was developed based on the result of Delphi round 1 together with a proposed DS-CDF (Data Scientist - Competency Development Framework) to be evaluated on its feasibility aspects if it were to be adopted for the Malaysian IT public sector. A binary scale (5 = Agree. This skill is important, 2 = Disagree. This skill is not important) was used.

4 Findings 4.1

Findings and Result of Delphi Round One

The questionnaire was distributed to seventeen (17) experts using email of which 15 responded. They rated 41 data scientist skills (from literature review) using a six (6) Likert scale (5 = Extremely important, 4 = Very important, 3 = Moderate important, 2 = Slightly important, 1 = Not important and 0 = I’m not Sure) to prioritise the most important data scientist skills. The Delphi round one process is conducted within two weeks from the invitation process by email, reminder to the participants, and receiving the feedback. Altogether seven male and eight female experts were involved. Age-wise 40% of them are within 31 to 40 years old, 20% are within 41 to 50 years old and 40% are 51 years old and above. Most of them are from the Government agencies (73.33%), two (2) experts from the private company (13.33%), one (1) from GLC (6.67%) and one (1) from university (6.67%). Most of the experts have more than five (5) years experience and majority of them have Masters (40%) and PhD (40%). A summary of the experts’ profile is depicted in Table 3. Based on mode, 41 data scientist skills were rated as extremely important, very important, and moderately important. Another five (5) skills added by experts are critical thinking, business strategic, forecasting, retail, and finance. Two (2) skills have been eliminated from the list (highest percentage under not important score) which are cloud computing and distributed system. Hence, 44 data scientist skills are considered for Delphi round two.

102

W.Z. Abidin et al. Table 3. Summary of experts’ profile for round 1

Expert

Gender Age Profession Type of (Years) organisation

Expert 1

Male

Expert 2

Male

Expert 3

Male

Expert 4

Male

Expert 5

Female Above 50 Female 31–40

IT Expert

Expert 8

Female Above 50 Female 31–40

IT Consultant IT Officer

Expert 9

Male

Expert 10 Expert 11

Above 50 Female 31–40 Female 31–40

IT Consultant Lecturer IT Expert

Expert 12

Male

Expert 13

Female 41–50

Data Analyst IT Expert

Expert 14

Female 31–40

IT Officer

Expert 15

Male

Director

Expert 6 Expert 7

Above 50 Above 50 Above 50 31–40

41–50

41–50

IT Consultant Data Scientist IT Expert IT Expert

IT Expert

Government Agency Private Company Government Agency Government Agency Government Agency Government Agency Government Agency Government Agency Government Agency University Government Agency Private Company Government Agency Government Agency Government Link Company (GLC)

Experience in data science/big data analytics/database 1–10 Years

Qualification

PhD

21 Years and Above 21 Years and Above 11−20 Years

PhD

Master

1–10 Years

Master

11–20 Years

Degree

1–10 Years

PhD

1–10 Years

Degree

21 Years and Above 11–20 Years 1–10 Years

Master PhD Master

11–20 Years

Master

1–10 Years

PhD

1–10 Years

Degree

1–10 Years

PhD

Master

Table 4 summarises the findings for Round 1 and Round 2, showing the selected skills while Table 5 shows the rejected skills.

4.2

Findings and Result of Delphi Round Two

The objective of Delphi round two is to finalise results from Delphi round 1. In Delphi round two, the questionnaire was distributed to fifteen (15) experts using email. However, only eight (8) experts responded.

Skills Sets Towards Becoming Effective Data Scientists Table 4. Summary of the findings for round 1 and 2 Category/Skills Analysis Data Visualisation Data Modelling Decision Making Ethics Communication Basic Database Algorithm Business Hadoop Statistic Data Manipulation Data Processing SQL Hive Economic Mathematics R Programming Basic Programming Python Probability Artificial Intelligence NLP MapReduce Data Mining Data Warehousing Machine Learning Matlab Business Intelligence System architecture C/C ++/C# Java SAS Privacy and Security Pig Clojure Simulation Mahout Scala

After Delphi R1 5 - (1) 5 - (2) 5 - (3) 5 - (4) 5 - (5) 5 - (6) 5 - (7) 4 - (1) 4 - (2) 4 - (3) 4 - (4) 4 - (5) 4 - (6) 4 - (7) 4 - (8) 4 - (9) 4 - (10) 4 - (11) 4 - (12) 4 - (13) 4 - (14) 4 - (15) 4 - (16) 4 - (17) 4 - (18) 4 - (19) 4 - (20) 4 - (21) 4 - (22) 4 - (23) 3 - (1) 3 - (2) 3 - (5) 3 - (6) 3 - (7) 3 - (6) 3 - (10) 3 - (11)

After Delphi R2 No change No change No change No change No change No change No change No change No change No change No change No change No change No change No change No change No change No change No change No change No change No change No change No change No change No change No change No change No change No change No change No change 3 - (3) 3 - (4) 3 - (5) No change 3 - (7) 3 - (8) Added 3(9)

RANK From FINAL Literature 1 18 2 28 3 29 4 38 5 39 6 37 7 27 8 21 9 40 10 2 11 17 12 34 13 35 14 31 15 32 16 41 17 19 18 4 19 1 20 3 21 25 22 24 23 20 24 7 25 30 26 36 27 16 28 26 29 33 30 15 31 8 32 5 33 22 34 14 35 13 36 9 37 23 38 12 39 (continued)

103

104

W.Z. Abidin et al. Table 4. (continued) Category/Skills

After After Delphi R1 Delphi R2 Apache Spark Added 3(10 NoSQL Added 3(11) Social Network Analysis Added 3(12) Storytelling Added 3(13) Business Strategy Added 3(14) Forecasting Added 3(15) Critical Thinking Added 3(16) TOTAL 44 46

RANK From FINAL Literature 40 41 42 43 44 45 46 46 41

Table 5. List of skills rejected after Round 1 and 2 Category/Skills

From Literature Ruby 6 Cloud computing 10 Distributed System 11 Retail NA Finance NA

Rejected After Delphi Delphi Delphi Delphi Delphi

R R R R R

2 1 1 2 2

Table 6. Summary of experts’ profile in Round 2 Expert

Gender Age

Profession

Type of organisation

Experience in Qualification data science/big data analytics/database

Expert 1

Female 31–40

IT Officer

1–10 Years

Degree

Expert 2

Male

31–40

IT Expert

11–20 Years

Master

Expert 3

Male

50 and Above IT Consultant

Female 31–40

20 Years and Above 1–10 Years

PhD

Expert 4 Expert 5

Male

50 and Above IT Consultant

Male

41–50

20 Years and Above 11–20 Years

PhD

Expert 6 Expert 7 Expert 8

Female 41–50 Lecturer Male 50 and Above Data Scientist

Government Agency Government Agency Government Agency Government Agency International Company International Company University International Company

IT Officer

Data Scientist

1–10 Years 20 Years and Above

Degree

PhD PhD PhD

Skills Sets Towards Becoming Effective Data Scientists

105

The experts rated 44 data scientist skills based on result of Delphi round 1. Binary (2) Likert scale (5 = Agree. This skill is important, 2 = Disagree. This skill is not important) was used to finalise the most important data scientist skills. The Delphi round two process was conducted within two (2) weeks from the invitation process by email, reminder to the participant, and receiving the feedback. There are eight (8) experts involved, 5 male and 3 female. In terms of age range, 37.50% experts are within 31 to 40 years old, 25.00% are within 41 to 50 years old and 37.50% experts are 51 years old and above. Most of them are from Government Agency (50.00%), three (3) experts from the international company (37.50%), and one (1) expert from university (12.50%). Most of the experts have more than five (5) years experience and 62.50% of the experts have PhD. Table 6 shows the summary of the experts’ profile in Round 2. From the 44 skills shown, 41 was rated as important by the experts whereas three (3) skills were rated not important namely Ruby, Retail, and Finance. In this round, the experts gave their full commitment by giving many suggestions and recommendations on how to improve the DS-CDF and have added new data scientist skills. See Table 7. Table 7. Summary of comments from experts Expert Expert 1

Feasibly for adoption Yes

Expert 2

No

Expert 3

Yes

Expert 4

Yes

Comments “Currently, every skills needed for each level is appropriate and can be achieved by Government IT Officer”. “Too complicated and IT Officers usually need to cover multiple activities e.g. operation, security, application development, assets, administrative”. “To set out a clear direction in the skills and competency dev programme as well for DS career path’’. “Data Scientist is significant today as growth of big data technology. Perhaps Statisticians also can consider to be Data Scientist”.

5 Future Works & Conclusion This study achieved its research objective which is to examine the most important data scientist skills at the end of Delphi round one and two. Delphi round 1 experts have rated the most important data scientist skills based on the given categories derived from literature review. Based on mode, 41 data scientist skills have been rated as extremely important, very important, and moderate important. Another five (5) skills were added by the experts are critical thinking, business strategic, forecasting, retail, and finance. Two (2) skills were rejected (cloud computing and distributed system). In Delphi round two, there are 41 data scientist skills which have been rated as important by the experts whereas three (3) skills are not important namely Ruby, Retail, and Finance.

106

W.Z. Abidin et al.

This finding can be useful to help curriculum developers of data science programs in universities and industry. The skills demanded of a data scientist is such a tall order ranging from able to do programming until making decisions. Therefore, in the immediate situation to overcome the gap it is recommended that the works of a data scientist be resolved by the use of a team of people with various skills sets. Acknowledgments. This study is funded by Advanced Informatics School, Universiti Teknologi Malaysia (UTM AIS).

References 1. Chen, H., Chiang, R.H., Storey, V.C.: Business intelligence and analytics: From big data to big impact. MIS Q. 36(4), 1165–1188 (2012) 2. Deloitte: 4 IT outsourcing risks and how to mitigate them (2012) 3. Manieri, A., Demchenko, Y., Brewer, S., Hemmje, M., Riestra, R., Frey, J.: Data science professional uncovered: how the EDISON project will contribute to a widely accepted profile for data scientists. In: 2015 IEEE 7th International Conference on Cloud Computing Technology and Science Data (2015) 4. Provost, F., Fawcett, T.: Data science and its relationship to big data and data-driven decision making. Big Data 1(1), 51–59 (2013) 5. Larose, D.T.: Discovering Knowledge in Data: An Introduction to Data Mining. Wiley, Hoboken (2014) 6. Gehl, R.W.: Sharing, knowledge management and big data: A partial genealogy of the data scientist. Eu. J. Cult. Stud. 18(4–5), 413–428 (2015) 7. Mohanty, S., Jagadessh, M., Srivatsa, H.: Big Data Imperatives: Enterprise ‘Big Data’Warehouse, ‘BI’Implementations and Analytics. APress, New York (2013) 8. Shum, S.B., Hawksey, M., Baker, R.S., Jeffery, N., Behrens, J.T., Pea, R.: Educational data scientists: a scarce breed. In: Third International Conference on Learning Analytics and Knowledge (2013) 9. De Vos, A., De Hauw, S., Van der Heijden, B.I.: Competency development and career success: the mediating role of employability. J. Vocational Behav. 79(2), 438–447 (2011) 10. Adolph, S., Tisch, M., Metternich, J.: Challenges and approaches to competency development for future production. J. Int. Sci. Publ. Educ. Alternatives 12, 1001–1010 (2014) 11. Spencer, L.M., Spencer, S.M.: Competence at Work: Models For Superior Performance. Wiley, New York (1993) 12. Stadelmann, T., Stockinger, K., Braschler, M., Cieliebak, M., Baudinot, G., Ruckstuhl, G.: applied data science in europe–challenges for academia in keeping up with a highly demanded topic. In: European Computer Science Summit (2013)

E-Learning Platforms Analysis for Encourage Colombian Education Lizeth Xiomara Vargas Pulido ✉ , Nicolás Olaya Villamil, and Giovanny Tarazona (

)

Universidad Distrital Francisco José de Caldas, Bogotá, Colombia [email protected], [email protected], [email protected]

Abstract. The incorporation of Information and Communication Technologies in societies has been given according with countries development, for that reason national politics are directly involved in how society use technology and how the different sectors of economy advance depending of resources’ access. It is well known that the education level is not the unique factor that determines the devel‐ opment of a country. However, the evolution of this sector has determined how is the behavior of countries to manage knowledge and accept technology, reason why this document is done by authors in order to evaluate how technology works in education sector, what are the existence e-learning tools and which is the acceptation factor from teachers through some applied interviews in engineering faculties from Colombia (the studied scenery), to give a critical analysis about the impact of the use of learning management systems as a way to improve educational system and enhance general development in developing countries, keeping in mind e-learning platforms characteristics and need from Colombian context. Keywords: E-learning · Information and communication technologies (ICT) · Learning management systems (LMS)

1

Introduction

Education can be defined as the training aimed to develop moral, emotional and intel‐ lectual capacity according with the cultural and communal living, main objective that pretends also to generate knowledge on the process. Over the years, pedagogy and educational develop have focused into take real advantage of all resources in order to transmit knowledge, build and make grow up skills for accomplishing multiple proposes depending of formation type and subject. The outbreak of Information and Communication Technologies (ICT) have completely empowered the knowledge triangle, that is composed by a physical space (classroom), the subject matter (books and other tools), and the class (knowledge’s transmission). Nowadays, education moves around many styles, models and modalities, among its modalities are: in-person classes, face-to-face tutorials, virtual classes and a mixture of those. Whereby, books, games, and some other physical tools are being

© Springer International Publishing AG 2017 L. Uden et al. (Eds.): KMO 2017, CCIS 731, pp. 107–118, 2017. DOI: 10.1007/978-3-319-62698-7_10

108

L.X.V. Pulido et al.

changed into technology in order to make the education service and knowledge afford‐ able for anyone in the world [1]. There is where e-learning was born as a way to impart education under different styles and pedagogical models making proper use of technology, as of the interaction with technological devices, people get knowledge from subject material (as electronical books, virtual games and multimedia among others) and from other people (through tools like chats, video calls and some other). Such as ICT are in continuous growing the develop of e-learning as pedagogical model has grown up around the world, and given the quantity of the existence of e-learning tools, each time it is more difficult to choose the proper platform according with the tutors and users’ needs. The requirements that users and bidders can have from a e-learning tools vary according with all pedagogical structure and work area, but too with the development and social conditions of the place in which the tools are going to be used. It’s well known that technology is developed in order to facilitate life but its acceptation from people depends of many factors and total success on this process cannot be guaranteed, reason why this study emerges in order to know what are the most used platforms in a devel‐ oping country as Colombia and in a specific area as engineering faculties, what are the needs from teachers and students taking into account some interviews and what are the requirements to become, the most used tools in real functional e-learning tools, to extend the use of those and make a substantiated call of attention to Colombian and Latin American Society about how the correct use of available technology can strengthen education process and contribute to national development.

2

Methodology

This document presents the study done by authors to identify which of the existence elearning platforms are the most used and how those can be improved according with Colombian context, in first section there is an introduction that tries to explain font reasons of the research and the evaluation scenery used for it. After that conceptualiza‐ tion is done in Sects. 3 and 4, defining e-learning meaning and the characteristics such as functioning of learning management systems in order to let know theory basis on the scope according to available data. Section 5 gives a brief of the profile of the most used e-learning platforms taking into account all information got by authors from official documents, Sect. 6 shows all the results got from personal interviews done to engineering faculties teachers of some universities of Bogota city, as selected scenery of study and Sect. 7 tries to show an analysis from study results and Colombian reality to define which are the best e-learning platforms, what are their strengths and weaknesses and what is needed to improve the impact of e-learning in the country and for the development of educational sector. Finally, in Sect. 8 concluding remarks are presented.

E-Learning Platforms Analysis for Encourage Colombian Education

3

109

E-Learning

E-learning is a pedagogical model that mix different educational elements [2] like onsite and virtual classes, and practices in delayed and real time [3]; that is based in the correct use of telecommunications and computing resources in order to increase people’s knowledge [4], share and generate new knowledge. This pedagogical model (through its multiple appearances like “virtual campus”, “active classroom” and “virtual platform” among others) has become in the most spread out tool in the educational world, that makes it in a dynamic tool able to update in real time as well as provides individual and group work spans, what constitute it in the perfect work spot for students. E-learning is shaped as of the interaction of the next factors [5]: (a) Education as a constructive personal and group process. (b) Technology through the direct interaction with devices making use of ICT. (c) Organization and team work that configures the aim of teaching and learning process. According with some global register for the year 2011 e-learning industry moved around 35.600 million of dollars and for 2013 it got around 56.200 USD millions what suppose a growth of 55.2% in 3 years [1]. For 2015, e-learning growth prediction aimed that on-line teaching would duplicate until to overtake de 100.000 USD and that for 2017 the sector is going to be able to reach 225.000 USD million. On the other hand, aggregate growth rate of the industry goes on 7.6%, that in regional statements let to Latin America with a growth rate of 15.2% for 2016 [6]. This industry is developed through e-learning platforms or Learning Management Systems (LMS) that are the adapted interface to permit the correct functioning of educational processes.

4

Learning Management Systems

Web education has resort to standardized tools traditionally, to conduct the interaction among the different participants of the system (trainers, learners, overseers, tutors and system operators). The set of tools received the name of E-learning platforms or LMS, acronym of Learning Management Systems, that are in charge of offering a held envi‐ ronment in which educational institutions can manage forums, feedback tools, virtual content and communication instruments with a view to provide friendly and comfortable experience to the final user (the student) [7]. An E-learning platform can be defined as a software or a set of computing applica‐ tions that allow to create and manage spaces destined to teaching and learning via internet. It is on charge of the management and control of users, courses, communication services and some other activities of virtual education [3, 8, 9]. Its main purpose is to manage learners’ process and performance monitoring of every person associated to training activities on the system [10]. There are three kind of LMS, developed according to general system needs, being the first of those commercial platforms, nowadays those are very versatile in order to facilitate the development of virtual courses and the accomplishment of proposed objec‐ tives for this, in all virtual roles (academic or manager) [11]. Second kind are free

110

L.X.V. Pulido et al.

software platforms, with free access on software by public license, but it does not mean that those are free of charges by use [12], and the last kind of platforms own to private developments that are done to fulfill specific requirements. In order that those platforms accomplish the expected functions, it is necessary to have some minimal tools like [8]: • Control tools and/or content distribution: Makes reference to the storage and files’ organization. • Communication tools: That allow to interact among all participants (teachers and students), those can be synchronous (like chats) or asynchronous [13]. • Follow-up and evaluation tools: To elaborate and classify the different kinds of eval‐ uation and self-assessment [14]. • Management tools: Those organize users’ registration, passwords, roles, among other tasks. Additionally, those allow the creation, modification and elimination of educa‐ tional spaces. • Group management tools: To have flexibility on the modification of students for the purpose of generating collaborative work in virtual scenes. Every e-learning platform should have the next characteristics [8, 15]: • Interactivity: It is associated with the communication among students and teachers, keeping in mind that platforms should have the capacity to generate the greater possible integration between the platform content and the learners, pretending to become them the protagonist of their own learning. • Flexibility: With the capacity to customize in an easy way to the environment to be implemented. That covers the adaptation of the organization structure, to the study methods, content and pedagogy. • Scalability: As the platform capacity to keep a good quality level independently of the number of registered and connected users. • Standardization: As the proper use of resources from third parties (courses and mate‐ rials), generating new virtual material able for those who use the platform. Besides, it pretends to avoid the traffic of obsolete information. • Usability: To do own task to achieve main goals as efficacy, efficiency and effec‐ tiveness. • Functionality: According with needs and requirements of users. It is directly related with the scalability. • Persuasion: To generate loyalty on every participant of the platform. • Accessibility: Facility to get quick and possible free access to the on-line system.

5

Common Used Platforms

Taking into account general information on the available data and studied authors for conceptual framework the most used learning management systems are presented towards to have a clear comparison about their properties and requirements, that allow to show according with the study which is the best for Colombian society, especially for engineering faculties’ teachers, as described in the next section.

E-Learning Platforms Analysis for Encourage Colombian Education

111

(a) A Tutor: It is a system of learning content’s management, of accessibility and adaptability to learners, teachers and stewards needs, including disabled people since it allows the use of assisted technology for web access [16]. Tutor’s software is designed in PHP and a little part in Java, it works with an Apache server, and database motor is MySQL. It can run on Windows, GNU/Linux, UNIX, Solaris and has support in 32 languages, through FAQ’S (instructions for the general installation and documentation of the platform) [15] One of its main characteristics is its easy access for people without knowledge on informatics, just using the personal e-mail. Although it’s interface is really friendly and simple to use, the interfaces by roles are very different, thus it is difficult for instructors to know what trainees have on the screen in order to give some technical assistance [10, 17] (b) Chamilo: Open code platform with free software and GNU/GPLv3, what means that this frame gives the liberty of using, modifying, improving or distributing its code. It works on Linux, Windows and OS-X systems, helps the teacher providing different pedagogical methodologies that can help to the student to gain knowledge in a person‐ alized way, what makes that learner strength communication skills in each course by means of tools like video conferences, customized interfaces, group announcements, etc. [18] Chamilo tries to ensure a great level of availability in education offer for the minimum cost by free distribution (with direct translation to 55 languages [10]) towards to arrive to place where education access is limited (as in developing countries like Colombia, the studied scenery) [19]. (c) Claroline: This platform allows virtual learning and work, its coverage is not just on educational field but in corporative too; free software with open code able to create new on-line content [15]. Developed under PHP language and managed by MySQL system, this LMS works on Linux, Unix, Mac OS X and Windows with some free web browsers as Mozilla and Netscape and Internet Explorer owners. It permits to publish content in any file format and it has available translation to 30 languages [10]. Among some task are student group creation, public and private forums, access to the student performance [17]. Despite that it is one of the most used platforms worldwide, an important lack of this one consist on the absence of video calls or video conferences, what limits its field of action. Furthermore, setting management is also limited compared with other platforms [16]. (d) Moodle: Learning management systems of open and free code. It counts with GNU public license, what means that modifications are allowed under permission for protecting the original license. It can be installed in any device that execute PHP language and works

112

L.X.V. Pulido et al.

with a SQL data base, it runs in its basic form in Unix, GNU/Linux, OpenSolaris, FreeBSD, Windows, Mac OS X, and NetWare among others [15, 20]. Moodle proposes around 20 different available activities, among those forums, glos‐ saries, task, surveys, research data bases and some others, all those easy to adapt to any course content and allowing to as students as trainers to upload information useful for the development of the course and future courses. It allows co-working between profes‐ sors for sharing subject information and getting touch through communication technol‐ ogies to get a well management of platform resources. Additionally, this system counts with consultant mode with which teachers and students can take decisions trough votes and the task unit has some management settings as checkpoint for delivery time. The biggest difficulty of this platform is represented by its unfriendly interface to the user [10, 17]. (e) Schoology: Cloud system platform, it does not require installation and saves memory on devices. It supports a great kind of files of content, included SCORM 2004. It works without problem in any device with IOS or Android system [21], allowing to sync google and some other servers facilities through a light application as some other platforms which count with application for different devices [22]. Its functioning is based on social networks what facilitates the communication among the participants and its multiple settings give to trainers the possibility to create multiple profiles depending of the quantity of courses or subjects they want to impart; on the other side, its aspect of social network permit to teachers to get connected for sharing experiences and knowledge [21]. Its main disadvantage is that just certified professors of educational institutions can create new courses and the existence of hacking of system [10]. (f) Blackboard: It is one of the commercial educational platforms that requires payment of access to develop its software, its accessibility is proposed by the World Wide Web Consortium (W3C) and it counts with wide support given that system managers, teachers, students it provides many online communication tools as free calls, online manuals, help communities and forums [16]. The greater advantage of this platform is that it allows the interaction with other learning management systems and some social networks from any device in real time [10]. Its virtual communities deliver the easy sharing and storing of users’ information, also counts with specific programs for disabled people and promotes co-working in and out of educational process [16]. Some of its facilities are files transfer, calendars, forums, shared board, audio records and video conferences, tools that reinforce the straight participation of users in the courses, the main weakness of this platform is the hard understanding of the interface and safety faults [17, 23].

E-Learning Platforms Analysis for Encourage Colombian Education

6

113

Comparison Study

Taking into account general information on the available data of the consulted sources, authors of the present document decided to apply a interview to around 100 teachers, to know which are the most used platforms and why. The study was applied in Bogota the capital city of Colombia, in engineering faculties, all that in order to know the specific needs of teachers in one or two of the roles offered by the learning management systems, avoiding the dispersion generated by different kind of interest, with the selection of some faculties and keeping in mind that engineering faculty is one of the most involved in the usage of technology and e-learning tools for developing all proposed skills in each engineering program. There is important too to denote that the interview is done under Colombian reality and knowing that collected information is surely biased, without giving importance to the selected community to apply the interview, information to be collected is going to have some dispersion because answers about the usage of e-learning tools as the eval‐ uation of Information and Communication Technologies implementation, it does not exist register about the field and it is not possible to pick up an exact region with uniform data, reason why the sample community selected can represent the reality about the use given by Colombian teachers to the learning management systems. Collected data is presented hereafter: According with results of the interview devel‐ oped 77% of teachers have knowledge about the meaning of E-learning, the other 23% do not register notions about the term, without forgetting that these percentage of people could have contact with one e-learning platform before. From people who knows about the concept of e-learning the reported usage of the different platforms named before showed a really strong tendency, as observed in Fig. 1, Moodle with a 78.8% and Black‐ board with 58.8% are the most used learning management systems, not just by teacher but by institutions too, for instance National Service of Education Learning denoted SENA (for its acronyms in Spanish) is a public institution of technical and technological studies which use blackboard to impart virtual courses and mix on-site modalities in order to implement ICT in all educational spaces, on the other hand as public as private

Fig. 1. E-learning platforms usage.

114

L.X.V. Pulido et al.

universities show a special affinity for Moodle (without exact register of how much institutions prefer it). Teacher were questioned about their roles at the moment of using a platform in order to know which percentage of them use the student role, registering in Fig. 2 a 66,7% for affirmative response and 33,3% for negative one, it means that more of the half of the studied community have knowledge about what do students receive from the platform, when they want to impart any course or complement the on-site contents with virtual material.

Fig. 2. Usage from teachers of student’s role in platforms

After that, teachers who registered the usage of the student role as shown in Fig. 3, qualify their experience using the different platforms as excellent as excellent in a 37.9% and outstanding with a 27.6%, the 29.3% told it was acceptable no good no bad, and the last 5.2% qualified it as bad, it means that virtual tools used to impart courses can be well selected and well used by professors, inasmuch as they know how the learners received all the information and what could be their possible limitations.

Fig. 3. Teachers experience in student’s role.

They were examined too for knowing if they have imparted virtual classes or courses, through any of these platforms, answers registered in Fig. 4, 64.1% for affirmative response and 35.9 for negative; teachers that gave positive answer expressed their favoritism for Moodle and Blackboard platforms to convey their courses.

E-Learning Platforms Analysis for Encourage Colombian Education

115

Fig. 4. Usage of e-learning platforms to impart virtual classes or courses.

The use of e-learning tools is a way to get connected with technology un an advan‐ tageous form, because knowledge is in constant broadcast and growth. Professors feel motivated to use these learning management systems because they can offer different tools to help students in learning process, their knowledge can be imparted from any place in the world and at any time too and share virtual content without caring about how large or extensive can it be. Moreover, two of most important characteristics of elearning platforms for teachers were feedback unit and an organized system of content, these parts are evaluated by them as vital to have a good experience with platforms, and most of them registered great experiences using the platforms as professors (Fig. 5 shows that 49,9% qualify it as excellent and 30.2% as outstanding, 17% considered it acceptable and the 2.9% expressed their experience as bad, according with statistics from personal interviews).

Fig. 5. Teachers experience in trainer’s role.

Finally, teachers told that the biggest difficulty for them and their colleagues is the existence lack of knowledge and practice in the use of Information and Communication Technologies, such as the proper use of new devices, the need of implementation of elearning tools in institutions and the hard access to technology resources in some places.

116

7

L.X.V. Pulido et al.

Comparison Study

All what was mentioned in the above section represent the Colombian reality about how e-learning has been incorporated in real life, without forgetting that there is not exact available data which gives reason to pick up a scenery of study over other. The lack of data about official statistics in the field shows that maybe the sample community taken for the study is not enough to claim something specific, but there can be inferred some important things. First of all, is that a great quantity of teachers has knowledge about the existence of e-learning platforms even if they use it or not, that the most used platforms in Colombia are Moodle and Blackboard, because trainers think that those two systems are the most complete of the e-learning market and on the other hand, those are the most used by educational institutions in the country; most of the half of the interviewed people have use a learning management systems to impart or complement their courses and more of the 50% expressed their opinion about the importance of implementing technology in educational spaces as a way to reduce the illiteracy gap and enhance general develop‐ ment for the country. Results from some additional questions done in the personal interviews have shown that one of the most common ideas of teachers from their experience with the students’ role was outstanding in the way that platforms accomplish with learning process but they detected some specific problems that can have a bad reaction in educational processes like the lack of interactivity of platforms and the unfriendly face of the inter‐ faces. However, they recognize that this problem can be generate because teachers sometimes do not have training or do not care about the management of content. Furthermore, they reported that from trainers’ experience they have some difficulties to use platform tools and manage the courses. The named above makes necessary two important things, first of all skill training as for students as for teachers in order to get a proper use of the learning management systems (specifically Moodle and Blackboard for Colombian society) and second the demand by all participants of e-learning platforms, for solutions in light of problems generated by unfriendly interfaces and the faults on interactivity, a way in which it can be solve is with some modifications on code (in Moodle) evaluating the flexibility of that one or requesting some updates in this areas (specifically in Blackboard platform).

8

Concluding Remarks

According with the theory review and the applied study, there can be said that a great part of teachers recognizes the existence of e-learning as educational model and as the way to incorporate technology into the education process, even if their institutions have implemented those systems or not, that the most used platforms in Colombia are Moodle and Blackboard (keeping in mind that some of them work with the two platforms at time in different courses), and that these are of their preference because they consider those are the most complete of the ones that they know or can have in their hands.

E-Learning Platforms Analysis for Encourage Colombian Education

117

Results presented and their respective analysis not just let know the current situation of the country but it allow to consider two key actions in order in enhance the use of technologies as the offered by e-learning market, first of all the training about the use of those platforms for all participants, students, teachers, managers and operators, that because in Colombian society not everybody has direct access to technology, so there is a big quantity of people who do not know about what is the functioning of devices and how they can use information technologies for their lives and to get education with the same quality of face-to-face learning process. The other action, is that Colombian government must work to ensure the investment is generated in two aspects: the devel‐ opment of these platforms, because some of they are open source, so it could be modified to adapt them to our society requirements; and training teachers to open minds, seeking to reduce the gap between those who know or not the impact of e-learning on use of technology as a tool and educational model; All to promote educational development.

References 1. Santamans, J.M.: El mercado global del e-learning, vol. 53(9) (2013) 2. Area, J., y Adell, M.: e-Learning: Enseñar y aprender en espacios virtuales, en DE. Tecnol. Educ. La Form. del Profr. en la era Internet, no. January 2009, pp. 391–424 (2009) 3. Dans, E.: Educación online: plataformas educativas y el dilema de la apertura. Univ. y Soc. del Conoc. 6(1), 22–30 (2009) 4. Álvarez, R.B.: El e-learning, una respuesta educativa a las demandas de las sociedades del siglo XXI. Pixel-Bit Rev. medios y Educ, vol. 35, pp. 87–96 (2009) 5. Duart, J.M., Lupiáñez, F.: Las TIC en la universidad: estrategia y transformación. Rev. Univ. y Soc. del Conoc. 2(11), 111 (2005) 6. Docebo Report: E-Learning market trends and forecast: 2014 – 2016 Report. A Rep. by Docebo, no. March, p. 480 (2014) 7. Correa, J.M., Paredes, J.: Cambio Tecnologico, Usos De Platafor- Mas De E-Learning Y Transformación De La Enseñanza En Las Universidades Espa- Ñolas: La Perspectiva De Los Profesoresj. Rev. Psicodidácrica 14(2), 261–278 (2009) 8. Rodríguez, J.S.: Plataformas tecnológicas para el entorno educativo. Acción Pedagógica 14, 18–24 (2005) 9. Fernández-Pampillón Cesteros, A.: Las plataformas e-learning para la enseñanza y el aprendizaje universitario en internet. In: Las Plataformas Aprendizaje: Del Mito a La Real, pp. 45–73 (2010) 10. Castro, S.M., et al.: Analizamos 19 Plataformas de e-learning Investigación colaborativa sobre LMS, vol. 1 (2013) 11. Sánchez, A.G., Rosas, J.C.: Los componentes de un sistema de educación virtual: El subsistema tecnológico. Odiseo, Rev. electrónica Pedagog., vol. 5 (2005) 12. Stallman, R.M., Stallman, R.M.: Software libre para una sociedad libre (2004) 13. Cejudo, M.L.: Moodle como entorno virtual de formación al alcance de todos. In: Comunicar, pp. 197–202 (2007) 14. Jäntschi, L., Bolboacǎ, S.D., Marta, M.M., Laszlo, A.: E-Learning and e-evaluation: A case study. In: 2008 Conference on Human System Interactions HSI 2008, pp. 840–845 (2008) 15. Cejudo, S.D.: Elearning. Análisis De Plataformas Gratuitas, p. 118 (2003)

118

L.X.V. Pulido et al.

16. Raquel, E., Arguello, G., Jenny, C., Enriquez, R.: Evaluación de plataformas tecnológicas para la teleformación o e-learning para el ámbito universitario, tomando como caso de estudio educativa (2006) 17. Modelo, P., Motic, P., Mienert, I.A.: Características comparativas de plataforma de código abierto desarrollada expresamente para modelo pedagógico motic, vol. 2 18. Pineda, V.O., Leyva, J.T.: Entornos virtuales de enseñanza y aprendizaje: la educación entre lo presencial y lo virtual Virtual environments for teaching and learning, vol. 10, pp. 435–442 (2013) 19. Académico, V.: Universidad Nacional de Educación ‘Enrique Guzmán y Valle’ La Cantuta (2010) 20. Marcela, M., Lic, C., Pianucci, I., Lucero, M.: Criterios de Evaluación de Plataformas Virtuales de Código Abierto para Ambientes de Aprendizajes Colaborativos. Ambient. Virtuales como apoyo al Aprendiz. Colab. Dep. Informática, pp. 1–5 (2010) 21. Moodle, C.D.E.E., Sidweb, S.Y., Jacqueline, L., Bonifaz, R.: Instituciones De Educación Superior En El Ecuador Durante El Proceso De Enseñanza-Aprendizaje, vol. 1999, pp. 331–340 22. Castillo, P.M.C.: Análisis psicopedagógico desde el paradigma constructivista de una estrategia de aprendizaje a través de la plataforma virtual Schoology para favorecer la Orientación vocacional de estudiantes del Colegio Marista de Alajuela, p. 190 (2014) 23. Sara, F.: Trabajo Final del programa FONTE (Formación Online de Tutores de Español) de Dina Krochuk Título: Estrategias de aprendizaje

Using a Simulation Game Approach to Introduce ERP Concepts – A Case Study Marjan Heričko1 ✉ , Alen Rajšp1, Paul Wu Horng-Jyh2, and Tina Beranič1 (

)

1

2

Faculty of Electrical Engineering and Computer Science, University of Maribor, Maribor, Slovenia {marjan.hericko,alen.rajsp,tina.beranic}@um.si School of Science and Technology, Singapore University of Social Sciences, 461 Clementi Road, Singapore 599491, Singapore [email protected]

Abstract. ERP systems are huge and complex systems, therefore, the introduc‐ tion of these solutions into a study program and process is quite challenging. Especially if there is only a small ERP course within the scope of which students are expected to gain an understanding of ERP systems from both a functional (business process) and implementation perspective. Fortunately, business simu‐ lation games have proved their efficacy in enhancing the learning of businessrelated subjects. In this paper we present some preliminary results of a study aimed at investigating the appropriateness and effectiveness of the business simu‐ lation game approach for introductory ERP lectures/classes at the Master Degree ERP course provided to students of informatics and technologies of communi‐ cation. The results demonstrate that the Game Simulation method was well accepted by both students as well as instructors, and recognized as a valuable teaching method for an ERP course. Keywords: Knowledge acquisition · ERP system · Business simulation game · ERP course

1

Introduction

Enterprise Resource Planning (ERP) systems are packages of software composed of several modules (e.g. human resources, sales & distribution, finance & accounting, production) made to provide cross-organization integration of data with the use of embedded business processes [1]. Highly volatile and changing markets created a need in companies for effec‐ tive ERP systems to maintain their competitive advantages, by closely integrating the flow of material, finance and information [2]. Systems with such a broad range of closely inte‐ grated functionalities are very complex. They present a challenge for new users, who can struggle to grasp their essence and efficiently use their functionalities. Even navigation within such a system can be a challenge for a beginner, including ERP course participants. When selecting an appropriate approach to introduce the basics of the ERP systems and core business processes and functions we must understand that our students are digital natives. Unlike preceding generations, digital natives are more inclined to create © Springer International Publishing AG 2017 L. Uden et al. (Eds.): KMO 2017, CCIS 731, pp. 119–132, 2017. DOI: 10.1007/978-3-319-62698-7_11

120

M. Heričko et al.

and construct their own knowledge rather than receive pre-made instructions. They prefer to discover new knowledge and learn in a collaborative learning environment while also exhibiting a preference for teamwork incorporating cooperative learning and constructivist principles [3]. Further, they prefer to learn in flexible, personalized and customized schedules, in an environment that makes learning interesting; in structured environments; in environments that use technology to enable them to be more productive and connected; in environments in which individuals are respected and all members of the group are supported; in environments that are goal and achievement oriented [3]. They expect immediacy in all that they do [4]. Digital natives possess sophisticated knowledge of, and skills using, information technologies. Because of their upbringing and experiences with technology, digital natives have particular learning preferences or styles that differ from earlier generations of students [5]. The digital native generation is the most accustomed to technology generation ever. This also has some drawbacks, such as an extremely short attention span, as we as instructors have noticed in our own classes. Thus, it is necessary to introduce more effective techniques and approaches such as business simulation games. Business simulation games are educational tools that, by utilizing technology, allow students to make business decisions typically in a roundbased, controlled, and risk-free environment. There is no doubt that the use of simulated activities is becoming recognized as an important tool that might assist in digital native education. Since they are “hands-on”, students are actively involved and they become participants, instead of mere listeners or observers. Studies have shown that knowledge retention is directly correlated with the involve‐ ment of the person in the subject at hand. The knowledge retention rate of the hands-on learning approach is estimated to be up to 75% (Fig. 1), which is only preceded by the notion of teaching others, according to research conducted by the National Training Laboratories [6].

Fig. 1. Pyramid of learning (adapted from: [7])

Different ways of delivering knowledge exist. In the scope of the ERP courses, a large variety of teaching methods are being used - lectures and practical exercises and projects are still prevalent [8]. We examined the potential of gamification to introduce

Using a Simulation Game Approach to Introduce ERP Concepts

121

students to the basics of Enterprise Resource Planning (ERP) system (namely SAP ERP). According to [9] Gamification is adding mechanics of gaming to non-game activities to change people’s behavior. In a business context, it is the process of integrating game dynamics into a website, business service, online community, or marketing campaign to increase participation and engagement. This lead to simulation games being created for use in education. A broad definition of simulation games is [10] “Exercises that have basic characteristics of both games and simulations, players are constrained by a set of explicit rules to that game and by a predetermined end point. The elements of the game constitute a more or less accurate representation or model of some external reality with which players interact by playing roles in much the same way as they would interact with reality itself.” Research conducted in 2012 [8] at German-speaking universities found that simu‐ lation games are only used in 7% of all ERP courses. Other teaching methods in ERP courses include lectures, practical exercise, case studies, projects, seminars and assign‐ ment papers – all of them having higher relative frequencies than simulation games. In order to improve and enhance students’ experiences, we decided to introduce an ERP simulation game developed and researched by the HEC Montreal business school [11]. This game teaches players with core knowledge and understanding of SAP, an ERP solution that is according to Gartner Research and [12] one of the leading ERP manage‐ ment systems. The purpose of our study was to investigate the appropriateness and effectiveness of a business simulation game approach for introductory ERP classes at the Master Degree ERP course provided to students of Informatics and Technologies of Communication. With these objectives, we aimed to find answers to the following questions: Do the students find the business simulation game to be a good approach to ERP introduction? and How much do they learn through the introductory session/workshop using the Busi‐ ness Simulation Game and the ERPSim tool, developed by HEC? This research paper is structured as follows. In the related works section, similar research in using ERP simulation games for knowledge acquisition and learning is presented. A section on Business simulation Games & Learning provides a theoretical background and history of business simulation (and similar) games as well as different domains in which this approach has already been introduced and applied. Subsequently, an in-depth description and presentation of the workshop execution is given, i.e. how the ERP simulation game was used and applied in our environment. Students’ feedback and final survey results are presented and discussed in the fifth chapter/section. The instructor’s insights section presents some findings and observations that are based on data obtained through semi-structured interviews of instructors and an online survey. In conclusion, the findings are summarized and some directions for further research are given.

2

Related Works

Quite a few similar studies have been done in the field of the development of ERP simulation games and teaching ERP through simulation.

122

M. Heričko et al.

In [13] the authors presented the developed distribution ERP game, the business simulation game used for introductory sessions in our ERP course. Leger et al. argue that by giving a firsthand experience of using an ERP system, students can learn the system and their core concepts very quickly, which is what we were aiming to achieve with our ERP introductory workshop. The principles of the game they developed, how it is meant to be played and recommendations about executing the workshop were also presented, but ended up different than our implementation. One iteration of the game is divided into three quarters, each lasting 20 game days (which take 25 min). In the first quarter/round, the participants only have a limited set of available functionalities, and later on additional functionalities are introduced and made available in the second and third quarter/round of the game. The HEC Montréal also published a paper [14] about the challenges faced by IT trainers in adapting to this innovative training approach and some general guidelines on how to create a sufficient learning environment using this approach. They also presented an example of a course on ERP that is partly supported for use of the distribution game, the introduction of a manufacturing game as well as an extended manufacturing game. A study on “Enhancing Student Learning of Enterprise Integration and Business Process Orientation through an ERP Business Simulation Game” is presented in [15]. It describes the execution of a full course focused on ERP, where 6 out of 13 weeks of the course were spent playing the simulation and an additional two weeks were based on the assessment of the ERP simulation (ERPSim) game assessment. They found that, overall, all knowledge increased significantly and that using a course that is partly provided as a game provided higher overall attitude of participating students, than tradi‐ tional ERP courses, that were being taught previously. A similar study on ERP simulation games [16] found that the simulation approach was proven effective at engaging and educating students about decision making and the effects of those decisions on business performance. The difference between these research works, except for the one [13] describing the distribution game, and our research, is that all the above-mentioned studies have focused on using ERP simulation games for the whole or at least half of the course whereas our focus was on using an ERP simulation game only for the introduction to the ERP system and related concepts. Our introductory ten hour workshop was given in three separate sessions (4 h + 4 h + 2 h).

3

Acquisition of Knowledge and Business Simulation Games

Knowledge acquisition in business simulation games follows the Simplified Kolbian experimental learning cycle [17] - he argues that learning and playing a business simu‐ lation can be divided into three steps as seen in Fig. 2: • Understand – The player must understand how to play the game and make some initial decisions or understand the feedback they have received (in later cycles). • Act – The player acts on their decisions • Reflect – The player reflects on feedback they receive.

Using a Simulation Game Approach to Introduce ERP Concepts

123

Fig. 2. Simplified Kolbian experimental learning cycle

The whole process of playing the game is called the experience. A brief history of business simulation games is presented in the journal article [18]. They found that the first studies in this field started around the year 1962. In 1974, the Association for Business Simulation and Experimental Learning (ABSEL) was founded. The topics that were presented at the ABSEL and used business simulation games for their support [18] are entrepreneurial skills, inventory management; specific job skills such as personnel administration, hiring, motivating, and leading; mathematical modeling; job-hunting skills; research and data analysis skills; collective bargaining; recruiting and applicant evaluation skills; creating advertisements; basic financial concepts; basic economic concepts; leadership skills; interpersonal skills; communica‐ tion skills; problem-solving skills, economic forecasting, conflict resolution, and the relationship between distinct business decision-making areas. Another study [19] reveals that direct predecessors of modern business simulation games can be found as early as 1932 in Europe. They also found that by the year 1961, over 100 business games were used in the US alone. Some of the major milestones that greatly expanded the spread of business simulation games as presented by [19] were: • Mainframe games (games that were executed on a mainframe computer), since previously games were hand written and as such much harder to operate, • Personal computers (PC): in 1984 the first PCs were sold by IBM • Graphical user interface (GUI): in 1985 Microsoft released the first Windows oper‐ ating system. GUI made games easier to operate and more user friendly • The World Wide Web: its invention in 1991 and spread afterwards made games possible to execute online. The first games were run on servers, but data was manually downloaded to local computers where calculations were performed and then uploaded back to the server (for each round). The next generation eliminated this need and nowadays (as of 2009), most online business simulation games are stored and performed on a central server.

124

M. Heričko et al.

Business simulation games are a field of study that has a lot of potential for study, and has been rapidly developing since its start.

4

Execution of the Introductory ERP Lectures

Since the focus (regarding the practical part, labs and exercises) of our ERP course has been on the SAP ERP solution we decided to use the ERP SIM distribution game to teach the basics of the SAP system as well as to cover the business perspectives of the ERP implementation. The introductory workshop was led by one of the coauthors of this paper, Dr. Paul Wu Horng-Jyh, who is a certified ERPSim instructor. ERPSim is the most basic implementation of ERP simulation games developed by HEC Montréal. The game challenges students to manage a water distribution company in Germany through a full business cycle (plan, procure, sell). In our game, this meant buying bottled water from a fixed distributor and selling it forward to stores as seen in Fig. 3.

Fig. 3. Simplified illustration of a player objective in the supply chain of the simulation

The players were given the freedom to use the following SAP transactions: Create Planned Independent Requirements (MD61); Material Requirements planning run (MD01); Automatic generation of Purchase Orders (ME59 N); Condition Maintenance: Change (VK32); ERPsim: Marketing Expense Planning (ZADS); ERPsim: Inventory Report (ZMB52); ERPsim: Price Market Report (ZMARKET); ERPsim: Summary Sales Report (ZVC2); ERPsim: Sales Report; ABAP Report: Financial Statements (F. 01); ERPsim: Purchase Order Tracking (ZME2 N). Not all transactions were available to participants from the first round, but were unlocked in rounds 2 and 3. The partici‐ pating students were divided into four competing teams. The details of the game used are described in [13]. However, we carried out the whole workshop with some notable differences, as can be seen in Fig. 4, especially regarding the team size and the progres‐ sive game pace, i.e. turn duration.

Using a Simulation Game Approach to Introduce ERP Concepts

125

Fig. 4. Structure of the ERP introductory workshop sessions

The team members were seated together for all rounds. Team sizes were 3 to 4 members, instead of 2. We played the game in three iterations instead of one. On day/ iteration 2 and 3 all functions were available from the first quarter (round). We changed the turn times, so that on the first day we used 60 s per turn, on the second day 45 s per turn and on the last day 30 s per turn. That was done because users are supposed to become more familiar with the system and would consequently need less time to adjust each parameter depicting the team’s business decisions. After the first and third day, all participants received a quiz of 20 questions to see how much knowledge they retained. Detailed information regarding the structure and objective of the quiz is presented in [20]. The initial introduction to ERP SIM was composed of a short presentation on ERP concepts and SAP functions, including what the goals and rules of the game were and an explanation about all of the transactions that are available. Participants also received official handouts prepared by HEC Montreal and that are part of the game. On them was a list of all the products and their prices, the names and the order of transactions that were meant to be used and a visual representation of the market structure. After each round the team results (i.e. profit made) were compared and discussed. This part was closely connected to the sharing strategies, where students and teams were encouraged to share strategies among themselves. As expected, profound strategies were completely revealed and shared primarily on the last day, because teams felt a bit of competitiveness among themselves and had “no advantage to lose” once the game was over. On the last day, we also carried out a survey aimed at collection data needed to find answers to our research questions. The results are discussed fully in the next chapter.

126

5

M. Heričko et al.

Students’ Feedback on the Business Simulation Game Approach

5.1 Research Method and Measures We prepared a survey to investigate how suitable and effective ERP simulation games are at introducing students to the domain of ERP systems and Business Management. The following research questions were formed: • RQ1 – How appropriate and effective is the business simulation game approach for ERP introductory sessions from the students’ perspective? • RQ2 – How appropriate is the concept of executing/applying the ERPsim game for the ERP introductory course? • RQ3 – Does a business simulation game increase the understanding and learning speed of core concepts of company/business management and the use of SAP basic functionalities and transactions? Data was collected from the workshop participants with an online questionnaire and from workshop instructors, including with semi-structured interviews. The number of obtained observations was 22, with 18 participants’ answers and 4 instructors’ views. The participant questionnaire consisted of questions and statements, where respond‐ ents were asked to evaluate and assess using a 5-point Likert scale. The statements were grouped into the following categories: • Statements about the concept of a business simulation game (4 statements) • Statements about the introductory workshop in general (9 statements) • Statements about knowledge, understanding and qualification of using each trans‐ action (11 statements) • Statements about the team aspect of the workshop (7 statements) Statements about knowledge, understanding and gained skills on particular transac‐ tion were assessed using a 5-point scale (Very Good, Good, Fair, Poor, Very Poor) and responses related to all other statements were on a 5-point scale (Strongly agree, Agree, Neither agree nor disagree, Disagree, Strongly disagree). In addition, free-text remarks, suggestions for improvement, etc. could also be provided by participating respondents. Through the results collected, we were able to examine the suitability of the approach, team dynamics (role changing), usage and understanding of particular trans‐ actions; the perceived need for collaboration when making business decisions, overall perception of the business simulation game approach, etc. Students were also asked to provide feedback on whether the questions in the survey were clear and understandable. Half of the respondents strongly agreed with the state‐ ment and the other half agreed with it, thus giving validity to the survey. In addition, a questionnaire was pre-tested and refined by the instructors not involved in its development.

Using a Simulation Game Approach to Introduce ERP Concepts

127

5.2 Results Discussion We were interested in the attitude of participants towards using simulation games in various courses, after experiencing the workshop. We received feedback by asking how much students agree with the following statements: 1. A business simulation game is an appropriate approach to introduce basic function‐ alities of SAP ERP. 2. A business simulation game is appropriate/suitable for learning basic navigation between screens and using controls in the SAP interface. 3. It would be favorable to use a business simulation game approach to introduce ERP concepts in the next academic year as well. 4. It would be favorable, beneficial and advantageous to introduce a similar business simulation game to other courses. The results are visualized in Fig. 5. All the students agreed or strongly agreed that simulation games are suitable in SAP and ERP courses and even agreed that simulation games should be expanded to, and incorporated into, other non-ERP focused courses.

Fig. 5. Usefulness of the business simulation game approach

We also examined whether the expected benefits of applying the simulation game approach have been reached and if the workshop was conducted in the appropriate and best possible way (see the visualization of answers in Fig. 6): 1. The introductory workshop contributes and improves students’ understanding of core business/company management ideas with regard to sales, distribution, marketing and finance. 2. The workshop contributes to the development of the technical skills necessary to use the SAP ERP solution. 3. The workshop demonstrates the need and benefits of different ERP modules inte‐ gration. 4. It is appropriate to organize the workshop into 3 stages on 3 consecutive days with three rounds of a simulation game. 5. After completing the workshop, students better understood the need for collabora‐ tion, communication and coordination between different roles and business functions (in a company).

128

M. Heričko et al.

6. While participating in the workshop, new knowledge and skills about using SAP ERP were obtained. 7. The provided instructions and guidelines were satisfactory, sufficient and appro‐ priate, 8. Determination & willingness to recommend participation in the introductory work‐ shop to other students. 9. The business simulation game based introductory workshop was fun.

Fig. 6. Statements related to the perceived benefits of ERPSim introductory sessions

Students mostly felt that they increased their knowledge and understanding of SAP ERP system and company role management (statements 1, 2, 3, 5, 6). Worth mentioning and emphasizing – a great majority of participants would suggest the workshop to other students (statement 8) and all of them agreed that the workshop was fun, which is a quality that is hardly achievable in most traditional courses. When asked about concrete examples of fair understanding, knowledge and quali‐ fication, we received feedback as seen in Fig. 7.

Fig. 7. Results regarding the understanding of, and skills for using, SAP transactions

A great majority of participants gained (according to their opinion and self-assess‐ ment) at least a fair understanding, knowledge and qualification using each transaction. This means that the workshop still has some potential for improving the knowledge that

Using a Simulation Game Approach to Introduce ERP Concepts

129

the students can receive and acquire. One of the crucial factors behind the imperfect knowledge of participants may had been that team members did not switch roles regu‐ larly, i.e. after each iteration and/or round. This probably led to some specialization between team members, which to a certain degree simulates real world environments. We were also interested in the team aspects of the workshop. We asked participants to rate the following statements and received the results seen in Fig. 8: 1. Our main goal was not to win, but to gain/obtain new knowledge. 2. The team ranking after each round had an impact on, and influenced, the motivation of participants in the next round. 3. All members of our team actively participated in the business simulation game. 4. The main motivation of participants was getting the best result in comparison to other teams. 5. Information that teams exchanged/shared after the completion of different stages and rounds was useful and contributed to a better result in the next stages and rounds. 6. Our team size was appropriate. 7. Communication between the team members was good.

Fig. 8. Results regarding team goals and dynamics

Students generally agreed to almost all of the statements, with the exception of what their goal was, where some felt that it was winning rather than attaining new knowledge (statement 1). This can also be seen also in the fact that some participants believed that information shared by teams was not useful (statement 5).

6

ERP Business Simulation Game from an Instructors’ Perspective

Instructors’ feedback was obtained using a short online survey as well as by carrying out semi-structured interviews with all 4 instructors involved in the ERP introductory workshop. We interviewed them on the suitability, advantages and disadvantages of using the ERP simulation in introductory sessions of an ERP course. They all agreed that the workshop was a suitable way of learning basic functionalities of SAP ERP, navigation, technical knowledge and that the benefits of interconnecting different ERP modules was properly shown. However, the instructors noticed a lack of changing roles

130

M. Heričko et al.

in teams, which is a problem we also outlined in the previous chapter, to help understand why some transactions were lower. Only one of the instructors suggested that the busi‐ ness simulation game approach should be used throughout the complete ERP course. When discussing transactions, instructors felt that their potential was mostly fully utilized, while some of the students did not feel the same. This could have the same cause: the participants were not forced to change roles, which led to the specialization of participants in specific roles. This remained unnoticed according to the instructors’ perspective as the simulation seemed to them to run at a relatively optimal level. When discussing goals, some felt students were divided as some interviewees felt that students were pursuing knowledge, while some thought that their main focus was just competing amongst themselves and winning the game. This is the nature of multi‐ player games in general, there is always some competitiveness present, but we do not find it to be a problem. Even if players compete and do not see attaining knowledge as their primary goal, if they gain knowledge (basically, it is quite evident and there are no doubts that they did) we have succeeded. Individual comments and suggestions that the instructors provided when asked if they had anything to add after completing the interview was that they mostly liked the competition factor, which led to higher motivation; while among the potential chal‐ lenges, it was mentioned that rankings were hard to change after the first round, so team rankings did not change a lot between rounds in an iteration. Consequently, the third iteration was used for experimental purposes by some teams since they had nothing to lose. However, this experimentation resulted in a better understanding of specific busi‐ ness decisions and strategies, thus contributing to the goals of the ERP course.

7

Conclusion

Based on the positive student and instructor feedback, we can conclude that business simulation games are the right approach when introducing ERP systems such as SAP to an ERP course. Students felt that they received sufficient knowledge on SAP and over‐ whelmingly found this experimental approach to be fun. Most of them agreed and were open to the idea of using the business simulation in other courses, where appropriate. As academics, we should use the full potential of simulation games to our advantage in spreading knowledge and view experimental approaches to teaching as potentially equally good as traditional approaches in terms of teaching in classrooms. The introductory workshop was an opportunity for participating students to experi‐ ence an integrated enterprise in action and to notice the benefits of module integration. The simulation game provided a unique experience, where students realized that even a complete enterprise resource planning system needs someone to operate it. Further‐ more, with the same system, different teams/managers produced highly different results. We also found that we do not need to use an ERP simulation for the whole course, as those before us studied, and can still enjoy the positive benefits of introducing a business game simulation in the course curriculum. Moreover, evidence exists that using simulation games as a sole instructional method actually provides less benefit than when they are embedded as merely part of a course [21].

Using a Simulation Game Approach to Introduce ERP Concepts

131

Those wanting to use ERP simulation games must also be prepared to face the chal‐ lenges of striking the right balance between participants’ desire for knowledge and wishes for winning the game, since not all knowledge is passed and shared until the end of the game, because each piece of information offers a certain competitive advantage to the team that possesses it. Often forgotten is also all the “hidden” knowledge and skills that the students gain when engaged in learning through the use of such a simu‐ lation game. Since they work in teams, they also practice teamwork, communication skills, strategic thinking and core concepts of managing a company. Each student can spend time reading literature about ERP systems, attend seminars, but only when they use the system in real or simulated life experiences, can they capture the true purpose of ERP systems. Directions for further research include: how to increase knowledge retention of less understood SAP transactions; is it possible to rank players in such a way so that team rankings only brings positive motivation and lower scoring teams do not get demoti‐ vated; how much knowledge do participants of such an introductory workshop gain in comparison with traditional introduction lessons; the long term advantages of using such games with regard to other competences (ex. strategic thinking, teamwork); would students with a few such courses in each year gain any noticeable advantages in compar‐ ison with students that did not participate in such workshops?

References 1. Esteves, J., Pastor, J., Esteves, J., Pastor, J.: Enterprise resource planning systems research: an annotated bibliography. Commun. AIS 7, 1–51 (2001) 2. Wei, C.-C., Chien, C.-F., Wang, M.-J.J.: An AHP-based approach to ERP system selection. Int. J. Prod. Econ. 96, 47–62 (2005) 3. Sweeney, R.: Millennial Behaviors & Demographics (2006) 4. Skiba, D.J.: The Millennials: have they arrived at your school of nursing? Nurs. Educ. Perspect. 26(6), 370 (2005) 5. Benett, S., Maton, K., Kervin, L.: The ‘digital natives’ debate: a critical review of the evidence. Br. J. Educ. Technol. 39(5), 775–786 (2008) 6. John, G.: The Business Simulations Blog: How to Use Educational Technology - Business Simulation Games (2011). https://www.cesim.com/blog/bid/66888/webinar-for-educatorshow-to-teach-with-business-simulation-games. Accessed 17 Mar 2017 7. Finland, C.: How to Use Educational Technology - Business Simulation Games. https:// www.cesim.com/blog/bid/66888/webinar-for-educators-how-to-teach-with-businesssimulation-games. Accessed 06 Apr 2017 8. Leyh, C.: Teaching ERP systems: results of a survey at research-oriented universities and universities of applied sciences in Germany. J. Inf. Syst. Educ. 23(2), 217–227 (2012) 9. Dirk, H., et al.: Gamification of ERP Systems – Exploring gamification effects on user acceptance constructs Philipp Herzig Susanne Strahringer Michael Ameling gamification of ERP systems – exploring gamification effects on user acceptance constructs. In: Multikonferenz Wirtschaftsinformatik 2012 - Tagungsband der MKWI 2012, pp. 793–804 (2012) 10. Dorn, D.S.: Simulation games: one more tool on the pedagogical shelf. Dorn Source Teach. Sociol. 17(1), 1–18 (1989)

132

M. Heričko et al.

11. Léger, P.-M.: Using a simulation game approach to teach enterprise resource planning concepts. J. Inf. Syst. Educ. Winter 17(4), 441–447 (2006) 12. Pezzini, M., et al.: Gartner Vendor Rating: SAP (2016). https://www.gartner.com/document/ 3488117?ref=solrAll&refval=181593552&qid=0dcb00951006c73a874a4e30d894a392. Accessed 10 Mar 2017 13. Leger, P.-M., Robert, J., Babin, G., Lyle, D., Cronan, P., Charland, P.: ERP simulation game: a distribution game to teach the value of integrated systems. Dev. Bus. Simul. Exp. Learn. 37, 329–334 (2010) 14. Léger, P.-M., Charland, P., Feldstein, H.D., Robert, J., Babin, G., Lyle, D.: Business Simulation Training in Information Technology Education: Guidelines for New Approaches in IT Training. J. Inf. Technol. Educ. 10, 37–51 (2011) 15. Seethamraju, R.: Enhancing student learning of enterprise integration and business process orientation through an ERP business simulation game. J. Inf. Syst. Educ. 22(1), 19–29 (2011) 16. Cronan, T.P., Douglas, D.E., Alnuaimi, O., Schmidt, P.J.: Decision making in an integrated business process context: learning using an ERP simulation game. Decis. Sci. J. Innov. Educ. 9(2), 227–234 (2011) 17. Monk, E.F.: Measuring the Effectiveness of Enterprise Resource Planning Education on Business Process Comprehension. Brunel University, School of Information Systems, Computing and Mathematics (2013) 18. Faria, A.J.: The changing nature of business simulation/gaming research: A brief history. Simul. Gaming 32(1), 97–110 (2001) 19. Faria, A.J., Hutchinson, D., Wellington, W.J., Gold, S.: Scholarship at UWindsor developments in business gaming a review of the past 40 years developments in business gaming over the past 40 years. Simul. Gaming 40(404), 464–487 (2009) 20. Wu, P.H.-J., See, J.T.-B., Heričko, M., Beranič, T.: Knowledge creation activity system for learning ERP concepts through simulation game. In: Knowledge Management in Organizations: 12th International Conference, KMO 2017, Beijing, China, 21–24 August 2015 (2017) 21. Ely, K., Sitzman, T.: A meta-analytic examination of the effectiveness of computer-based simulation games abstract keywords. Pers. Psychol. 64(2), 489–528 (2011)

Knowledge Creation Activity System for Learning ERP Concepts Through Simulation Game Horng-Jyh Paul Wu1 ✉ , See Tiong Beng1, Marjan Heričko2, and Tina Beranič2 (

)

1

2

School of Science and Technology, Singapore University of Social Sciences, 461 Clementi Road, Singapore 599491, Singapore [email protected] Faculty of Electrical Engineering and Computer Science, University of Maribor, Smetanova ulica 17, 2000 Maribor, Slovenia

Abstract. Learning ERP concepts is challenging as ERP systems are complex. Gamification promises to be an effective method to enhance students’ motivation to learn. However, how to effectively integrate simulation game into ERP concept learning is not straightforward. We propose a theory-driven methodology by formulating a ERP learning system based on simulation game. We argue that Activity Theory, SECI and CMC knowledge creation processes can be consis‐ tently unified for an expressive theoretical model that clearly defines the unit of analysis and demonstrates the linkage between the expansive, social and historical processes that give rise to a snapshot of a system at a particular moment. This paper proceeds to formulate a ERP learning activity system through simulation game. With the clearly defined unit of analysis, a set of measures can be constructed to measure the effectiveness of the activity system. A pilot study involving graduating students enrolled in an ERP course is conducted. The preliminary result suggests positive effects of the simulation game on the learning of ERP concepts. Lastly, this paper expands on the preliminary results and antic‐ ipate the application of this theory-driven methodology to formulate comprehen‐ sive learning activity systems and measures for effectiveness of learning ERP concepts. Further empirical studies are required to implement and assess the effectiveness of these activities systems such anticipated. Keywords: ERP · Simulation game · Activity theory · SECI · Cascading modes of communication

1

Introduction and Background Research

Enterprise Resource Planning (or ERP) is an important domain of knowledge in the modern society [1]. However, learning ERP concepts is challenging as ERP software is a very complex system involving many components that cover the entire spectrum of business functions in an organization. This paper outlines a theory-driven research methodology to study the learning of ERP-related concepts through a simulation game. The methodology starts with modelling the learning activities as an Activity System based on Activity Theory [2]. It is further analysed by knowledge creation process theo‐ ries of SECI [3] and Cascading Modes of Communication (CMC) [4, 5]. In our previous © Springer International Publishing AG 2017 L. Uden et al. (Eds.): KMO 2017, CCIS 731, pp. 133–143, 2017. DOI: 10.1007/978-3-319-62698-7_12

134

H.-J.P. Wu et al.

research, we have argued that these theories presuppose knowledge formation process is intertwined with community formation process. We further argued that they all posit dialectic tension exists along the epistemological, ontological and communicative dimensions of an activity system [2, 6]. The community is to be sustained only by resolving the tension through the effort to overcome boundaries and barriers of an activity system, called the boundary-crossing [7, 8]. One of the critical tasks of the activity system is thus to engender and sustain such dialectic tension during the entire learning process. In this paper, we proposed that playing a simulation game is an effec‐ tive way to sustain a learning activity system with sufficient dialectic tension and moti‐ vation to overcome them. This view is intuitively consistent with the recent gamification movement in education. However, in this paper we provide a systematic explanation as why this is the case, how to implement it and measure its effectiveness. More precisely, in the following sections, our focus will be on the following three points: (1) How to apply Activity Theory consistently with SECI and CMC process analysis (2) How to formulate ERP concept learning as an activity system (3) How to measure the effective of the Activity System such formulated.

2

Apply SECI/CMC Processes Analysis to Activity Theory (AT)

Activity System is a structural model, it can be evolved through an expansive learning process [8]. In [5], we have explained how this expansive learning cycle is similar to a CMC/SECI process: “The cycle consist of several learning acts: questioning present work practices (point 1 in Fig. 1) – in CMC terms: expressing authentic but problematic feeling or belief resulting from problematical internalization of accepted truth, analyzing historically the causes that have created problems in daily work (points 2a and b) – in CMC terms: clarifying and arguing a point by giving justification to doubts earlier expressed in socialization, modelling and examining for a new form of activity (points 3 and 4) – in CMC terms: combing all argumentation to arrive at an acceptable common ground to be committed together, testing and changing the activity and practices during

Fig. 1. Expansive cycle of learning process

Knowledge Creation Activity System for Learning ERP Concepts

135

the experimental phase and finally reflecting on the process and implementing and generalizing the final concept of the activity (points 5, 6 and 7) – in CMC terms, inter‐ nalizing the common ground committed through embodying the common ground in their actions.” Thus, process-wise, expansive cycle of learning process can be readily mapped into CMC and SECI process. However, for CMC/SECI analysis to applied to AT, we need to map CMC/SECI directly to the unit of analysis itself. In another word, how does the expansive learning/CMC/SECI process arise directly from the structure of the activity system? To answer this, we need to map CMC and SECI to complex dyadic structure of activity system. As shown in Fig. 2, which is adapted from Fig. 2.6 of [2], there exists the different modes of the activity system: Consumption, Exchange, Distribution, and Production, in an activity system. These dyad-structured modes manifest two important premises of activity theory as explained in [9]: mediation and intentionality/consciousness. Media‐ tion necessitates the dyad-structure among an intentional subject (or group), a tool and an object. Intentionality/consciousness is expressed by the intentional subjects through the modes and transmitted to all involving participants. Therefore, an expansive learning process, as demonstrated in Fig. 2 is but a well formed (or fruitful) traversal through the modes in a CMC/SECI process.

Fig. 2. The Structure of Activity in light of SECI Processes

According to Engestrom on activity system [2]: “This third lineage, from Vygotsky to Leont’ev, gives birth to the concept of activity based on material production, mediated by technical and psychological tool as well as by other human beings.” With this, we need to realize that AT foregrounded the importance of object and mediation of a learning process. In most SECI/CMC analysis, the objects were variously assumed to be a product (such as a kneading machine or an automobile) or a concept (such as Lead‐ ership) that is under development; but they were not properly defined and studied. The concept of mediation through technical, psychological means as well as through community of human beings in activity is also important. Thus by integrating AT with SECI and CMC, the activity system with its modes, mediation, tools and intentionality are mapped across all three theories as demonstrated in Table 1.

136

H.-J.P. Wu et al. Table 1. Unified model of AT, SECI and CMC

Modes

Consumption Exchange Distribution Production

3

Triad structure Subject Mediation/ Tool Subject Community Subject Rules Community Division of Labour Subject Tools

Object

Process (SECI) Intentionality (CMC)

Object Community Object

Socializaiton Authenticity Externalization Rationality Combination Normality

Object

Internalization

Reality

ERP Game as an CMC/SECI Activity System

Following1 Table 1, the ERP simulation game can be formulated as an activity system with the object of winning a game where there are individual members (subject) and the team (community) involving in an expansive learning cycles as shown in Table 2. In the Production mode, when the community formation has yet to happen, individual student was given essential knowledge of the tools, with mostly their background knowledge of the tool (the simulation game based on ERP application), starts to play the game and trying to win it. They were feedback by the environment with the reality whether they succeed or not. This is followed by the Consumption mode, where the individual is joined with other team members so that they can share with one another their authentic expe‐ riences of playing the game. By bringing their own background and prior understanding of ERP to bear in their sharing, they can even get to understand one another better. So the initial round is served as a bonding activity as well to engender the community formation process. In the Exchange mode, the individual started to engage each other through devising additional rules (game strategies) that they can use to win the game through a rational debate process. In the Distribution mode, the roles of each individuals will be determined so they can play the game as a team, each player is supposed to conform to the norm (rules) and responsivities (division of labour) that was agreed upon. This brings the process into a new cycle when a second round of the game is played, and the team enters a new Production mode where individual member exercise what each of them has learnt in the previous cycle about the simulation game, team members, strategy, and team work, in an attempt to win the game. Through the individuals’ efforts in forming the community and learning the ERP concepts, the expansive cycle continues to refine the mediation and intentionality of the system. With the above explanation, it follows that a typical design of an ERP game session has the following components [11]. 1. 45 min: Introduction to the Game. The instructor will go through the relevant modules of the game and using the instructional materials so that participants will know how to play the game.

1

We adopt ERPSim game developed by Prof. Leger’s group at HEC Montreal and is developed on top of SAP ERP system [10].

Knowledge Creation Activity System for Learning ERP Concepts

137

Table 2. Simulation game learning as an activity system Modes

Production Consumption Exchange

Distribution

Triad structure Subject Mediation/ Tool Student ERP Application Student Team (and Class) Student Business Context and Strategies Team (and Collaborative Class) Tasks

Object Winning the Game Winning the Game Team (and Class) Winning the Game

Process (SECI) Intentionality (CMC) Internalization

Reality

Socializaiton

Authenticity

Externalization Rationality

Combination

Normality

2. 3. 4. 5. 6.

15 min: break 75 min: Participant will play the 1st round of the game. 45 min: Review of the 1st round of the game. 60 min: break 45 min: Each group (6 of them) take turns sharing their key concepts learnt and decide on a common summary of lessons learnt. 7. 15 min: break 8. 75 min: The participants will pay the 2nd round of the game. 9. 45 min: Review of the 2nd round of the game.

Notice that, the above steps did not specify the teaching of ERP concepts explicitly. On the contrary, it emphasizes how the subjects (namely, the students) may engage themselves in an activity system through a learning process. Instructors serve mainly as facilitators during the process. One common questions in such student-centric approach has been how can we ensure the students learn what they are supposed to learn without the instructors specifically teach on the knowledge to be acquired? This is the question to be answered in Sect. 4 on how effective is the learning activity system to allow students acquire the complex ERP concepts. 3.1 Sustaining Student-Centric Learning Activity Beyond ERP Game Activity System Before proceeding to Sect. 4, it is noted that the simulation game needs to be integrated with traditional classroom teaching and learning activities for the entire duration of a course. Thus, concurrent to classroom-related activities, an activity system may be provided to sustain the student-centric learning activities started in the game simulation activities. Table 3 shows a high level structure of the Cascading Modes of Communi‐ cation Social Media Activity System.

138

H.-J.P. Wu et al.

Table 3. Blended cascading modes of communication (CMC) social media activity system CMC social media Blogs

Discussion forums

Wikis discussion forums

Simulation game or operating ERP applications in an organization

Blended online and offline activities Online Activities -Pre-course selfintroduction on Group Blogs Offline Activities -Chats during Class Breaks Online Activities -Debate on issues on Discussion Forum Offline Activities - Participation in discussions during Breakout sessions Online Activities -Wiki Group reports Offline Activities -Moderation Duty to facilitate discussion and produce group report -Presentation to External Panellists

SECI stage Socialisation (tacit to tacit)

Externalisation (tacit to explicit)

Combination (explicit to explicit)

Online Activities Internalisation -Reflection Reports (explicit to tacit) -Story for Preparing for Panel presentation Offline Activities -Playing the Simulation Game by embodying the knowledge acquired about ERP concepts

Traditional interpretations Knowledge conversion begins with the tacit acquisition of tacit knowledge by people who do not have it from people who do. Involves converting tacit into explicit knowledge, and holds the key to knowledge creation as new concepts are formed. A process of “systematizing concepts into a knowledge system”, which happens when people synthesize different sources of explicit knowledge into, for example, a report. Described as “a process of embodying explicit knowledge into tacit knowledge”. It is “closely related” to “the traditional notion of learning”, and to “learning by doing” although somewhat confusingly they also say that internalization is ‘triggered’ by learning-by-doing.

At the end, there would be 3 activity systems that would be interacting with one another in the final design of a course aimed at teaching and learning the ERP concepts. These are: 1. ERP Simulation Game Learning Activity System (Student-centric) 2. Classroom Seminar Activity System (Instructor-centric)

Knowledge Creation Activity System for Learning ERP Concepts

139

3. CMC Social Media Activity System (Student-centric) Note that in each of the above systems, both instructor and student will be involved. So neither student-centric nor instructor-centric excludes the participation of the other party. It is also noted all activity systems are blended as they have online as well offline activities. With respect to the emerging ideas of flipped-classroom methodology where students self-study instructional materials before coming to a classroom for discussion, our approach is similar in the sense that students’ learning activity is being emphasized; however it is also different that the learning activities in the flipped class‐ room approach focus on absorbing the instructional materials, while in the simulation game approach, the focus is on the interaction with the environment itself. Certainly, the final design can include both instructional materials and the simulated environment to engender students’ learning experiences, which form the basis of a student-centric pedagogical approach. Lastly, these 3 systems can be similarly designed, analysed, and measured in further details as demonstrated in this paper, which will be included in our future research.

4

Measure the Effects of ERP Game Learning Activity System

As mentioned in Sect. 3, a measurement of the effects of learning is called for specifically for a more student-centric learning activity. According to [12, 13], a set of objective measures of cognitive learning in the ERP games is specified that assesses the effec‐ tiveness of the learning process. It consisting of the following components: 1. the ERP application (command based, tool procedural, and tool conceptual) 2. the business context in which the ERP application is used (business procedural and business motivational) and 3. the collaborative tasks enabled by the ERP application (task interdependencies and collaborative problem-solving approach) With the theoretical underpinning of Activity Theory, it is not surprising how these measures have been arrived, as it is implied when the mapping of the ERP game in an activity system is completed as shown in Table 2. Namely, ERP application is the mediation in the Production triad; the business context is the rules (business strategies) in the Exchange triad; and the collaborative tasks are the division of labour in the Distribution triad. In sum, the above measures are essentially those of how well the subject and team (community) can utilize the corre‐ sponding tools in each of the modes of the activity system. With the application of these tools, students acquire necessary aspects of ERP knowledge from the feedback and challenges from the environment. Last but not the least, in [12], the subjective measure of the cognitive learning is also discussed. However, the focus of the measure remains on the various tools to play the simulation game. Nothing pertaining to the community where the students are learning was studied. As introduced in Sect. 1, nonetheless, all 3 underling theories of our research, AT, CMC and SECI are essentially social learning theories. The importance of the social dimension cannot be overlooked. How would this aspect of the learning be

140

H.-J.P. Wu et al.

measured then? It is realized that the above three aspects of ERP knowledge pertain to only 3 of the 4 modes in the ERP game activity system. The remaining one mode is that of Consumption (or Socialization) where subjects leverage on one another (the Community) to share their experiences (and consumption) of the object - playing the game (and the outcome of the game). In [4], we have proposed a Structural Equation Model (SEM) model, which is adapted from that in [14], that can be used to measure the effectiveness of the Community Formation achieved by the players in forming a learning community. The measures mainly focus on the attitude, personality and cogni‐ tive attributes of the participants (the Social Capital aspect) and their efficacy of their actions resulting from interacting with the community (the Social Cognitive aspect). 4.1 Measure the Effects of ERP Game Learning Activity System We conducted a pilot study based on the objective measures mentioned in Sect. 4. The subjects of concern are post-graduate students who are enrolled in a course on Enterprise Resource Planning. A sequence of 3 games, the first one serving as familiarization, is played as the pre-course exercise to motivate the students to learn about the subject matter of this course. A quiz consisting of 20 questions is administered before and after the ERP game sessions. There are 16 and 11 respondents to the quiz before and after the games, respectively. As comparison of the performance before and after is required, our sample size has to be the small one, which is 11. Furthermore, as explained in Sect. 4, the quiz concerning the objective assessment of the EPR knowledge is categorized at 3 distinct levels of knowledge, code-named as TS, BP and ES, respectively as follows: 1. TS: the ERP application - 2 questions 2. BP: the collaborative tasks – 12 questions 3. ES: the business context in which the ERP application is used – 6 questions There are 2 questions in the TS category, 12 in BP and 6 in ES. The results of the quiz are summarized and visualized in Fig. 3 below. As shown in Fig. 3, the overall performance level of the ERP knowledge improved from 55.5% to 59.5% before and after playing the ERP games 2 times. The same positive effect is seen for the BP and ES aspects of the ERP knowledge, where BP shows a slightly better results (7.7% improvement) than ES (7.3% improvement). What is surprising is there is a significant worse performance level of the TS knowledge category (27.3% deterioration) before and after the game. Given that ES covers the strategic aspects of the ERP knowledge, BP the integration of business process and TS the application itself, it seems particularly positive that playing the game has improved the 2 higher level of knowledge of ERP. On the other hand, it is rather surprising that the effectiveness of the lowest level of knowledge of ERP deteriorated after more games are played. However, upon further reflection, the results may not be so unexpected and can be interpreted as a shift of focus from lower level knowledge to the higher level one while students getting more familiar with the ERP application. Since TS concerns students’ familiarity of the ERP application system, this knowledge once mastered through memory may fade into the background when the students pay less attention to while focusing on ES and BP. As a result, when they are

Knowledge Creation Activity System for Learning ERP Concepts

141

Objective Assessments of ERP Knowledge 80.0% 70.0%

67.3%

68.2%

59.5%

59.4%

60.0% 60.0%

50.0%

55.5%

51.7%

40.0% 40.9%

Before

30.0%

Afer

20.0% 10.0% 0.0% TS

BP

ES

Overall

Before

68.2%

51.7%

60.0%

55.5%

Afer

40.9%

59.4%

67.3%

59.5%

Fig. 3. Objective assessment of ERP knowledge before and after the Game

quizzed on TS, they tend to make mistakes as they may not consciously try to recall the exact names of the transaction codes as well as be too “lazy” to reference the job aids that was provided to them. Another probable explanation is that the size of the questions 2 is too small to sensitivity to noise. To increase the reliability, more questions in this category are needed. However, the above only describes a pilot study with some preliminary observations and explanations of certain surprising results. The questions of the quiz will need to be further refined and the distribution of the number, re-balanced. Nonetheless, the initial indication suggests that the game indeed help to improve the overall level of knowledge of the students.

5

Conclusion and Future Work

Learning ERP concepts is a challenging task for students as ERP is a complex system. Gamification seems to be a promising way to facilitate a student-centric approach where students are proactively learning the ERP concepts collaboratively. In order to ensure the effectiveness of a particular gamification design, a rigorous methodology needs to be formulated to analyse and examine the design. This paper proposes a theory-driven formulation of a knowledge creation activity system for learning ERP concepts through simulation games. It is explained and validated through the following steps: (1) How to apply Activity Theory consistently with SECI and CMC process analysis (2) How to formulate ERP concept learning through simulation game as an activity system (3) How to measure the effective of the Activity System such formulated. Through Step (1), it is shown that all three theories: AT, CMC and SECI more expressive theoretical model that clearly defines the unit of analysis and demonstrates the linkage between the expansive, social and historical processes that give rise to the

142

H.-J.P. Wu et al.

structure at a particular moment. Through Step (2), it is shown ERP simulation game learning activity system is a system that starts at Production mode and ends at the same also. As a result of Step (3), a set of questions are formulated to measure the effectiveness of the simulation game learning activity system. The assessment covers 3 distinct level of the ERP knowledge TS, BP and ES. A pilot study was conducted based on the ques‐ tions, and the results indicates positive effect that simulation game has on learning the ERP concepts. In particular, the effects are positive at the 2 higher level knowledge (ES and BP). The result of the lowest level knowledge (TS) is not straightforward. However, some explanation is attempted that calls for further study. Nonetheless, the measures formulated above does not cover an important mode of the activity system, namely the Consumption (or Socialization). In the future, the measure should be included in the empirical study of the activity system, extended from that was reported in [14]. We further believe the methodology demonstrated in this paper can be adopted to formulate a 3 tiered-activity systems that will encompass all learning activities more comprehensively: 1. ERP Simulation Game Learning Activity System (Student-centric) 2. Classroom Seminar Activity System (Instructor-centric) 3. CMC Social Media Activity System (Student-centric) In Sect. 3.1, a preliminary formulation is described for activity system 3. The formu‐ lation needs to be completed and activity system 2 will also need to be formulated in the future to study all relevant learning activities.

References 1. Robert Jacobs, F., Ted Weston, F.C.: Enterprise resource planning (ERP)-a brief history. J. Oper. Manage. 25(2), 357–363 (2007) 2. Engeström, Y.: Learning by Expanding: An Activity-Theoretical Approach to Developmental Research. Orienta-Konsultit, Helsinki (1987) 3. Nonaka, I., Takeuchi, H.: The Knowledge-Creating Company: How Japanese Companies Create the Dynamics of Innovation. Oxford University Press, New York (1995) 4. Wu, H.-J.P.: Knowledge and community formation via cascading modes of communication with a case study and research design. In: The 8th International Conference of Knowledge Management in Organizations, pp. 189–204 (2013) 5. Wu, P.H.-J., Uden, L.: Knowledge creation process as communication – connecting SECI and activity theory via cascading modes of communication. In: Uden, L., Fuenzaliza Oshee, D., Ting, I.-H., Liberona, D. (eds.) KMO 2014. LNBIP, vol. 185, pp. 403–412. Springer, Cham (2014). doi:10.1007/978-3-319-08618-7_39 6. Wu, H.-J.P.: Unifying knowledge creation process through cascading modes of communication. In: Uden, L., Heričko, M., Ting, I.-H. (eds.) KMO 2015. LNBIP, vol. 224, pp. 15–25. Springer, Cham (2015). doi:10.1007/978-3-319-21009-4_2 7. Engeström, R.: Voice as communicative action. Mind Culture Activ. 2(3), 192–215 (1995) 8. Engeström, Y.: Expansive learning at work: toward an activity theoretical reconceptualization. J. Educ. Work 14(1), 133–156 (2001)

Knowledge Creation Activity System for Learning ERP Concepts

143

9. Nardi, B.A.: Activity theory and human-computer interaction. In: Nardi, B.A. (ed.) Context and Consciousness: Activity Theory and Human-Computer Interaction. The MIT, Cambridge (1996) 10. Léger, P.M.: Using a simulation game approach to teach enterprise resource planning concepts. J. Inf. Syst. Educ. 17(4), 441 (2006) 11. Wu, H.-J.P.: Learning enterprise resource planning (ERP) through business simulation game. In: Proceedings of the 11th International Knowledge Management in Organizations Conference on The Changing Face of Knowledge Management Impacting Society, pp. 5–6. ACM, July 2016 12. Cronan, T.P., Léger, P.M., Robert, J., Babin, G., Charland, P.: Comparing objective measures and perceptions of cognitive learning in an ERP simulation game: a research note. Simul. Gaming 43(4), 461–480 (2012) 13. Charland, P., Léger, P.M., Cronan, T.P., Robert, J.: Developing and assessing ERP competencies: basic and complex knowledge. J. Comput. Inf. Syst. 56(1), 31–39 (2016) 14. Chiu, C.M., Hsu, M.H., Wang, E.T.: Understanding knowledge sharing in virtual communities: an integration of social capital and social cognitive theories. Decis. Support Syst. 42(3), 1872 (2006)

Entrepreneurship Knowledge Transfer Through a Serious Games Platform The Venture Creation Game Case Dario Liberona1 ✉ and Cristian Rojas2 (

1

)

Department of Business Administration, Universidad Tecnica Federico Santa Maria, Valparaíso, Chile [email protected] 2 Department of Industrial Engineering, Universidad Tecnica Federico Santa Maria, Valparaíso, Chile [email protected]

Abstract. One of the challenges for productivity is how to improve the experi‐ ential and impact of learning in education, The research aims to evaluate the creation of a serious game using Nonakas SECI model to transfer experiential knowledge related to entrepreneurial practices to an experiential simulation game, the game was name “venture creation game”, the application and use of the serious game was evaluated in higher education students. The empirical results indicates that games have a very positive impact in the learning process off students and is a very efficient way to enhance the teaching of entrepreneurship in Higher education. Also trough the SECI model, aspects of entrepreneurship theory were developed. Keywords: Edutainment · Entrepreneurship · Serious games · Game learning · Gaming base learning · Startup education · Venture creation game

1

Introduction

We have been talking about the importance of Knowledge Management since at least 1959, when Peter Drucker in his book “Landmarks of Tomorrow” first described the rise of the “knowledge worker” (Drucker 1959). Forty years later Drucker declared that increasing the productivity of knowledge workers was “the most important contribution management needs to make in the 21st century.” (Drucker 1999). Since education is one of the key elements to productivity of the Knowledge Worker, one of the challenges is how to improve the experiential and impact of learning in this “knowledge Workers”. Knowledge and education are the pillars of competitiveness in the framework of international competition (World Economic forum 2012). Therefore and important part of the role of education in universities is related to help students to become better knowledge workers. How can we innovate in teaching methods so we can achieve this fundamental goal?, this paper is related to an experience of using serious games to teach entrepreneurial concepts, in particular those related to developing © Springer International Publishing AG 2017 L. Uden et al. (Eds.): KMO 2017, CCIS 731, pp. 144–156, 2017. DOI: 10.1007/978-3-319-62698-7_13

Entrepreneurship Knowledge Transfer Through a Serious Games

145

startups in emerging economies, In order to give a more experiential aspect to the acquisition of knowledge by students. The research group develop a serious game denominated “venture creation game” in order to review the process of developing a learning game and its effect in the learning experience of higher education students.

2

About Serious Games

The 21st century is characterized by the digital era, the evolution of technology leads to constant change, progression and development. People talking about Internet of things (IOT), Big Data analysis, mobile applications and entertainment, the world is changing and so is the way people are learning. The problem emerges from the educational systems and teaching technics that are basically the same in the last 20 years or more, there is a huge disconnection between technology and education and it’s only getting wider (Brian Solis 2014). Learning technologies (LT) are an opportunity to share and learn knowledge and abilities on different fields around the world. When you hear about learning tech‐ nologies the first think that comes into your mind is online courses such as E-Learning platforms or video tutorials, never the less there is an LT ecosystem where you can find alternatives to learning methods such as serious games. (Ulicsak 2010). “Serious games are games designed in which education (in its various forms) is the primary goal, rather than entertainment.”(Gross 2016) There are multiple appliances for serious games, on the fields of education, military, politics, and healthcare or corporate, all for purposes of training, skill development, and learning or complementary tools, this doesn’t mean that serious games are not enter‐ taining or fun, it’s just not the main objective. (Sloetjes 2014) The field of edutainment, serious games or learning game design fields, have made big improvements in the past decade, but there is still a long way to go before we can explode the full potential of educational games. In this area there are some recurrent questions that are rather reiterative such as, If learning games are beneficial why aren’t they integrated more in classrooms? What learning is really happening and achieve in games? How can learning games be designed or implemented to have even deeper and more meaningful impact on the learner? Moving to the next level will require much more thoughtful and strategic mapping of the learning experiences and outputs in games. It is clear that games can improve the learning experience, but there is much learning to be done from the methodology and for the relevance of content implemented, in the case of business learning there has been a long tradition of market simulation tools, financial models simulators, and digital exercises like entrepreneurial games, but we are at the verge of new experiences. The millennials and most of the workers and executives under forty five years old have had great exposition to digital games, more than ever.

146

3

D. Liberona and C. Rojas

Educational Games (Edutainment)

Educational games, can be traced back to the 1970’s where “non digital” games where used in schools for math-related or social science understanding. The digital application comes on the 1990’s with the first multimedia computers, evolving the term to edutain‐ ment, but because of the poor quality and lack of understanding the interest decreased. Never the less, serious games gain attention again during the 21th century with a situa‐ tional and constructionist approach in games. Research has shown a positive effect of games as educational tools in various skills such as: strategic thinking, planning, communication, collaboration, group decision making, and negotiating skills (Kirriemuir and McFarlane 2004; Squire and Jenkins 2003). Edutainment, comes from “education” & “entertainment” it is designed to generate motivation, interest and a better understanding throw technology using games, music, internet or television to help both students and teachers in the process of learning. The market of business simulation for education are mainly centered on marketing, finance, strategy and optimization. The market for entrepreneurial games is very limited, our related findings are shown in the next table: Serious games did not come into wide use until the 1990s with multimedia PCs, even though such games were created and used long before. At the time, educational games and other software evolved into “edutainment”. However, interest in edutainment soon decreased, partly because of the (poor) quality of the games themselves, and partly because of a growing interest in the Internet (Michael and Chen 2006). The problems encountered in edutainment are reflected in phrases such as “edutainment, an awkward combination of educational software lightly sprinkled with game like interfaces and cute dialog” (Zyda 2005), or “most existing edutainment products combine the entertainment value of a bad lecture with the educational value of a bad game” (Squire and Jenkins 2003). With the general renewed interest in serious games, game developers have moved from “skilland-drill interactive learning paradigms towards situational and construc‐ tionist approaches” (ELSPA 2006). Games in education is gaining acceptance, but their use is not widespread, and it is a controversial issue (ELSPA 2006; Michael and Chen 2006). Educational games is also faced with the challenge of providing research evidence of the acclaimed benefits, which currently is “complex and thinly spread”, possibly because the study of games and gaming relates to several different disciplines; “as a result of the diversity and complexity of games themselves, and the range of perspectives taken by researchers, there are few hard and fast findings in the literature” (Kirriemuir and McFarlane 2004, p.2). Despite the “few hard and fast findings”, research is showing positive effects of games as educational tools. Games can support development of a number of various skills: strategic thinking, planning, communication, collaboration, group decision making, and negotiating skills (Kirriemuir and McFarlane 2004; Squire and Jenkins 2003; see also Gee, unpublished manuscript). However, “hard facts and evidence” is for future research to provide. There is also a number of concerns to consider in order to realize the full potential of games as educational tools: resources (many schools have

Entrepreneurship Knowledge Transfer Through a Serious Games

147

computers that are too old for new games, technical support, time for teachers to famil‐ iarize themselves with the game, etc.), how to identify the relevance of a game to stat‐ utory curricula, difficulty in persuading school stakeholders to the potential benefits of computer games, etc. (ELSPA 2006).

4

The Venture Creation Game

During 2016, the Federico Santa Maria University developed a project devoted to analyze the Latin-American entrepreneur’s practices for improving the chances of surviving the Death Valley of Startups and improving the success rate. The experience of the project was to learn about entrepreneurial practices that contributed to the success of Startups (Fig. 1).

Fig. 1. The SECI model (Nonaka and Takeuchi)

Basically a SECI model approach was applied, (Nonaka 1994; Xu 2013), consisting in processes of socialization, externalization, combination, and internalization. The first one, socialization, is about an informal sharing of experience (e.g. between master and apprentice), for this interviews were conducted personal interviews with around 60 successful technological and innovative entrepreneurs in Chile. The second one, exter‐ nalization, is about the formalization of tacit knowledge, the interviews helped to iden‐ tify and document a series of practices and the general process of developing and startup journey theory model. The third one, combination, is about the construction of explicit knowledge from tacit knowledge, a series of practices were identify, and a series of surveys with the participation of more than a thousand founders of startups in Chile permitted to validate and weighted the practices. The fourth one, internalization, is the

148

D. Liberona and C. Rojas

transformation of explicit knowledge into tacit knowledge through appropriation and akin to learning by doing, this is where the “Venture Creation Game” was developed. There was also a review of existing related games in the world. (Table 1). Table 1. Serious games related to entrepreneurship (own) Name Entrepreneur simulation

Company GoVenture, MediaSpark Inc.

Country Canada

Terra NovUP

Terra NovUp

Chile

The start up game

Wharton Business School

EE.UU.

CleanStart

Learning edge, MIT

EE.UU.

Venture blocks

Venture blocks

EE.UU.

Description Online Simulation based on the operational phase. Include decisions on: marketing product mix finance team construction Board game based on the construction of the business model throw the theory of Alex Osterwalter. Roleplay: investor employee founder focused on: development management raise capital Wage compensation vs equity Online simulation. Based on the launching and operation of the business includes decisions on: product price team work raise capital Online simulation. Based on finding a need and creating a business idea.

With the analysis of gaming experts and a group of developers, designers the contents of the game were review, this acknowledging that a single game will not be able to hold of the concepts and practices of a startup journey. The conclusion was that there was a minimum need of at least three games in order to have a serious game related to entre‐ preneurship, one for the idea, business model or for the “preparation stage” (Fig. 2), another for the venture start and initial running, and one of the operation (breakeven point) and scaling or exiting the venture. Studding the so call “Startup journey” with different experts, there are four phases that can be identified as common on most of the Startup, they are: (1) Preparation for the Business: Identification of the problem/opportunity: Motivation of the founders, concept of an idea, problem that needs to be solved. This is the stage of preparation and the general practices identify are related to anticipate, market projections, competitors reaction, try to identify the opportunity. (2) Starting The Business: Is the stage where the Business model is define, and the company starts to operate, and the general practices identified are related to activate a serious of processes like recruiting, networking, getting financial and other resources. (3) Running the Business through the Death Valley: Launching of the business, starting operations and passing through the surviving period where good management and orchestration are fundamental to survive and achieve the Break-even point.

Entrepreneurship Knowledge Transfer Through a Serious Games

149

Fig. 2. The startup journey, stages. (own elaboration)

(4) Scaling or Exiting the Business: Once the company is stabilized you must look at ways to continue the journey, how to grow bigger or sell it to the market (profita‐ bility). It is essential to adapt, sometimes pivot and improve processes and customer satisfaction. The Venture Creation game is based on the second phase (business model), this is because it is one of the critical moments of a Startup and there is a huge gap of knowledge and understanding of the practices or decisions entrepreneurs have to make to create a solid base for the implementation and correct identification of the problem/opportunity. The Game answers to this three basic questions: (1) What is the Value proposition? Answer: Correct Segmentation for a customized product, clear definition of the problem, and a proposal of the characteristics of the product. (2) How Profitable is the business? Answer: Market size, and cash flow understanding (3) Is the correct team for the business? (high performance team) Answer: Identify the right teammates, their abilities needed to increase the proba‐ bilities of success in the business. Based on our research there is no game specifically dedicated to the understanding the startup business process. This serious game is an online simulation game which doesn’t need multiple players to play, is purely focused on the modeling of the business plan and it will incorporate some operational aspects of the business idea, ending with the round to raise of capital that involves a simulated panel of investors (Fig. 3).

150

D. Liberona and C. Rojas

Fig. 3. Venture creation game market place.

The venture creation game is set on a City landscape, where you are the leader of a Startup. The primary rules have been settled. First the Startup is about creating an App related to sports activity, specifically about creating a healthy gym routine for a deter‐ mined type of client. The Entrepreneur has to walk through the city taking decisions about his startup. Going to the University, gather information at home, recruit the team at the event center, the players (students) should find out more about the segments at the

Fig. 4. Venture creation game – home practices option example.

Entrepreneurship Knowledge Transfer Through a Serious Games

151

gym or park. Strengthen the team at the cowork building and create the company at the lawyer office are part of the activities (Fig. 4). Some of the decisions taking place at the simulation, are based on the four categories at the bottom (Team, network, segment, competitors and cash flow). Here we are at the house, where the first decisions take place, knowing about the segments, webpage and gathering information (Fig. 5).

Fig. 5. Venture creation game customer segments characteristics.

4.1 Clients Profiles There are four types of client profiles among which the player has to choose which segment will approach his Startup App (Fig. 6).

Fig. 6. Venture creation game, competitive map.

152

D. Liberona and C. Rojas

In this section, the entrepreneur can see where the competitors are positioned, and what are the better opportunities of making a profitable startup over the years. At the end of the game the entrepreneur has to decide the qualities of the App that are based on Social and Complexity attributes (Fig. 7).

Fig. 7. Venture creation final round of investment.

At the end of the game, there is an evaluation of the decisions taken during the game. And evaluation of the attributes selected and the information gathered in all five cate‐ gories. There are three investors that are willing to pay 500.000 USD each and give a resourceful feedback. The process of creating the game is very complex.

5

Results

The game was use in four different classes, two engineering classes (fifth and sixth year) in two different universities, and international exchange students Business class (from 8 different European countries) and one MBA program class in entrepreneurship. The total number of participant students was 86 and the ones that participated in the survey were 77. The classes were pilot classes and the game was used has part of an entrepreneurial module in Management and Strategic Management classes. A survey was conducted among student, with the following results (Tables 2, 3, 4 and 5):

Entrepreneurship Knowledge Transfer Through a Serious Games

153

Table 2. Level of knowledge and experience related to entrepreneurship Exposure to entrepreneurship Little knowledge Some studies I have collaborated in startups I have studied extensively I have been a founder member

16,9% 51,9% 19,5% 9,1% 2,6% 100,0%

Table 3. Survey answers to general evaluation of serious games What is the contribution of the game to learning Very low Low Medium High Very high

0,0% 1,6% 28,6% 36,5% 33,3%

Table 4. Evaluation of the venture creation game How would you evaluate the venture creation game for learning Bad Regular Good Very good Excellent

0,0% 15,9% 14,3% 44,4% 25,4%

Table 5. How was the level of contribution to learning with the serious games Did the game contibuted to learning about entrepreneurship No contribution at all Learned some new concepts It made learning easier and fun A much better understanding of the concepts Really helped to clarify and learn entrepreneurship concepts

0% 33% 24% 29% 14%

If educating the player should be the primary goal of serious games like Michael and Chen (2006) proposed, the results on the learning process was achieved, the students got better results than just preparing their classes and learned and discusses about reasons and concepts of failure and learning. Class attendance also increased in classes that involved the game simulation (Fig. 8).

154

D. Liberona and C. Rojas

Fig. 8. Class experience of the game

6

Conclusions

Now a days, it has become difficult to gain students attention and collaboration, there has been several learning technologies for education that are being deployed at class‐ rooms, such as the use of interactive education, the use of online and e-learning plat‐ forms, the cellular use of technologies for labs, the use of sophisticated simulators (Enterprise games), active classrooms, and the use of gaming has a way to attract students. Certainly there has been a change in the culture of young students, all millennials are used to play games, some of them role playing, this kind of learning technologies, have a lot of applications in terms of Business and modeling studies, the University has embrace the use of the game in different educational games, being very appropriate for professional graduate students. The students give a lot of credit to the venture creation game, they were eager to play it a few times, and it was an easier way to understand key entrepreneurial concepts, in a rapid and very applicable fashion. From the universe of surveyed students 52% agree they have taken entrepreneurship courses before and 23% have participated in Startups as founders or collaborators, and from the analysis 67% declared they have a better understanding of the concepts after playing the game and no one on the survey declared that the game was not a contribution in learning about entrepreneurship concepts. After playing the game for the first time, students showed interest and wanted to know why they failed and what should they do better next time to be a better entrepreneur and raise more capital, and they prepared for the next game session class. 71% of the students actually rated the game as excellent or very good. Students, the industrial world, they are always complaining that University is all about gaining knowledge and studying, but they are not preparing professional for the real world. Serious games is a way to simulate and experience a more realistic approach to real life decisions that finally are going to shape and give value to your professional profile The future of research will be to work on another game related like a balanced score card, in order to have a serious games program that will allow the development and analysis of differences in the learning experience of courses.

Entrepreneurship Knowledge Transfer Through a Serious Games

155

One of the challenges is for game designers to better align the learning goals with game mechanics to produce more deeply engaging and effective learning game experi‐ ences, this is very difficult since you need Knowledge experts and the ability to transmit and to agree with the game developers, designers and programmers who must collabo‐ rate to produce an educational tool that is innovative and an engaging learning experi‐ ences. In the past 10 years, the field of learning games has been growing a lot. There are several opportunities for the use of games as a way to transfer knowledge in a more experiential way to teach graduate students to take better decisions while they develop startups or use and practice some management tools. The process of creating a learning game is indeed very complex if like in this case new knowledge has to be identify and created regarding a topic, using the Nonaka model is very helpful, but in this case it prove to be a long and laborious project, and large discussions were part of it, the experience of transferring an already existing program to a game should be a lot easier. There are several challenges to be met, the experience of using knowledge transfer to create an educational game that could transmit the experience of tacit knowledge was very successful but deeper analysis will be conducted. The knowledge transfer in learning games is Learning games seems to be a new and emerging field that operates at the intersection of a lot of professionals such as game researchers, educators, programmers, designers, learning designers, and topics experts, who have to collaborate to produce and innovative way to engage students in a powerful way and generating new classroom dynamics. There is a challenge in how this experiences could be better evaluated and the devel‐ opment of the teaching experience general guidelines will be done during the present year.

References Solis, B.: The Future Of Learning Is Stuck In The Past: Why Education Is Less About Technology And More About Behavior. http://www.briansolis.com/2014/03/future-of-learning-takeslearning/. Accessed 10 Dec 2016 Drucker, P.: Landmarks of Tomorrow. Harper & Brothers, New York (1959) Drucket, P.: Management Challenges for the 21st Century. Harper & Brothers, New York (1999) ELSPA: Unlimited learning: Computer and videogames in the learning landscape. Royaume-Uni: ELSPA. http://www.org.id.tue.nl/ifip-tc14/documents/ELSPA-report-2006.pdf. Accessed 16 Nov 2016 Freitas, S.D., Liarokapis, F.: Serious Games: A New Paradigm for Education? Introduction: Serious Games: A New Paradigm, pp. 9–23. https://doi.org/10.1007/978-1-4471-2161-9 Gross, B.: Handbook of Research on Serious Games for Educational Applications. Editorial Advisory Board, pp. 402–405 (2016) Ulicsak, M., Wright, M.: www.futurelab.org.uk/projects/games-in-education. Accessed 15 Nov 2016 Michael, D.R., Chen, S.L.: Serious Games: Games that Educate, Train, and Inform. Thomson, Boston (2006) Nonaka, Takeushi: The knowledge creating company: how Japanese companies create the dynamics of innovation, p. 284. Oxford University Press, New York (1994)

156

D. Liberona and C. Rojas

Susi, T., Johannesson, M., Backlund, P.: Serious games – an overview (2007). http:// scandinaviangamedevelopers.com/downloads/HS-IKI-TR-07-001_PER.pdf Freitas, S.D., Liarokapis, F.: Serious games: a new paradigm for education. In: Ma, M., Oikonomou, A., Jain, L.C. (eds.) Serious Games and Edutainment Applications, pp. 9–23. Springer, London (2011) Silva, V.V.: Método de Diseño de Modelos de Negocios Tecnológicos (2016) Squire, K., Jenkins, H.: Harnessing the power of games in education. Insight 3(1), 5–33 (2003) Sloetjes, M., Hoogendoorn, E. (n.d.): Serious games: more than just edutainment, pp. 1–3 (2014) Ulicsak, M.: Games in Education: Serious Games. Bristol Future Lab (2010). http:// media.futurelab.org.uk/resources/documents/lit_reviews/Serious-Games_Review.pdf. Accessed 11 Feb 2017 Xu, F.: The formation and development of ikujiro nonaka’s knowledge creation theory. In: von Krogh, G., et al. (eds.) Towards Organizational Knowledge: The Pioneering Work of Ikujiro Nonaka, pp. 60–76. Palgrave Macmillan, Basingstoke (2013) Zyda, M.: From visual simulation to virtual reality to games. Computer 38(9), 25–32 (2005)

Knowledge and Service Innovation

Knowledge for Translating Management Innovation into Firm Performance Remy Magnier-Watanabe ✉ and Caroline Benton (

)

Graduate School of Business Sciences, University of Tsukuba, 3-29-1 Otsuka, Bunkyo-ku, Tokyo 112-0012, Japan [email protected]

Abstract. This paper examines the role of tacit and explicit knowledge in translating innovation measures into firm performance in Japanese companies. While innovation has been found to be a source of higher firm performance, this research is considering whether innovation measures adopted by the firm trans‐ late directly into higher firm performance or whether these innovation meas‐ ures generate tacit and/or explicit knowledge which themselves produce higher corporate performance. Using a questionnaire survey and conditional process analysis, this paper found that there was no direct effect of innovation measures onto firm perform‐ ance, and that instead, both tacit and explicit knowledge fully mediated the rela‐ tionship between innovation measures and firm performance. Previous research did not consider the role of knowledge as interface to translate management inno‐ vation into firm performance. This paper uncovers the mediating role of knowl‐ edge, potentially elucidating past inconclusive results. Keywords: Management innovation · Tacit knowledge · Explicit knowledge · Firm performance · Japan

1

Introduction

Innovation as a major force of economic development was first highlighted in 1911 by Joseph Schumpeter [21] who proposed that change and new combinations brought about by entrepreneurs fuels economic growth. Over the last 50 years, however, the importance of leading and adapting to innovation has increased dramatically with the accelerating pace of technological progress that has been fuelled by the advances in information technology, which has had a tremendous impact on all industries. Kurzweil [13] states that the history of technological progress provides compelling evidence that change is not linear but exponential and will accelerate further in the future. As such, leading and/ or adapting to innovation and change are among the most important priorities for firms and organizations of the 21st century. In such a turbulent environment, the management of innovation at firms is increasingly difficult, and has spurred much investigation into the management innovation and how these have, or have not, led to higher corporate performance.

© Springer International Publishing AG 2017 L. Uden et al. (Eds.): KMO 2017, CCIS 731, pp. 159–169, 2017. DOI: 10.1007/978-3-319-62698-7_14

160

R. Magnier-Watanabe and C. Benton

This research examines the interface role of knowledge in translating management innovation into firm performance. Specifically, it analyzes the mediating effect of knowledge in the relationship between management innovation and firm performance. In the next section, we review the literature on management innovation and knowledge, and we develop hypotheses to investigate the current understanding of the relationship between management innovation and firm performance, and whether tacit and explicit knowledge mediate that relationship.

2

Literature Review and Hypotheses Development

2.1 Knowledge Knowledge has been described as representation of the real world and conceptualized as a product of the interaction between individual cognition and reality [23]. More generally, knowledge has been defined as information that has been proven true or useful through experience, and thereby embodies a high-value form of information that is structured for making decisions and taking actions. In other words, knowledge is neces‐ sary for making well-reasoned decisions in any context in both one’s private (personal) and public (business) lives. Polanyi [20], however, infers that knowledge is not simply singularly defined. He contends that “we can know more than we can tell” (p. 4) and that “it is not by looking at things, but by dwelling in them that we understand their joint meaning” [20, p. 19]. Building on this concept that not all knowledge is easily verbalized or conveyed, Polanyi [20] categorized two types of knowledge: tacit and explicit. Tacit knowledge is cognitive knowledge that is highly individual and difficult to express with language or numbers; for example, beliefs, points of view, technical skills and know-how are all part of tacit knowledge. Explicit knowledge, on the other hand, is objective and rational knowledge, and can be expressed with words or numbers; texts, equations, specifications and manuals. He continues by describing how the process of formalizing or archiving only externalizable or recordable knowledge to the exclusion of any tacit knowledge (know-how, expertise) is self-defeating and that tacit knowing is critical in achieving comprehension. For example, the expert and deeply-held tacit knowledge of master artisans or craftsmen, leading surgeons, or even a beloved teacher, cannot be gained merely through books, but through years of experience and doing. Also, to understand the true value or significance of any knowledge, one must reflect and understand the meaning of the knowledge in the context of the real world, or how it affects and is affected by the surrounding context. In the study of business and management, the paradigm that espouses organizations as centers of information processing has shifted to one where organizations are viewed as venues of knowledge creation [17]. Information here is defined as data that has been given meaning through connecting pieces of data and other information, while knowl‐ edge is information that has been given meaning through contextualization and thus is useful and purposeful. Nonaka [16] ’s seminal work on the “knowledge-creation theory” incorporates Polanyi’s dual classification of knowledge and suggests that organizational knowledge is created through the spiral process of tacit and explicit knowledge

Knowledge for Translating Management Innovation

161

conversion [15], defined as Socialization, Externalization, Combination, and Internali‐ zation (SECI). Nonaka [16] suggests that such incorporation of tacit knowledge is necessary for truly radical innovation. Indeed, according to Harlow [11] ’s research on American and Canadian firms, higher levels of tacit knowledge resulted in higher innovation performance. Less clear, however, was the relationship between tacit knowledge and financial measures of performance. Seidler-de Alwis and Hartmann [22] investigated the use of tacit knowl‐ edge in organizations and found that tacit knowledge, compared to explicit knowledge, can be of greater importance for innovation. Explicit knowledge by definition is public knowledge, or knowledge that is easily accessible, and thus can be readily copied, making it less valuable or sustainable as a source of competitive advantage. While past research has shown a positive link between knowledge management and firm performance [3, 18], empirical studies on the effects of tacit and explicit knowledge have not been consistent. Park et al. [19] remark that the “accumulated research on the performance effects of tacit and explicit knowledge has provided inconsistent results” (p. 89) and note that in some cases explicit and tacit knowledge transfers have been found to result in both positive and negative performance effects. Becerra et al. [2] have suggested that these inconsistent results are due to the specific conditions or measures under which knowledge is transferred. In this paper, to evaluate the effect of knowledge on firm performance, we hypothesize the following: Hypothesis 1: Tacit knowledge activities have a positive effect on firm performance. Hypothesis 2: Explicit knowledge activities have a positive effect on firm performance. 2.2 Management Innovation Management Innovation (MI) is defined as the establishment of new structures, novel processes, original systems and programs or practices in firms [7, 14, 24]. Walker et al. [24] noted however, that while the “antecedents, processes, and consequences of inno‐ vation in organizations have been studied by management scholars since the 1960s” (p. 416), the focus has been on technology-based product and process innovations rather than on non-technological innovations. Accordingly, and pursuant to their findings, they recommend further research that moves away “from the common practice found in studies of innovation in organizations to model a direct and independent effect of management innovation on performance.” In other words, the mechanism of the effect of management innovation on firm performance is still unclear and requires further investigation. Reviews of the literature on innovation in firms “continually suggest that its results are inconsistent” [4]. Atalay et al. [1] found that while product and process innovations have significant and positive impact on firm performance, no such relationship was found between non-technological innovation such as organizational and marketing innovation and firm performance. In contrast, results of research by Walker et al. [24] showed that there were “no differences in the direction and the strength of the association of MI [management innovation] and TI [technological innovation] on organizational performance” (p. 418), and that the organizational competencies gained from initiatives

162

R. Magnier-Watanabe and C. Benton

such as new ways of structuring and coordinating organizations and knowledge manage‐ ment activities were shown to be essential, especially in competitive markets. As previous research results have been inconsistent with regard to the effect of management innovation on firm performance, we hypothesize the following: Hypothesis 3: Management innovation has a positive effect on firm performance. We also investigate how management innovation supports knowledge creation. Specifically, we look at the relationship between non-technological innovation and performance by addressing Walker et al.’s. [24] recommendation. Since innovation is defined as the introduction of something new or a new idea, method or device, it assumes the creation of something that has not been developed before, and thus requires new knowledge or insight [5]. Accordingly, the authors suggest that the key may lie in how management innovation in organizations gives emphasis to tacit and/or explicit knowl‐ edge creation. Therefore, we hypothesize the following: Hypothesis 4: Management innovation has a positive effect on tacit knowledge activ‐ ities. Hypothesis 5: Management innovation has a positive effect on explicit knowledge activities.

3

Method

3.1 Sample and Measures The hypotheses presented above make up the research model depicted in Fig. 1. To test this model, a questionnaire survey of Japanese managers and staff of Japanese domestic companies in a wide range of industries was conducted. The survey instrument was first built in English, and then translated and administered in Japanese by the authors who are fluent in both languages. The data was gathered in March 2015 using a Japanese Internet Survey service with a large database of potential respondents throughout Japan.

H4

Management innovation

Tacit knowledge

H1

Corporate performance

H3

H5

Explicit knowledge Fig. 1. Research model

H2

Knowledge for Translating Management Innovation

163

Respondents were selected among Japanese employees working in Japan. Respond‐ ents numbered 310, consisted of 78% male and 22% female respondents from a wide range of industries and working in different functions, and were split evenly between those employed as general staff with no supervisory duties (155) and managers with subordinates (155). A majority of respondents were 40 years old or older, with an abso‐ lute majority having worked for their current company for 10 years or more, 27% of which worked at locations with less than 10 employees, 22% at sites with between 10 to 49 employees, and 23% at locations with 500 employees or more. Firm performance is typically measured using objective or subjective indicators, or a combination of both [9]. While objective financial measures are straightforward, their selection is a matter of availability, which is often linked to whether the firms under study are privately- or publicly-held. Firm performance was measured using three reflective questions on the respondents’ company performance in relations to that of its main competitors’, following Geringer and Hebert [8] ’s recommended use of subjective measures. The first one asked them about their firm performance in general in compar‐ ison to their main competitors’, the second one about labor productivity at their site compared with other establishments in the same industry, and the third one about their site’s financial performance, or profitability, compared with other establishments in the same industry. Management innovation was assessed using reflective questions from Eldring [6] who includes items based on Porter’s four types of strategies of cost leadership, differ‐ entiation, focus, and hybrid. These questions consider whether the firm has in place specific initiatives for innovation: integration of performance measurements for inno‐ vation activities in the performance measurement for executives; initiatives for finding, developing, and retaining key people driving innovation; recruitment and training investments to help reduce skills shortages; delegation of decision-making powers of innovation for line and project managers; material incentives for innovation managers (salary, bonus, promotion, etc.); intangible incentives for innovation manager (spaces, public recognition, challenging tasks, etc.); incentives for employees to develop their ideas (employee suggestion system); and organizational initiatives to more efficiently use human resources (team work, innovation circles, etc.). it is important to note that Eldring [6] ’s scale is especially relevant when looking at non-technological innovation, as is the purpose of this paper. Its indicators reflect management innovation through novel human resource practices. For tacit and explicit knowledge, with the former more important in personalization strategies and the latter more predominant in codifications strategies, reflective questions were derived from Hansen et al. [10] ’s findings on how consulting firms manage their knowledge. The first five items, whereby higher scores reflect an emphasis on explicit knowledge, assessed whether: the company’s business model focuses on using knowl‐ edge that can be used many times; the firm is more focused on gaining a large market share rather than high profit margins; the company’s knowledge management strategy is focused on ICT systems that codify, store and disseminate knowledge; human resource training is mainly done using manuals, computer systems or in large groups; and the company’s important knowledge can be codified, written down or stored in computers. The next five items, whereby higher scores reflect the prominence of tacit knowledge,

164

R. Magnier-Watanabe and C. Benton

assessed whether: the company’s business model focuses on providing highly custom‐ ized solutions to our customers’ unique problems; the company is more focused on generating high profit margins rather than gaining a large market share; the company’s knowledge management strategy is focused on facilitating conversations and face-toface exchange of knowledge; human resource training is mainly done through mentor‐ ships and in small groups; and the company’ important knowledge is not easily written down, but is the expertise and know-how of our employees that were nurtured through personal experience [10]. 3.2 Validity and Reliability Factor analyses were conducted with each subset of questions pertaining to each vari‐ able – management innovation, firm performance, and the use of tacit and explicit knowledge – to ensure that the questions displayed highest loadings on the intended constructs and to assess discriminant validity. Question items with excessive crossloadings, freestanding as one-item factors, or considerably reducing factor reliability were removed. Marked differences were found between the intended constructs and those obtained with the current sample. All factors were found to be reliable with Cronbach alpha scores above 0.7. For management innovation, a single factor with eigenvalue above one was gener‐ ated rather than the intended four factors. This single factor explained 61% of the total variance. For firm performance, a single factor with eigenvalue above 1 was obtained, explaining 76% of the total variance. For knowledge type, two factors with eigenvalues above 1 were achieved, the first one consistent with an emphasis on explicit knowledge and explaining 37% of the total variance, the second one indicating the importance of tacit knowledge and explaining 17% of the total variance, for a combined total variance explained of 54%.

4

Results

4.1 Exploratory Statistics Based on the respondents’ use of both explicit and tacit knowledge in their organizations, we assessed the levels of management innovation and relative firm performance. First, we compared respondents based on their self-reported organizational use of explicit knowledge, then based on their use of tacit knowledge. Those reporting to use more explicit knowledge (N = 70 vs. N = 150) worked in organizations with greater usage of management innovation (M = 3.66, SD = 0.60 vs. M = 3.14, SD = 0.73; t(218) = −5.134, p = 0.000) and higher relative firm performance (M = 3.23, SD = 0.75 vs. M = 2.73, SD = 0.82; t(218) = −4.316, p = 0.000). Considering their use of tacit knowledge, we found similar results, whereby those reporting to use more tacit knowledge (N = 121 vs. N = 77) worked in organizations with greater usage of management innovation (M = 3.54, SD = 0.64 vs. M = 3.02, SD = 0.77; t(196) = −5.089, p = 0.000) and higher relative firm performance (M = 3.12, SD = 0.79 vs. M = 2.50, SD = 0.83; t(196) = −5.294, p = 0.000).

Knowledge for Translating Management Innovation

165

Mean of Self-Reported Tacit Knowledge Use

In our sample, a relative majority of respondents rate their use of explicit and/or tacit knowledge as average (score of 3). And more importantly, an absolute minority of respondents reported using high levels of explicit knowledge and low levels of tacit knowledge in their organizations. Assuming that the sample is representative of the population of Japanese employees, this would suggest that Japanese firms make more use of tacit knowledge compared to explicit knowledge, which is consistent with previous research [15]. Several statistically-significant differences for management innovation and relative firm performance, using independent sample T-tests, were found by comparing the 3 groups of respondents based on their use of both explicit and tacit knowledge. The high explicit and low tacit knowledge group was omitted as it consisted of only eight indi‐ viduals. As aggregate statistics only give the big picture, we divided the population into a taxonomy based on the respondents’ levels of explicit and tacit knowledge use in their organizations. We calculated the mean of the 4 items of each factor for every respondent and clusters were made according to the mean of each respondent’s aggregate score and labelled as follow (Fig. 2).

5

Low Explicit High Tacit

High Explicit High tacit

4

18.7%

14.8% Others 45.5%

3 2

Low Explicit Low Tacit

High Explicit Low Tacit

1

18.4%

2.6%

1

2 3 4 5 Mean of Self-Reported Explicit Knowledge Use

Fig. 2. Taxonomy of respondents’ perceived usage of tacit and explicit knowledge

When the mean of the respondent’s aggregate score on explicit and tacit knowledge use are both greater than 3, he/she falls in the category of high explicit and high tacit knowledge use (N = 46, 14.8%). When the mean of the respondent’s aggregate score on explicit and tacit knowledge use are both lower than 3, he/she falls in the category of low explicit and low tacit knowledge use (N = 57, 18.4%). When the mean of the respondent’s aggregate score on explicit knowledge use is greater than 3 and that on tacit knowledge use is lower than 3, he/she falls in the category of high explicit and low tacit knowledge use (N = 8, 2.6%). When the mean of the respondent’s aggregate score on explicit knowledge use is lower than 3 and that on tacit knowledge use is greater than

166

R. Magnier-Watanabe and C. Benton

3, he/she falls in the category of low explicit and high tacit knowledge use (N = 58, 18.7%). The remaining respondents (N = 141, 45.5%) ranked either their tacit or explicit knowledge use, or both, as average, and make up the “other” category. The results clearly suggest that firms with higher levels of management innovation and with higher relative firm performance display greater self-reported use of both explicit and tacit knowledge. However, we cannot tell from the data whether low tacit knowledge and high explicit knowledge use, or the opposite combination of high tacit knowledge and low explicit knowledge use, are favorable since the differences in management innovation and firm performance were not statistically significant (Fig. 3).

4.0 3.5 3.0 2.5 2.0 Firm performance (mean) Low explicit / Low tacit

Management innovation (mean) Low explicit / High tacit

High explicit / High tacit Fig. 3. Firm performance and innovation measure by taxonomy of respondents

4.2 Mediation Model The following statistical tests use SPSS and PROCESS, a freely-available computational tool for SPSS that specifically addresses mediation, moderation, or conditional process analyses [12]. Table 1. Regression coefficients, standard errors, and model summary information for the presumed management innovation influence multiple mediator model Antecedent

Consequent M1 (Tacit K.) Coeff. SE 0.324 0.054

X (Management innovation) M1 (Tacit K.) – M2 (Explicit K.) –

– –

p 0.000

M2 (Explicit K.) Coeff. SE p 0.381 0.053 0.000

Y (Corp. Performance) Coeff. SE p 0.067 0.061 0.272

– –

– –

0.276 0.224

R2 = 0.145 F(1, 308) = 36.051, p = 0.000

– –

– –

R2 = 0.105 F(1, 308) = 52.357, p = 0.000

0.056 0.057

0.000 0.000

R2 = 0.155 F(3, 306) = 18.644, p = 0.000

Knowledge for Translating Management Innovation

167

We assessed the effect of organizational management innovation on relative firm performance both directly and indirectly through the levels of tacit and explicit knowl‐ edge. The regression coefficients, standard errors, and other statistics pertinent to the model are summarized in Table 1, and the path coefficients are shown on the statistical diagram in Fig. 4.

0.324**

Tacit knowledge

Corporate performance

Management innovation

0.381**

0.276**

Explicit knowledge

0.224**

Fig. 4. Standardized regression coefficients for the relationship between management innovation and firm performance mediated by tacit and explicit knowledge. **p < 0.001

The total effect is statistically significant (0.242, p < 0.001) and signifies that higher organizational management innovation result in higher relative firm performance. Addi‐ tionally, only the indirect effects (0.090 and 0.085 respectively) through tacit and explicit knowledge are statistically significant. It is important to note that the significance of the indirect effects is not assessed based on the statistical significance of the paths that define them but rather on asymmetric bootstrap confidence intervals which are entirely above zero (0.041 to 0.141 and 0.030 to 0.160, respectively) [12]. The normal theory-based Sobel tests (Z = 3.778, p < 0.001 and Z = 3.410, p < 0.001, respectively) agree with the inference made using a bias-corrected bootstrap confidence interval [12]. These findings indicate a full mediation of both tacit and explicit knowledge on the relationship between organizational management innovation and relative firm performance, thus providing support for H1, H2, H4, and H5, while there is no effect of management innovation on firm performance, suggesting H3 is not supported. Tests of moderation effects of both tacit and explicit knowledge by evaluating the statistical significance of the interaction terms revealed no such effect.

5

Discussion and Conclusion

5.1 Aligning Management Innovation and Knowledge Management A relative majority of respondents rate their firm’s use of explicit and/or tacit knowledge as average (score of 3) (Fig. 2). However, statistical results showed that firms with higher use of explicit and tacit knowledge reported higher firm performance, supporting

168

R. Magnier-Watanabe and C. Benton

Hypotheses 1 and 2. In addition, firms with more management innovation had employees reporting using both more tacit and explicit knowledge, thus providing support for Hypotheses 4 and 5. These two findings suggest that management innovation results in both higher tacit and explicit knowledge use, which positively affect firm performance. In our analysis of the initial research model, we found no direct link between management innovation and firm performance. Thus Hypothesis 3 was not supported. This result is consistent with some previous literature, which found no correlation between the two constructs [1]. However, we did find that there is a full mediation of both tacit and explicit knowledge on the relationship between management innovation and relative firm performance. In other words, while management innovation programs by themselves may not directly increase firm performance, aligning these programs with knowledge management initiatives enhances corporate performance of the Japanese companies surveyed in this study. This highlights the need for management innovation that supports knowledge creation and recognized the role of knowledge as interface between management and performance. Finally, the finding of the mediation of knowledge on the relationship between management innovation and relative firm performance could be a possible explanation for the inconsistent results found in previous studies. Indeed, in this research, tacit and explicit knowledge played a key interface role in linking management innovation and firm performance. 5.2 Limitations This study uses data for dependent and independent variables that were collected simul‐ taneously in the same survey instrument, raising the risk of common method variance. However, this paper has uncovered an important link between management innovation and firm performance, and future research should collect such data separately for further validation. Last, the data for this research includes only responses about Japanese companies in Japan and future research should evaluate whether the results obtained extend to firms in other countries. The authors suggest both quantitative and qualitative research across countries for upcoming investigations.

References 1. Atalay, M., Anafarta, N., Sarvan, F.: The relationship between innovation and firm performance: an empirical evidence from Turkish automotive supplier industry. Procedia Soc. Behav. Sci. 75, 226–235 (2013) 2. Becerra, M., Lunnan, R., Huemer, L.: Trustworthiness, risk, and the transfer of tacit and explicit knowledge between alliance partners. J. Manage. 45(4), 691–713 (2008) 3. Chen, E.T., Monahan, J., Feng, D.: A longitudinal cross-section examination of the implementation of knowledge management systems and firm performance. J. Int. Technol. Inf. Manage. 18(2), 223–238 (2009) 4. Damanpour, F., Wischnevsky, J.D.: Research on innovation in organizations: distinguishing innovation-generating from innovation-adopting organizations. J. Eng. Technol. Manage. 23, 269–291 (2006)

Knowledge for Translating Management Innovation

169

5. Damanpour, F.: organizational innovation: a meta-analysis of effects of determinants and moderators. Acad. Manage. J. 34(3), 555–590 (1991) 6. Eldring, J.: Porter’s 1980 Generic Strategies, Performance and Risk: An Empirical investigation with German Data. Diplomica Verlag, Hamburg (2009) 7. Evangelista, R., Vezzani, A.: The economic impact of techno-logical and management innovations: a firm level analysis. Res. Policy 39, 1253–1263 (2010) 8. Geringer, J., Hebert, L.: Measuring performance of international joint ventures. J. Int. Bus. Stud. 22(2), 249–263 (1991) 9. Greenley, G., Foxall, G.: Multiple stakeholder orientation in UK companies and the implications for company performance. J. Manage. Stud. 34(2), 259–284 (1997) 10. Hansen, M.T., Nohria, N., Tierney, T.: What’s your strategy for managing knowledge? Harv. Bus. Rev. 77(2), 106–116 (1999) 11. Harlow, H.: The effect of tacit knowledge on firm performance. J. Knowl. Manage. 12(1), 148–163 (2008) 12. Hayes, A.F.: PROCESS: a versatile computational tool for observed variable mediation, moderation, and conditional process modelling [White paper] (2012). http:// www.afhayes.com/public/process2012.pdf 13. Kurzweil, R.: The law of accelerating returns (2001). http://www.kurzweilai.net/the-law-ofaccelerating-returns 14. Lam, A.: Organizational innovation. In: Fagerberg, J., Mowery, D.C., Nelson, R.R. (eds.) Oxford Handbook of Innovation, pp. 115–147. Oxford University Press, Oxford (2005) 15. Nonaka, I., Takeuchi, H.: The Knowledge-Creating Company. Oxford University Press, New York (1995) 16. Nonaka, I.: A dynamic theory of organizational knowledge creation. Organ. Sci. 5, 14–37 (1994) 17. Nonaka, I., Umemoto, K., Senoo, D.: From information processing to knowledge creation: a Paradigm shift in business management. Technol. Soc. 18(2), 203–218 (1996) 18. Palacios Marqués, D., Garrigós Simón, F.J.: The effect of knowledge management practices on firm performance. J. Knowl. Manage. 10(3), 143–156 (2006) 19. Park, C., Vertinsky, I., Becerra, M.: Transfers of tacit vs. explicit knowledge and performance in international joint ventures: the role of age. Int. Bus. Rev. 24(1), 89–101 (2015) 20. Polanyi, M.: The Tacit Dimension. Peter Smith, Gloucester (1966) 21. Schumpeter, J.: The Theory of Economic Development; An Inquiry into Profits, Capital, Credit, Interest, and the Business Cycle. Harvard University Press, Cambridge (1961). Translated from German by Redvers Opie 22. Seidler-de Alwis, R., Hartmann, E.: The use of tacit knowledge within innovative companies: knowledge management in innovative enterprises. J. Knowl. Manage. 12(1), 133–147 (2008) 23. von Krogh, G.: Care in knowledge creation. Calif. Manage. Rev. 40(3), 133–153 (1998) 24. Walker, R., Chen, J., Aravind, D.: Management innovation and firm performance: an integration of research findings. Eur. Manage. J. 33(5), 407–422 (2015)

Product vs. Service War: What Next? A Case Study of Japanese Beverage Industry Perspective Zahir Ahamed1 ✉ , Akira Kamoshida1, H.M. Belal2, and Chris Wai Lung Chu3 (

)

1

Yokohama City University, Yokohama, Japan [email protected], [email protected] 2 Universiti Utara Malaysia (UUM), Sintok, Malaysia [email protected] 3 University of Surrey, Guildford, UK [email protected]

Abstract. Products, services or servitization concept; nothing is new anymore to the marketer these days. Historically, it is observed that the development of market has been shifting from one-phase to another phase in every 10 years. Since 1950s to the 1990s, the market experts have changed their focus from the produc‐ tion of goods to quality of products, selling to marketing approach, and products maintenance to service orientation respectively. In mid 2000s, the Japanese beverage industry has realized a big change, while many leading companies were diversified their business-focus and entered into rivals’ core business segments. As a result, the marketplace is quickly flooded as same category products and services-integration concept became spotlight to the marketer in the beginning of 2010s. However, today the services offer by soft drinks manufacturer cannot alone differentiate the firm from its rivals that pressure them to re-think its business strategy and finding new mechanisms for long-term sustainability. Therefore, this paper aims to examine the current market situation more precisely, identifying value deficiencies, and proposes a conceptual “Relationship Business Model”. Data was collected from three top leading Japanese beverage firms namely, CocaCola, Suntory, and ITO EN limited (CSI). The firms’ answer to a survey included multiple choices and open questions about the current market situation, chal‐ lenges, and key factors for future business success, and so on. Keywords: Commoditization · Productization · Servitization · Relationship Business Model · Re-thinking business strategy · Beverage industry

1

Introduction

In today’s competitive marketplace, products, services or product-service integration cannot carry any additional values to its customers or differentiate the firm from its rivals due to its similar offerings. Traditionally, it is observed that the competition between manufacturing firms has tended to focus on the goods itself. The firm acquired customers or increased its revenues through better quality of goods offering than those of its competitors. However, when goods were becoming more and more alike in terms of their quality and performance, the firm has started to shift its focus to another level of © Springer International Publishing AG 2017 L. Uden et al. (Eds.): KMO 2017, CCIS 731, pp. 170–181, 2017. DOI: 10.1007/978-3-319-62698-7_15

Product vs. Service War: What Next?

171

competition, i.e., product’s service component, which in later became spotlight to the marketer as one of their key strategic choices, so called, servitization of business [1, 2]. But, nowadays the services offering by a manufacturing firm has observed a very common business trait that cannot attract the customers significantly. As a result, the firms are realizing a massive pressure throughout the industry that pushes them to iden‐ tify an alternative business strategy and the source of long-term sustainability. In these circumstances, we realized the importance of ‘Relationship Concept’ and its implication in all marketing activities that may directed the firm’s proactive engagement in creating, developing and maintaining committed, interactive and profitable exchanges with selected customers over time. Meanwhile, the nature of the supplier-customer relation‐ ship has long been the subject of many research and debate both from academics and experts in the industry. In an attempt to draw all of the various relationship types together academics have produced many business relationship models. Probably one of the most prominent of these models was conceived by Golicic, Foggin, and Mentzer in 2003, who proposed a supplier-customer relationship was based on three basic configurations; arms-length, cooperation, and coordination respectively [3]. However, the development processes of interactive and beneficial relationships are still poorly understood and remain a new and complex concept. Hence, the purpose of this paper is to propose an interactive beneficial relationship development process model between provider and customer. In this study we aimed at exploring the current relationship stage between provider and customer of Japanese Beverage industry, and identifying the factors that influence to the effectiveness of developing an interactive and beneficial relationship. The main study materials were gathered with the help of field study, questionnaire, open discus‐ sion, and observations of three leading beverage companies in Japan, namely, CocaCola, Suntory, and ITOEN (CSI). The structure of this paper is as follows: after this introduction section the literature review is presented. Section 3 describes the research methodology was used in this paper. This is followed by Sect. 4, which presents a reallife case study of Japanese beverage industry including, current market situation, chal‐ lenges and future direction of success. The next section we articulate the output of our survey and depicted an interactive beneficial relationship development process model. Section 6 raised and an open discussion including the implication of our proposed model, merits and demerits and future debate of respective field. Finally, the paper is concluded with a constructive research summary and suggestions for the future research in the field of relationship marketing strategy.

2

Literature Review

Nowadays, either manufacturing or service industries both are increasingly stands as competitive and offering a greater choice than ever, which in turn showed the firm’s tendency toward commoditization in real marketplace. Particularly in manufacturing industries, it is very common and unique phenomenon of evolving marketing competi‐ tion characterized by increasing homogeneity of products, higher price sensitivity among customers, lower switching costs, and greater industry stability. During 2000s

172

Z. Ahamed et al.

period, a similar practice has been observed in Japanese beverage industries where many manufacturing firms transform of its branded products or services into a commoditylike one. In this regard, pricing became as the key buying factors for the customers, which are characterized by undifferentiated offerings and high transparency. Table 1. Relationship management related some literature review No. 1

2

3

4

5

6

7

Key findings of study Service quality Gaps Model

Title Source A conceptual model of service [6] quality and its implications for future research The market’s expectation of the quality of a Customer satisfaction, market [7] firm’s output positively affects customers’ share, and profitability: Findings overall satisfaction with the firm from Sweden These expectations are largely rational, albeit with a small adaptive component Putting the service-profit chain [8] The service-profit chain establishes relationships between profitability, customer to work loyalty, and employee satisfaction, loyalty, and productivity Managing customer relationship is an Managing customer [9] integral part of a company’s strategy, and its relationships input should be actively considered in decisions regarding the development of organizational capabilities, the management of value creation, and the allocation of resources Customer relationship management and A model of customer [10] business intelligence model, Factors of relationship management and business intelligence system for business intelligence systems organizational success for catalogue and online retailers Measurement model of customer Customer relationship [11] relationship management (CRM) capabilities management capabilities: Measurement, antecedents and consequences Measurement, antecedents and are influences that supports of customer consequences orientation, customer-centric organizational system and CRM technology on CRM capabilities, as well as the influence of CRM capabilities on organizational performance Demographic aspects and context effect on The Impact of Customer [12] customer relationship Relationship Marketing on Customer Satisfaction of the Arab Bank Services

In response to the commoditization of marketplace, many leading firms were increas‐ ingly accepted the concept of productization and servitization, and implemented into their respective business portfolios. In 2007, Baines et al. refers the productization as the evolution of the services component to include a product or a new service component

Product vs. Service War: What Next?

173

marketed as a product, while servitization directed the firm’s strategic innovation of creating value and processes to transform from selling products to selling an integrated product and service offering that delivers value in use and collectively meet the desired needs of client [2, 4]. However, today both strategic options are no longer make sense to the marketers that force them to think about an alternative business strategy to sustain its revenue growth and profitability. In these circumstances, relationship marketing has been observed as a major shift in marketing theory and practice. Most researchers agree that relationship marketing is the opposite of transaction marketing, in which transactional exchange involves a single, short time exchange with a distinct beginning and ending where relational exchange involves multiple linked exchanges extending over time and usually involves both economic and social bonds. According to the Harker in 1999, relationship marketing occurs when an organization engaged in proactively creating, developing and main‐ taining committed, interactive and profitable exchanges with selected customers or partners over time [5]. At a micro level, it is concerned with the nature of the relationships between firm and customer, which emphasize a long-term relationship that takes account of the customer’s needs and values. However, from macro perspective, it is directed the firm’s relationship with all of its stakeholders, thus the strategic objective of the firm should develop a mix and portfolio of the relationships. There are some relevant studies including commoditization, productization, servitization, and relationship management studies are shown as Table 1.

3

Methodology

3.1 Data Collection The methodology used in this paper was interviews and discussions with the employees of three multinational Japanese beverage firms respectively, Coca-Cola, Suntory, and ITOEN (CSI). Prior to the actual interviews and discussions, a pilot study was conducted with a prolong engagement of the firms’ corporate marketing team over six months period. This helps the authors to explain the objective of interviews to the respondents that ensure more consistency and clarity of the questions asked, which resulted to collect in depth and accurate feedback from interviewees. We conducted a total of 30 in-depth interviews separated into two distinct phases. Each of the interviews lasted between 40 and 70 min, and was recorded and subsequently transcribed into a verbatim. The first phase of these interviews was developed based on the field study where the interviewees were asked about the industry trend, current market situation and the challenges that they confront with today’s sales and marketing operations. The interviewees were designated in this stage as an area manager, branch manager, team leader and the peoples who engage in daily route sales and marketing activities. The second phase of the interviews was composed based on an accumulated result of the field study and interviews that we conducted in the first phase. The objective of this phase is to derive a conceptual model of an interactive beneficial relationship process model between provider and customer where they asked about current marketing

174

Z. Ahamed et al.

strategy, challenges, future policies and their opinion on specific issues of developing relationship marketing process. The interviewees were designated in this stage as the director of sales and marketing, human resource manager, marketing and communica‐ tion managers, operational managers, and service employees. In addition to primary data collection, the secondary data (company documentation and archival records) were collected as well, in order to achieve a theoretical triangulation. 3.2 Data Analysis and Procedure The data were analyzed using a thematic framework initially developed from the relevant literatures. A coding framework was then developed, and used Nvivo software (QSR International) for managing the vast amounts of data, annotations and memos recorded within the transcripts.

4

Case Study: Japanese Beverage Industry

4.1 Background and Trends The Japanese beverage market is comparatively rich in terms of its qualities and varieties of products and services offering than many other countries. The market size is calculated approximately JPY3,678 bil. by 2015 that observed as 3% real growth rate in 2016. Major players in this industry are Coca-Cola, Suntory, and Ito EN, which represents the annual sales as JPY1004 bil., JPY807 bil., and JPY397 bil. respectively by 2015 (source: company annual report, 2015). The market share is composed as 27% by Coca-Cola who are the dominant in this market just after U.S., Suntory is the second largest player accounted as 22%, and Ito EN 11% who are operating their business under carbonated drinks, fruit and vegetables juice, coffee drinks, sports drinks, and teas segments (Fig. 1).

Fig. 1. Market trends in-terms of values and segment

Product vs. Service War: What Next?

175

4.2 Market Analysis “Products vs. Service War” The Japanese beverage industry is realized as a matured market in terms of its quality and variety categories of products and services. In 2000s period, the market has been observed a massive diversification while many leading firms are extending their business portfolios from core brand to rivals business segment randomly. For an example, the market leader Coca-Cola’s debut as a green tea maker “Ayataka”, Suntory’s Iyemon, and Itoen’s Tullys Coffee. As a result, the market has crowded immediately with same categories of products and started “Products War”. A particular example of this products war are shown in Fig. 2.

Fig. 2. Products war between two leading brands of vegetable juice

The above Fig. 2 is shown a products war between two leading companies’ vegetable juice segment, which is first introduced by ITO EN limited in 2004 as the name of “Ichinichibun no Yasai” that mixed with 16 items of vegetables in the begining. But, just in 3 months later the rival company namely, KAGOME offering the same category of product by mixing with 17 vegetables items. However, this war is continued by increasing the items of mix vegetables and concluded as 30 items at the end. Another example of “Services War” that we observed in this industry from the beginning of 2010 to till now. At the initial stage, the services offering by a manufac‐ turing firm was to follow-up the customers, which in later extended as their products displays, event execution, and finally long hours of staying at customers’ business place for its products maintenance. Meanwhile, today these services are also realized a very common business practices for all manufacturer that does not carry any significant values to the customers. Thus, companies are facing a big challenges and looking more reliable business strategy that we have discussed in Sect. 4.3. 4.3 Challenges and Future Direction The Japanese beverage industry is highly competitive in terms of product innovation, quality, price, and promotional activities. However, the level of services is increasing day-by-day without a free of charge and reached almost in an extreme position that headed the firm to gain a very thin margin. On the other hand, declining the population

176

Z. Ahamed et al.

growth and increasing aged peoples reducing the volume of real consumption. Thus, the major challenges in this industry are identified as to improve the profitability by reducing cost and sustaining the revenue growth by acquiring valued customers and maintaining relationship with them. In these circumstances, our research is proposed a conceptual model of developing an “interactive relationship process model” based on interviewees of three leading beverage firms (CSI) that may ensure the firms’ sustainability and create customer lifetime values (CLV). The next section is depicted and discussed how a provider-customer relationship can be developed and create values for each other.

5

Relationship Development

5.1 Interactive Relationship Development Process Model Due to remarkable changes in the business environment and the continuous shifting of the market as well as its demands, the typical provider can no longer sustain in compet‐ itiveness by offering pure goods or services differently [13]. Firms have realized that, the strategic and beneficial relationship with customers is an effective way to sustain its business. In this study, the interactive relationship development process model as shown in Fig. 3, is proposed as a mechanism to build such strategic and beneficial relationship with customers.

Fig. 3. Interactive relationship development process model

The insight of this model is to utilize in business to business (B to B) perspective. Nevertheless, numerous models that approach of making the relationship between prof‐ itability and revenues in the service science field. One of them is service-profit chain [8]

Product vs. Service War: What Next?

177

linking the revenue and profitability growth with internal (employees and service quality perspectives), and external (customer perspectives) issues. Again, some approaches are very renowned to innovate business model in the perspective of organizational sustain‐ ability, such as business model canvas [14] and St. Gallen business model navigator [15]. Where, business model canvas helps to recognize that, what company do present situa‐ tion, what and where they need to change for meeting proposed value [13]. According to this recognition company can build a new business model for supporting companies in the product or service innovation game that can create and deliver service value in order to stay ahead in the market [13]. All over again, St. Gallen business model navigator provides the key issues to generate their new business model by knowing the target customer, value proposition to the customer, value chain for creation of proposed value, and the revenue structuring to cope that value [15]. However, the mentioned approaches are functioning as a general ways of doing business at the aim of sustaining in the market. Our proposed relationship process model is more specific to build an interactive relationships between firms and customers with hypothesizing that, organizational business sustainability is mainly driven by estab‐ lishing beneficial relationship between providers and customers, which is through into organizational design, ability, and selective customer segment by forming relationship enhancement space. Additionally, it explains the drivers of 1st two blocks as well as feedback of 2nd two blocks. In this proposed model, the flow of the processes is as follows: organizational sustainability is depending on its selective customer loyalty. Loyalty is happening, when co-create value by sharing customer’s expectations and experience with a provider’s facility, and the value co-creation is influenced by making customer satisfaction including with the confirmation of benefits for both parties. However, value co-creation and making loyal customer is a difficult task for product-based companies, because according to our data, it is required to build an interactive beneficial relationship between providers and customers. If so, then, providers need to make-sure their capability by reformatting their organizational design, cross functional collaboration competence, business partnership oriented setup, increase peoples’ skills & knowledge, as well as workstation design for extraordinary service-based solution delivery. This model is constructed with four blocks accordingly; providers, beneficial rela‐ tionship, customers, and outcomes that all inter-bonding to each other. The model suggested that the providers require to increase its organizational capability for devel‐ oping relationship and enhance this relationship by on-time value proposition through sharing information, customer education, business and/or personal communication, building trust and fulfill the commitment (which is considered in our model as relation‐ ship building & enhancement space) in the view of long-term benefits for both parties. Thereby, meeting customers’ desireness directly lead to the result of satisfaction that in largely influenced by the value providers’ offerings and the determined-value of customers, which in turn co-creating value and making customer as a fan and loyal to the organization. Loyal customers are the best marketing influence to acquire market share. They are willing to pay a price premium and bring more economic value in addi‐ tion to profitability.

178

Z. Ahamed et al.

5.2 Key Drivers of Building an Interactive Beneficial Relationship There are some key driving forces to guarantee of building relationship and its enhance‐ ment between provider and customer. Our data are told that, for a provider, the main drivers are executive marketing, value sharing & communication, employee education with up to date information, reward & recognition, and adequate tools to satisfy customers. The ‘executive marketing’ is considering here as the marketing activities between executive level of provider and customer side (such as; marketing manager of a service providing company and the marketing manager of the recipient-company). Since, the top level executive is the main responsible for decision making and directing firm operation by its own power rather than low level employees, the ‘executive marketing’ is the starting point to shape more valuable relationship with the customer. In ‘value sharing and communication’ phase, the service providing company is required to always keeps and maintains communication with the recipient-company and sharing all necessary updated information to them. Increasing the employees’ skills of service providing company’s with the view of ‘service oriented business thinking education and training’ is another key driving force for building relationship with the customer. The provider should take care about employees’ satisfaction as well. It could be by ‘reward and recognition’. Because, employee dissatisfaction is the main cause to intention of leaving the company, and the low employee turnover is linked to high customer satis‐ faction [16, 17]. Consequently, ensuring the ‘adequate method’ to measure the customers’ expectation and their satisfaction level is necessary for keeping in the perfec‐ tion of organizational business performance. As we already have mentioned that all initiatives should be interacting with the customers’ experience and their expectations, thereby making relationship and its enhancement space building is required. In this space, providers’ main intent to create or enhance the existing relationship more beneficial and long-lasting one. In order to do this, the provider should sharing all necessary information including, market trend, performance, and competitors’ position while they propose new values to the customers. In this space, the provider should also be responsive to develop an ‘interactive commu‐ nication’ with customers, which meaning is not only a communication about business purposes but also personal. The personal communication is more active way of improving the closer relationship with customers that ensure long-term benefits. There‐ fore, we believe that developing a personal relationship, a provider can easily access to the customers’ mind and becoming close-to-closer as like as a biz-couple (biz-couple refers to a relationship, in which both business partners are interdependent with each other and proactively recognize or loyal to meet the partner’s demand) that ultimately impact on real business performance. The personal relation also very vigorous to make ‘trusty and committed’ of each other. In that situation, if service providing company’s employees offer any business related proposal to the customer, then, there is a high possibility to accept it, which in resulting to create a mutual value for both parties and ensure win-win benefits for long-term perspective.

Product vs. Service War: What Next?

6

179

Discussion

As we observed that the market is too much crowdie with same categories of products and services due to the massive diversification of industry leading players into rivals’ core segments that can not differentiate the firm from its competitors and/or not attracting the customers significantly. So, marketer runs to identifying the new way of developing their competitiveness in the marketplace and finding a sustainable source of revenues growth and long-term profitabilities. In this circumstances, we assumed that the devel‐ oping an interactive relationship between provider and customer can be a good strategic choice for the firms long-term perspective. Thus, we proposed a conceptual model of developing an interactive relationship process based on the data collected from three industry leading beverage firms (CSI) that attempts to guide the firms for building and enhancing more beneficial relationship. The main message of this model is making the only relationship with customers doesn’t work well, it also requires making this rela‐ tionship beneficial and long lasting for both parties benefits. However, the beneficial relationship building is a difficult challenge for the company, because it necessitates to redesign organizational internal issues and external issues, which is related to customers direct participation. Hence, if a company is being very good in its internal aspects and not good at its external feature, then it may affect on organizational performance and in the broader sense to affect it to meet the philosophy of a company. This proposed model gives priority to the prominence of the links amongst organizational design that leads to available resources and logistics support, corporate collaboration, skills development, reward and recognition for employee satisfaction, and building relationship space for ensuring value co-creation with the customer. Interactive relationship process model helps to management of a company to over‐ come the challenges of making beneficial relationship with its customers. Thus, service provider and recipients can share their knowledge, resources, experience, and expecta‐ tion with each other to build or enhance a relationship with the aim of satisfying each other and ensure a sustainable growth and profitability for both parties. We trust that, our proposed model is workable for manufacturing industries who are targeting another corporation as their primary customer (B to B). For example; the customers of beverage manufacturer are initially consider as a supermarket chain, convenient store, drug store and so on. In addition to this, the model can be applicable in knowledge-based organization such as; consulting firm, pure marketing company as well as logistics firms. However, our model may create debate in theoretical aspects. Because, there is no enough explanation about novelty of new theory. Additionally, some service industries such as; hotel & tourism, restaurant, hospital, etc. may face difficulty to apply this model.

7

Conclusion

Commoditization, productization, servitization, and relationship marketing are four keywords that consist with this paper simultaneously. As we have discussed in early part of this paper and case study as well that the diversification of major leading Japanese

180

Z. Ahamed et al.

beverage firms are highly commoditized the marketplace and create a fierce competition within the industry. In order to attract the customers or differentiate the firm from its rivals, many of them in later transformed their business focus toward services phenom‐ enon along with its existing products offering, such as products maintenance, products displaying, and event execution, etc. But within a short period of time, the strategy ‘product service system’ became a very common business practice in almost every marketer that steadily loosing the firm of its distinguishing values from rival company and/or could not carrying the additional values to the customer. As a result, most of the manufacturing firms in this industry are realized a massive pressure and looking a reli‐ able alternative business strategy that may sustain the firms in long-run perspective. Therefore, our main study is set out with the aim of proposing an alternative business mechanism, i.e., an “Interactive Beneficial Relationship Process Model” that may help the firm to carry a sustainable revenue growth and profitability. In order to achieve that, we have collected the data from three top leading Japanese beverage firms (CSI) through an in-depth interviews, open discussion, and field study over six months period. According to the data analysis, we found that the key drivers for developing an inter‐ active relationship are executive marketing, value sharing and communication, customer education, and building trust & commitment between provider and customer. Conse‐ quently, the result suggested that building a beneficial relationship also require a personal contact rather than business relation only. In addition to these, the major challenges also identified in this paper through an open and constructive discussion with market experts, i.e., to improve the profitability by reducing cost and sustaining the revenue growth by acquiring valued customers as well as maintaining relationship with them. In this regard, as our research is proposed to develop an interactive beneficial relationship between provider and customer and raised the importance of maintaining relationship, so our next research could be to iden‐ tify the way of making this relationship more profitable considering the end-users into the account (B2B2C).

References 1. Zahir, A., Kamoshida, A., Inohara, T.: The role of organization in changing process towards servitization of business. In: The Paper is Presented in International Conference on Knowledge Management Organization, Tokyo Institute of Technology, 27–28 September 2011, Tokyo, Japan, ID.11, pp. 1–13 (2011) 2. Zahir, A., Kamoshida, A., Inohara, T.: Organizational factors to the effectiveness of implementing servitization strategy. J. Service Sci. Manag. 6(2), 177–185 (2013) 3. Golicic, S.L., Foggin, J.H., Mentzer, J.T.: Relationships magnitude and its role in interorganizational relationship structure. J. Bus. Logist. 24(1), 57–76 (2003) 4. Baines, T.S., Lightfoot, H.W., Benedettini, O., Kay, J.M.: “The servitization of manufacturing”, a review of literature and reflection on future challenges. Cranfield University, Cranfield, UK (2008) 5. Harker, M.J.: Relationship marketing defined? An examination of current relationship marketing definitions. Mark. Intell. Plan. 17(1), 13–20 (1999) 6. Parasuraman, A., Zeithaml, V.A., Berry, L.L.: A conceptual model of service quality and its implications for future research (1985)

Product vs. Service War: What Next?

181

7. Anderson, E.W., Fornell, C., Lehmann, D.R.: Customer satisfaction, market share, and profitability: findings from Sweden. J. Mark, 58, 53–66 (1994) 8. Heskett, J., Jones, T., Loveman, G., Sasser, W., Schlesinger, L.: Putting the Service - Profit Chain to work. Harv. Bus. Rev. 86, 118–129 (2008) 9. Bolton, R.N., Tarasi, C.O.: Managing customer relationships. In: Malhotra, N.K. (ed.) Review of Marketing Research, pp. 3–38. Emerald Group Publishing Limited, Bingley (2007) 10. Phan, D.D., Vogel, D.R.: A model of customer relationship management and business intelligence systems for catalogue and online retailers. Inf. Manag. 47(2), 69–77 (2010) 11. Wang, Y., Feng, H.: Customer relationship management capabilities: measurement, antecedents and consequences. Manag. Decis. 50(1), 115–129 (2012) 12. Al-Hersh, A.M., Saaty, A.S.: The impact of customer relationship marketing on customer satisfaction of the Arab Bank Services. Int. J. Acad. Res. Bus. Soc. Sci. 4(5), 67 (2014) 13. Belal, H.M., Yoneda, T., Takahashi, N., Hirata, N., Amemiya, K., Yamamoto, M., Shirahada, K.: Approach for organizational service climate creation: action research in a Japanese monitor maker. In: 2014 Portland International Conference on Paper Presented in Management of Engineering & Technology (PICMET), pp. 2449–2454. IEEE, July 2014 14. Osterwalder, A., Yves, P.: Business Model Generation a Handbook for Visionaires, Game Changers, and Challengers. Wiley, New York (2010) 15. Gassmann, O., Frankenberger, K., Csik, M.: The St. Gallen business model navigator (2013) 16. Pasupathy, K.S.: Sustainability of the service-profit chain. Doctoral dissertation, Virginia Tech (2006). https://vtechworks.lib.vt.edu/bitstream/handle/10919/26257/KalyanDissertation.pdf? sequence=1&isAllowed=y 17. Heskett, J.L., Schlesinger, L.A.: Putting the service-profit chain to work. Harv. Bus. Rev. 72(2), 164–174 (1994)

Comparing the Perceived Values of Service Industry Innovation Research Subsidiary Between Reviewers and Applicants in Taiwan Yu-Hui Tao(&) and Yun-An Lin National University of Kaohsiung, Kaohsiung, Taiwan [email protected], [email protected]

Abstract. A service industry innovation research (SIIR) project is a government subsidiary program in Taiwan for service-oriented small and medium enterprises (SMEs). A prior study was conducted to understand the perceived SIIR value by the awarded SMEs. This research collected the open-end survey data of 37 SIIR project reviewers with 10 identified value categories to balance the single perspective of 55 applicants with 11 perceived value categories. The similarities and differences were analyzed and led to three major findings. The comparative research result provides insightful knowledge for governmental funding agencies to reposition the project objective and procedural adjustments for sustainability. The result also presents a valuable reference for similar government subsidiary projects in Taiwan and in interested countries. Keywords: Small business innovation research
Service industry innovation research
Small and medium enterprises
Value system
Comparative study

1 Introduction An unofficial estimate indicates that government funding for private companies conducting research projects is an annual business of about NT$10 billion [1]. Service industry innovation research (SIIR) is the only government subsidiary for service-oriented small and medium enterprises (SMEs) after the US-originated small business innovation research (SBIR) was introduced to Taiwan in 1988 [2]. SIIR encourages SMEs to invest in new service products, new business models, new marketing models, and new business application technologies to increase competitiveness, in which technology is a key enabler for many applicants to apply technologies to their project proposals. Moreover, electronic commerce is the largest domain in receiving applications and approving subsidiaries to the service-industry SMEs. The SIIR annual budget is only NT$100–200 million [3], but the benefits it generates is significant [4]. Between 2006 and 2012, about NT$2.02 billion was invested in the SIIR project for awarding 1,171 SME projects that generated 7.5 times of the total income increase with a monetary value of NT$15 billion. However, the following issues persist. A total of 2,718 applicants did not pass the project review, majority of the SMEs in Taiwan are unaware that this SIIR project is available for service industries, and many complaints have been collected from the © Springer International Publishing AG 2017 L. Uden et al. (Eds.): KMO 2017, CCIS 731, pp. 182–189, 2017. DOI: 10.1007/978-3-319-62698-7_16

Comparing the Perceived Values of Service Industry

183

SMEs through the official annual performance tracking survey or other means. The complaints include excessive paperwork, strict accounting audit process, quantitative key performance indicators that are difficult to achieve within a year, often compressed project time due to the lengthy review and contract signing process, and academic reviewers who are unfamiliar with the industrial requirements. Nearly 4,000 SME applicants have been accumulated over the years, and many of them are repeated applicants. However, the SIIR project office continues to desire more new SME applications during the annual call-for-proposal period to meet the project objective of covering more new instead of repeating SME applicants. From the knowledge management perspective, rich tacit knowledge can be extracted from stakeholders to be stored and circulated for future utilization. Few research studies investigating SIIR can be found in the literature. Therefore, more research by academic communities should be encouraged to understand how the government could increase the performance of the SIIR project investment and meet the actual needs of SMEs in innovation research effort. Many of the failed applicants and some of the awarded SMEs did not agree with the SIIR objective or process. The perception gap of the SIIR SMEs is worthy of investigation, although SIIR had its position from the day its funding was approved by the Legislation Yuan in Taiwan. An investigation on the participating experiences of SIIR-awarded SMEs revealed that some means–end chain (MEC) studies in the literature adopted the existing personal value system in their value part of the attribute– consequence–value chain. Therefore, we are motivated to conduct a drill-down research to derive a value system for SIIR and other government funding projects because the value system should be domain or context dependent. Other government subsidiary projects may adopt or revise the value system better than start from scratch by building an initial value system template for SIIR. This paper is organized as follows. Section 2 briefly introduces the value systems. Section 3 presents the research design. Section 4 gives the analyses and discussion. Section 5 provides the conclusion and guidelines for future work.

2 Background Literature on Values This section briefly describes the background of values in the literature, which includes the traditional literature on personal values, corporate values, and perceived SIIR values by awarded SMEs in previous studies. Individual motivations for Internet users were divided into functional, experiential, and social motivations for classifying the perceived values of the second life [5]. Functional values include learning, shopping, making money, creating, researching, commercializing, and meeting/presenting; experiential values include exploring, entertainment, playing, vicarious experiences, diversionary, escapism and fantasy, and hobbies; and social values include socializing, dancing and clubs, romance, and cybersex [5]. In addition to Internet use, customer involvement, satisfaction, and service quality are used as connotations for classifying the values (objectives) in the exploration of customers’ store loyalty [6]. Transaction-specific, utility-oriented, and experiential perspectives were adopted when reviewing customer value in the study of women’s fascination with wedding photographs [7].

184

Y.-H. Tao and Y.-A. Lin

Specific value scales are used in MEC studies. The three most often used scales for measuring personal value were identified from the reviews of attribute–consequence– value [8]: Rokeach Value Survey (RVS), value and lifestyles (VALS), and a list of values (LOV). RVS contains 18 items in both instrumental values (e.g., changes in life and open heart) and terminal values (e.g., safety and freedom) [9]. VALS clusters consumers according to a series of consumers’ activities, interests, and opinions through an 800-item list or a reduced 36-item list [10]. LOV is a nine-item list that includes self-respect, being respected, self-fulfillment, sense of belonging, excitement, fun and enjoyment of life, warm relations with others, sense of accomplishment, and security. LOV is closer to the basic human being beliefs and is highly related to daily routine life compared with RVS and VALS [11]. Thus, the proposed multi-item LOV [12] for measuring the nine values in LOV was used in reference [8]. LOV was also used in an online shopping channel study [13]. From the emergent intellectual capital discourse, Tseng and Goo [14] argued that company (market) value is a combination of tangible value, traditional capital (e.g., physical and monetary capital), intangible value (e.g., intellectual capital), human capital, structural capital, and relationship capital [15, 16]. Moreover, Tseng and Goo [14] reviewed resource-based value and financial perspectives as a general review. Wenstop and Myrmel [17] attempted to structure organizational value statements by categorizing them into created values (e.g., return on investment, quality, image, and citizenship), protected values (e.g., health, environment, safety, and rights), and core values (e.g., integrity, honesty, and respect) from the stakeholder’s perspective. Data were collected from 43 Norwegian and 43 American listed companies; the top six values identified for American companies were integrity, honesty, respect, diversity, openness, and fairness, and those for Norwegian companies were honesty, respect, integrity, diversity, openness, and innovativeness [17]. The value systems in the literature could not be appropriately utilized in the SIIR scenario. Thus, a previous study of this research [18] conducted an interview. A total of 179 values were collected from 55 interviewees, with an average of 3.25 values per interviewee. A total of 11 value categories were identified after the coding process by the two interviewers. The categories in descending order were company growth, brand image, operational excellence, management competitiveness, marketing competitiveness, subsidiary, R&D competitiveness, product competitiveness, and miscellaneous positive and negative values. Categories 4, 5, 7, and 9 could be further combined into overall competitiveness. Categories 10 and 11 could be further combined into miscellaneous values. Few negative values were reported in the first half of the interview, although the interviewees made some suggestions on what could be improved during the second half of the interview.

3 Research Design This study aimed to determine the perceived value categories of SIIR project reviewers with a similar approach to awarded SMEs from prior research. The only exception was that the soft laddering interview was replaced by an open-end- question survey, in which the same questions were used in the soft laddering interview with the SMEs.

Comparing the Perceived Values of Service Industry

185

This research was conducted because of the perception gaps between SIIR reviewers and applicants based on our accumulated knowledge, observations, and experiences serving as SIIR project proposal reviewers for over eight years. The opposing positions in processing the SIIR project funding make the stakeholder roles of applicants and reviewers interesting to be investigated further. The targeted sample population was about 500 college professors who have been involved in the SIIR project since its initiation in 2006. Two issues were encountered during data collection. First, no list of reviewers and matching pairs of awarded SMEs with the corresponding chief reviewers was available. Thus, identifying the appropriately targeted reviewers was difficult. Second, college professors in Taiwan are usually too busy to be willing to accept interview invitations or reluctant to fill out questionnaires because of frequent requests. Therefore, we could only identify the appropriate target reviewers through publicly available information over the Internet and collect data using the same open-ended questions for the SME awardees through questionnaires by email. Several channels of target identification were applied, and they included retrieving a limited number of case publications by the SIIR project office with the chief reviewers’ names, Googling SIIR-related keywords along with the associated reviewers’ information, one author’s review sessions with known co-reviewers, and a publicly available SIBR reviewer list in which some professors could also be SIIR reviewers. Most of the interesting soft laddering interview research studies conducted 30–60 interviews. Thus, the present research aimed for at least 30 interviews or more. The questionnaire was used to interview SIIR-awarded SMEs. The first part of the questionnaire confirms the listed awarded SME case if known, or an option is presented to change to/fill in a new one. The questionnaire asks the chief reviewers to describe their SIIR projects, write down at least three outcomes with corresponding perceived values, and provide further suggestions to the SIIR project office. The subjects were reviewers with relevant expertise in college teaching and research. Thus, summarizing three perceived values from a familiar case should be easy according to a testing interview with two SIIR reviewers. An experienced graduate student who had processed the SME perceived values in the previous research first checked and reorganized the written coding. The coding result was further clustered on the basis of the similarity of the concepts (codes) and ranked according to the occurrences of the collected interviewees’ data, which were checked by one of the authors. The final coding result was a gradually formed agreement between the graduate student and one author through a negotiation process until no further conflicting concept or corresponding coding item could be found. We applied the same procedure to the SME data from the previous research to build an equal basis for the next comparison to fairly compare the categories of the reviewers’ perceived values with those of the awarded SMEs.

186

Y.-H. Tao and Y.-A. Lin

4 Analysis This section profiles the chief reviewers who returned valid questionnaires, lists the ranked perceived values by both SMEs and chief reviewers, compares the similarities and differences of the ranked lists of perceived values, and presents the findings from the comparison. We identified 79 reviewers who were involved with the SIIR projects with different indications. Among them, 27 were authors of SIIR cases, 10 were co-reviewers with known passed SIIR cases, 19 were co-reviewers with known SIIR cases that did not pass, and 23 were reviewers with associated SIIR keywords in the Google search results. We invited these candidates to answer the questions through an email inquiry and sent email invitations to familiar professors who could have been involved in SIIR project reviewing. The SBIR review list was obtained from the Internet. Thirty-seven of the forty reviewers returned valid questionnaires. Their affiliation locations were distributed from north to south cities: ten from Taipei, one from New Taipei, four from Taoyuan, one from Hsinchu, two from Miaoli, three from Taichung, one from Nantou, two from ChiaYi, two from Tainan, nine from Kaohsiung, and two from Pingtung. The SIIR project years were distributed as follows: two in 2007, five in 2009, three in 2010, one in 2011, two in 2012, four in 2013, six in 2014, eight in 2015, three in 2016, and three undisclosed. A reviewer who recalled any SIIR-awarded case could take only 15 min to answer the short-answer questions. However, a reviewer who could clearly not recall any SIIR case could take more time to recover the case documents or to retrieve it from memory, especially when the case was several years ago. Therefore, collecting the questionnaire was difficult given that the limited feedback from some professors, their inability to remember all the case details, lack of time to fill out the questionnaire, inability to remember details from long ago, and having no passed case. Therefore, we reviewed only 40 returned questionnaires out of nearly 200 email invitations, three invalid returned questionnaires, and a few incomplete but valid questionnaires. A total of 110 values were collected from the 37 interviewees, with an average of 2.97 per interviewee. Ten value categories were confirmed by the other authors after the coding process was conducted by one co-author. These categories are shown on the right-hand side of Table 1. The 179 values collected from the 55 interviewees of SIIR-awarded SMEs were reexamined. The values have an average of 3.25 per interviewee. Eleven categories were derived, as shown on the left-hand side of Table 1. The categories in their descending order are organizational culture, brand image, operational effectiveness, employee quality, resource integration, innovative R&D, market coverage, external affirmation, core competency, external pressure, unmet expected values for SMEs, resource integration, expert advice and counseling, business growth, innovative R&D, operational effectiveness, core competency, organizational culture, brand and market development, external pressure, and unmet expected values. Similarities and differences were found among the categories. The overlapping value categories were organizational culture, operational effectiveness, resource integration, innovative R&D, core competency, external pressure, and unmet expected values. Brand image and market coverage in SMEs were covered in brand and market

Comparing the Perceived Values of Service Industry

187

Table 1. Value categories ranked by count Rank Value category (count) SME 1 Organizational culture (41) 2 Brand image (36) 3 Operational effectiveness (20) 4 Employee quality (17) 5 Resource integration (17) 6 Innovative R&D (12) 7 Market coverage (11) 8 External affirmation (10) 9 Core competency (9) 10 External pressure (3) 11 Unmet expected values (3)

Chief reviewer Resource integration (22) Expert advice and counseling (16) Business growth (15) Innovative R&D (14) Operational effectiveness (14) Core competency (8) Organizational culture (7) Brand and market development (7) External pressure (4) Unmet expected values (4)

development in the chief reviewers. Employee quality and external affirmation were found only in SMEs, and expert advice and counseling and business growth were found only in the chief reviewers. Several findings from the comparison are summarized as follows. First, a high percentage of perceived values is consistent between SMEs and reviewers, that is, 7 out of 11 and 10 categories, respectively. This finding implies that certain common values are recognized by conflicting stakeholders in the SIIR project reviewing process. Second, priorities are different between SMEs and reviewers. The awarded SMES valued organizational culture and operational effectiveness the most, whereas the reviewers valued resource integration and innovative R&D the most among the overlapping value categories. A possible implication is that the SMEs’ were more concerned about long-term company sustainability, whereas the reviewers were concerned more about the project-oriented performance, which could be due to the fact that the reviewers are requested by the SIIR project office to thoroughly review the project benefits brought to the company. Third, the differences in value categories strengthen the second point of different priority concerns between the SMEs and reviewers. Brand image (36) and market coverage (11) are two categories with large counts among SMEs, and they roughly match with the brand and market development, which has a small count, among the reviewers. SMEs value employee quality (17) and external affirmation (10) the most, whereas reviewers value expert advice and counseling (16) and business growth (15) the most. In terms of corporate values, most of the 11 + 10 values can be categorized into the intangible value, and only resource integration, business growth, and market coverage can be categorized into the tangible value of Tseng and Goo [14]. Majority of the 11 + 10 values fall into the created value (e.g., quality and image) and the core value (e.g., openness and innovativeness), as identified from the American and Norwegian companies in Wenstop and Myrmel [17]. Nevertheless, the identified 11 + 10 value

188

Y.-H. Tao and Y.-A. Lin

categories are difficult to be appropriately mapped to the existing corporate and individual values from the literature. How can these value categories be used in the future? One potential usage in research and practice is having the extracted value categories in Table 1 serve as a measurement scale in future applications for collecting user perceptions. The answer for each category may range from one to five (or seven) to represent negative to positive perceptions. Therefore, despite the positively oriented perceptions from the SIIR SME and chief reviewer subjects, the extracted value categories still represent the true perceived value system for two critical stakeholders in government subsidiary projects.

5 Conclusions Fifty-five SIIR-awarded SMEs were interviewed. A total of 179 potential values were extracted, were clustered into 11 categories, and then ranked according to the corresponding counts. Similarly, 37 chief reviewers surveyed and extracted 110 potential values, which were clustered into 10 ranked categories. The comparison between these two lists of perceived values by the conflicting stakeholders of awarded SMEs and chief reviewers reveals similarities and differences. This research recommends that the SIIR project office and corresponding high authorities should proactively incorporate the SMEs’ perceived value system into the next SIIR project proposal writing and evaluation procedure for potential improvements. Three findings were derived from the similarities and differences in value categories. They provide insights for the Department of Commerce and SIIR project office for future strategic and tactical deployment to sustain SIIR projects that must be justified and approved by the Legislation Yuan in Taiwan. The strategies and tactics may focus on project repositioning and promotion for improved long-term return-of-investment performance by the project officers. This research-based evidence can help to clarify the confusing sources of messages, especially for SMEs that are not familiar with SIIR. The word-of-mouth values from the 55 SIIR-awarded SMEs and 37 chief reviewers are a good reference for interested, qualified, and potential SMEs. Acknowledgements. The research is partially supported by Ministry of Science and Technology with grant number 103-2410-H-390-018-MY2.

References 1. Leading Consulting Group: Taiwan Research and Development Subsidiary Roadmap (in Chinese). http://www.leadconsult.com.tw/cov.asp 2. Lai, G.: The trend and experiences of Small Business Innovation Research from small and medium enterprises in primary countries for Taiwan. Govern. Res. Plan. 35(5), 139–148 (2011). (in Chinese)

Comparing the Perceived Values of Service Industry

189

3. Leading Consulting Group: Taiwan Research and Development Subsidiary Project Listing (in Chinese). http://www.leadconsult.com.tw/cov.asp?ID=2 4. Yu, S.: Current status of service industries innovation research. CPC Knowledge Management Center (2013). (in Chinese) 5. Zhou, Z., Jin, X.-L., Vogel, D.R., Fang, Y., Chen, X.: Individual motivations and demographic differences in social virtual world uses: an exploratory investigation in Second Life. Int. J. Inf. Manage. 31, 261–271 (2011). http://dx.doi.org/10.1016/j.ijinfomgt.2010. 07.007 6. Lee, W.-I., Chang, C.-Y., Liu, Y.-L.: Exploring customers’ store loyalty using the means-end chain approach. J. Retailing Consumer Serv. 17, 395–405 (2010). http://dx.doi. org/10.1016/j.jretconser.2010.04.001 7. Huang, S.-C., Chen, C.-Y.: Exploring women’s fascination with wedding photographs from the perspective of value and attractive quality theory. J. Manag. Syst. 18(4), 581–605 (2011). (in Chinese) 8. Chiou, W.Z., Lee, M.-F.: An application of means-end chain to the purchasing behavior of mobile phones. Commun. Manag. Res. 1(2), 213–237 (2002). (in Chinese) 9. Rokeach, M.: The Nature of Human Values. Free Press (1973). doi:10.2307/2149267 10. Kahle, L.R.: Social Values and Social Change: Adaption to Life in America. Praeger, New York (1973) 11. Kahle, L.R., Kennedy, P.: Using the list of values (LOV) to understand consumers. J. Consumer Mark. 6(3), 5–12 (1989) 12. Bearden, W.O., Neteyemer, R.G., Mobley, M.F.: Hand Book of Marketing Scales: Multi-Item Measures for Marketing and Consumer Behavior Research, 2nd edn. Sage Publications, Inc., London (1999) 13. Ho, Y.C., Lu, C.J., Fok, C.K.: The impact of on-line shopping channel on customer value judgment: a means-end chain model. J. Far-East Univ. 24(2), 119–130 (2007). (in Chinese) 14. Tseng, C.-Y., Goo, Y.-J.J.: Intellectual capital and corporate value in an emerging economy: empirical study of Taiwanese manufacturers. R&D Manag. 35(2), 187–201 (2005). doi:10. 1111/j.1467-9310.2005.00382.x 15. Roos, J., Roos, G., Edvinsson, L.: Intellectual Capital – Navigating the New Business Landscape. Macmillan Press Ltd., Basingstoke (1998) 16. O’Donnell, D., O’regan, P., O’regan, V.: Recognition and measurement of intellectual resources: the accounting-related challenges of intellectual capital. In: PAKM 2000 Third International Conference on Practical Aspects of Knowledge Management, Basel, Switzerland, October 2000 17. Wenstop, F., Myrmel, A.: Structuring organizational value statements. Manag. Res. News 29 (11), 673–683 (2006). http://dx.doi.org/10.1108/01409170610715990 18. Tao, Y.-H., Uden, L.: The perceived value of small industries innovation research in Taiwan. In: The 11th International Conference on Knowledge Management in Organizations, Hagen, Germany, 25–28 July 2016

Validation Tools in Research to Increase the Potential of its Commercial Application Anna Závodská1(&), Veronika Šramová2, and Anne-Maria Aho3 Research Centre, University of Žilina, Univerzitná 8215/1, 010 26 Žilina, Slovakia [email protected] 2 University of Žilina, University Science Park, Univerzitná 8215/1, 010 26 Žilina, Slovakia [email protected] School of Business and Culture, Seinäjoki University of Applied Sciences, Kampusranta 11, Frami F, 603 20 Seinäjoki, Finland [email protected] 1

3

Abstract. Researchers are being challenged to show how their research is contributing to commercialization in the private sector, especially when they received public funding. At the same time, the private sector is challenged to further develop their business models so they reflect on the continuous innovation is all fields. For innovation to be successful, it is important to validate it with customers and so co-create the value. This paper increases the awareness of the paramount role that validation techniques play in research by providing recommendations to properly validate research efforts. A validation tool based on the co-creation of value with customers was proposed. Keywords: Innovation development
Research


Validation

Interviews

Value

co-creation




Product

1 Introduction Innovation is a crucial activity for any type of organization in order to remain competitive. Innovation can be achieved by change in the process, product or business model. Thanks to the change in the business model many startups have launched their successful products. This can be hardly done in research but what differentiate startups from research institutions is the ability to quickly validate their idea and shape it according to the customer needs so it seamlessly solves their problems. Researchers have realized that their research results have to be not only innovative but also commercially applicable and reflect on demand. To achieve that, it is important to validate their research ideas with those who are their potential customers. Moreover, they need to co-create the research outcome with them. Unfortunately, many research innovations are not successful, usually due to the uniqueness which can be hardly monetized in the commercial sector. The biggest problem of commercial application of research results is lack of validation and co-creation of value.

© Springer International Publishing AG 2017 L. Uden et al. (Eds.): KMO 2017, CCIS 731, pp. 190–199, 2017. DOI: 10.1007/978-3-319-62698-7_17

Validation Tools in Research to Increase the Potential

191

This paper begins with brief review of innovation, followed by co-creation of value and validation techniques. Subsequent section describes the case study involving commercial company and research institution. This is followed by discussions of why the research results are not successful on the market. The paper concludes with suggestions for further research.

2 Theoretical Background to Innovation and Value Co-creation The generation of new ideas and their commercialization has traditionally been done internally [2]. However, many companies such as startups have realized the importance of sharing their innovative ideas and their validation with potential customers. Many books devoted to this subject has emerged [1, 3]. Based on the customer validation techniques startups are able to quickly validate the potential of their ideas and consequently better shape their products accordingly. Moreover, the result is much more competitive as if they develop their idea without any validation. We believe that researchers could be more successful in their commercialization activities if they use startup approaches such as customer interviews for their research ideas validation and customer development instead of product development. Therefore, we suggest using these approaches in research and development as they might have significant impact on the applicability of research results in practice. Below, we describe the main definitions of innovation and approaches that have positive impact on the innovation and its diffusion.

2.1

Innovation

“An innovation is an idea, practice, or object that is perceived as new by individual or other unit of adoption. [18]” Innovation consists of the generation of a new idea and its implementation into a new product, process, or service, leading to the dynamic growth of the national economy and the increase of employment as well as to a creation of pure profit for the innovative business enterprise [6]. Peter Drucker (2002, p. 96) defines innovation as “the effort to create purposeful, focused change in an enterprise’s economic or social potential,” and underscores compelling customer value, opportunity, and impact [9]. Concerning the purpose of this paper is important to understand the open innovation as an approach that needs to be used in modern research institutions. Chesbrough et al. [8] defined open innovation as ‘the use of purposive inflows and outflows of knowledge to accelerate internal innovation, and to expand the markets for external use of innovation, respectively. Open innovation assumes that internal ideas can also be taken to market through external channels, outside the current business of the firm, to generate additional value.

192

2.2

A. Závodská et al.

Co-creation and Its Role in Innovation

“Collaborative innovation is first and foremost a mindset. Companies seeking to engage in collaborative innovation will have to earnestly examine their culture and beliefs, organizational design and technological infrastructure before committing to new and productive partnerships.” [2]. Achieving collaboration has to be a deliberate, strategic and, a planned function of the business. Collaboration as a human behavior can be both complex and interactive and it is hard to develop metrics that measures collaboration attitudes, behaviors, and outcomes [5]. Collaboration is a deep-seated human response to turbulence which individuals or individual organizations are unable to manage. Process, capability, value proposition and channel have been proposed as the four key drivers which lead to collaborative structures. The development of a new value propositions requires proposition-based collaboration. The best strategy for organizations is high collaboration and high competition, in a spirit of networking excellence, with the major benefit of competitive advantage (learning!) for all the participants [19]. Montoya-Weis and Calantone’s study [16, 1994] attributed the failure of a new product in market because firms failed to understand the needs of the customers. For the new products to succeed in the markets, firms should be responsive to both current and potential customers’ needs. A customer-oriented culture facilitates innovative capability of the firm.

2.3

Validation and Its Role in Innovation

Balanced Scorecard As an example of a strategic management perspective on value creation, the Balanced Scorecard is presented [10, 11]. The balanced scorecard links performance measures and provides answers to basic questions from different perspectives. From a customer perspective: how do customers see us? From an internal perspective: what must we excel at? From an innovation and learning perspective: can we continue to improve and create value? From a financial perspective: how do we look to shareholders? [10]. Kaplan & Norton [12] introduced strategy maps as useful tools to visualize how intangible resources translate into corporate goals. According to the authors, strategy maps give employees a clear understanding how their jobs are linked to the overall objective of the organization [12, 15]. Bremser and Barsky [4] integrated the Stage-Gate approach to R&D management by using BSC. They created the framework in order to present how firms can link resource commitments to these activities and the firm’s strategic objectives. García-Valderrama et al. [7] found two reasons for the usefulness of BSC in the context of R&D. Firstly, there are difficulties to be found in the employment of some of the indicators traditionally utilized in the measurements of the returns from these activities (Donnelly 2000). Secondly, we lack a common vision of selection of indicators and their dimensions. In addition, we have a lack of alignment of the

Validation Tools in Research to Increase the Potential

193

measurements of the success of R&D activities. According to the recommendation of the literature on management of R&D, BSC is one of the instruments for the measurement of the success of R&D activities [4, 13, 14, 17]. Customer development Every enterprise goal is to secure prosperity of business and to achieve it through manufacturing or services [20]. Customer development helps towards achieving this goal. Customer development works for companies of all sizes, not just startups. Every hour spent on customer development saves five, ten, or even more hours of writing, coding, and design. Customer development, with its focus on small-batch learning and validation, can promote internal innovation [1]. Von Hippel notes that involving customers in the innovation process has several advantages [9]: 1. The lengthy trial-and-error period in understanding detailed customer needs is shortened because trial and error is accomplished by the customer. 2. Customer input can actually lead to a design ready for manufacturing. 3. Small customer niches that are otherwise too expensive to serve can be reached. The term customer development is meant to parallel product development and also does not replace product management and product vision. It is not user research. Customer development is a hypothesis-driven approach to understanding: who are your customers, what problems and needs they have, how they are currently behaving, which solutions customers will give money for (even if the product is not built or completed yet), how to provide solutions in a way that works with how your customers decide, procure, buy, and use. Customer development simply adds two components: a commitment to stating and challenging your hypotheses and a commitment to learning deeply about customers’ problems and needs [1].

3 Research Methodology Qualitative research was used as a core approach for data collection. This approach has involved the following methods: • Case study of the Machinery Company in Finland. • Case study of the Research Centre at the University of Žilina in Slovakia. • Analysis of validation techniques which can be used in research for improving its commercial application. We used comparative analysis for comparison of chosen startup and other validation techniques in order to analyze which techniques can be used for validation of research at universities. Research approach involves case studies from which data were cross-examined. Two different organizations were chosen intentionally to demonstrate the way how they do research and create innovations in various industries which have both research and development as a core activity in common in order to be competitive and be sustainable in long-term. Machinery Company was chosen as a successful company with high level of innovation activities and co-creation of value with customers. The case of Research Centre was chosen to demonstrate innovation activities which are hardly applicable in commercial sector.

194

A. Závodská et al.

Machinery company is a multinational company producing machines and systems for sheet metal working. The empirical domain of the study is the company’s software innovation process, which supplies embedded software for metal sheeting machines. Research Centre of the University of Žilina functions under the jurisdiction of the University of Žilina, but as a separate entity and is completely dedicated to research. It was build three years ago in order to trigger applied research at the University. It was part of the EU project. It has several unique laboratories where the research is conducted. Traditionally, researchers carry on their research without previous demand. As a result, many of the patents have never been used in commercial sector. Therefore, the main question of our research was: is the new product (research outcome) more successful on the market if the researchers validate their research idea with customers and consequently collaborate on developing it? In the Machinery Company methodology involved a case study research. In data collection, several methods were used: interviews, workshops and observation of the software engineering process. Data for comparative analysis were collected from websites, scientific papers and literature concerning validation techniques and examined by content analysis. Analysis of all data gained from these secondary sources of research were performed and enriched by authors’ opinions.

4 Analyses of Case Companies Companies try to bring new innovations, new products for solving customers’ problems. Usually these innovations are not successful. Below, we present two cases of different organizations. One is fully commercial company which is solely dependent on their customers and their purchases. The other one is mostly financed by public funds either from government or from EU funds.

4.1

Changing the Whole Business Philosophy Towards a Successful Innovation in theMachinery Company

Competition can concern the price of the product or core competences. This puts pressure on the need to improve the efficiency and customer orientation of the development processes. Current processes are not effective. This is reflected in the lack of information about customer value, incorrect timing and inaccuracy in the development processes. Because of the strategic role and changing environment of software engineering process, the improvement methods based solely on the software engineering process are not sufficient. In order to improve innovation process and create value for customer new methods are needed. A new evaluation tool is specified based on the analysis of the software innovation process and the changing environment. The four perspectives of the tool can be defined as follows: improved customer knowledge, improved process and integration, improved profitability, ability of continuous development.

Validation Tools in Research to Increase the Potential

4.2

195

Problematic Research Commercialization at the Research Centre

Slovak universities have no tradition in building and managing research institutions. Commercialization of research results has always been done intuitively by university departments or individual researchers. Moreover, applied research is lagging behind and commercialization of research results has no successful outcomes. Many of the patents has never seen the lights of the real world. As a result, six Slovak universities received funding from the European Union to build science park or research center in 2013 so they solve the problem with research and its commercialization. This paper investigates the Research Centre which is an organizational unit of the University of Žilina in Žilina and moreover, it is an EU project. Thanks to this project many unique technologies were bought as well as a lot of researchers get a chance to work at the Science Park or Research Centre in order to conduct an advanced research and contribute to improvement of applied research and its commercialization. However, due to the delay in public procurement, technologies were procured very late and the second phase of project has not yet been approved so the institutions lack finances to cover the costs of operation of technologies as well as keeping jobs of researchers working at them. Currently, the biggest problem with performing an outstanding research is lack of researchers as well as collaboration with companies. Collaboration can bring ideas into research as well as ensure the cash inflow. Many patents could be potentially sold out if the researchers find right application. This can be achieved by customer validation where they can obtain valuable feedback on their prototype or technology by customer. This can be then applied to further product development and customers can be involved in the development. Both sides can benefit from co-creation process.

5 Discussion Based on the previous analysis the following recommendations are provided to organizations fighting problems in research and its commercialization. • Changing organization’s culture which supports open innovation and collaboration. Change of an organizational culture towards open innovation and collaboration in research is well described in [8, 9]. • Initial testing of ideas where researchers discuss their ideas with other researchers and validate them with other experts. • Preliminary market pre-screening in order to test the initial idea and confront it with existing research results. Researchers should conduct technology analysis – going through patents and analyzing if there is a negligible or significant technological advancement. • Conducting interviews with potential customers. There can be used various validation techniques for that such as balanced scorecard, Steve Blank’s interviews with customers [3], etc. • Product development focused on continuous co-creation of value with customers resulting in prototype.

196

A. Závodská et al.

• Ensuring that customers understand benefits that product is offering them. Explaining the right benefits of products can guarantee that the product is well promoted and so easier to sell to customer. • Choosing the best option for technology transfer, finding the best business model. Sometimes selling licenses can bring more money to institution than to cover the costs of internal technology transfer. • Trial sell of prototype to chosen customers and gaining relevant feedback from them. • Product launch with complete marketing and other plans. Based on the recommendations the following tool is proposed. This tool is focused on the part of research validation and developing a prototype which we consider as a crucial part that helps deciding about the success of the future research outcome. It is very important to realize that it is not always possible to find a compromise and to fulfill all customers’ requirements and wishes, especially in research. For this reason, cooperation of customers and researchers is very important because this enable both parties to achieve mutual satisfaction.

5.1

Developing the Evaluation Tool for Value Creation Based on Balanced Scorecard andOther Validation Techniques

We proposed a tool for validation of R&D ideas and tested it in the Machinery Company. The results were positive, mainly in terms of improving decision-making process of managers deciding about the development of new ideas. We further elaborated the tool so it could be used also in the research and university environment. Currently, the significance of the development of value creation is well understood at companies but it is important to develop such a tool for the research institutions as well. The aim of this tool is to concretize the required perspectives into normal working practices. Evaluation is done using the following perspectives: customer, financial, internal process and learning and growth, based on the perspectives of balanced scorecard. In addition to this, the competitive position is considered. The idea of the tool is to draw attention to different aspects of the innovation project even before the development project begins. To implement the tool scoring of key elements of the project from different perspectives is crucial. Scoring can be done by each institution concerning the importance of each factor for them. We propose set of questions that can be used in any type of business if they are modified to the needs of any researcher based on the nature of the innovation. In the following some concrete factors to be scored of different perspectives in the machinery industry are described: • The customer perspective: Increasing efficiency, shortening of set-up time and improved usability. In addition, competitive position can be evaluated by the following factors: Adding/sustaining competitive advantage and influence on the competitive position.

Validation Tools in Research to Increase the Potential

197

• The financial perspective: Estimated customer price, fixed and variable costs, installation and maintenance costs per unit. Based on this information, contribution margin and break-even point can be calculated. • The process perspective: Shortening of start-up time and installation time, re-usability of software modules and increased knowledge of product development/ learning. • The learning and growth perspective: Increased knowledge of product development/ learning. When the questions are clear the measurement method has to be proposed. Managers has to decide how they will evaluate if the research outcome or product is going to be approved for development or not. In the research institution, different questions can be proposed such as: number of patents, number of citations, number of papers (current contents), scientific awards, scientific breakthrough, etc. which are totally different from companies, however, could be more important than financial benefit for the certain institution. Therefore, the scores and the value they can give to each factor can vary. Managers can decide whether financial benefit from the patent or number of publications from the invention is more important for them concerning each research outcome. Scorecards can look differently. We can use traffic lights, points or whatever scale we consider is best for each evaluation (Fig. 1).

Fig. 1. Proposed mix of four-perspectives’ questions and scoring system

When looking at the customer perspective we need to realize the importance of customers in the development process. It is not only about our perceived value that we think that product can bring to customer (and scores we relate to each factor) however it is more about the value we co-create with the customer based on the interviews. For interviewing the customer, we propose to use the methodology developed by Steve Blank [3]. There is a whole customer development methodology described in the book.

198

A. Závodská et al.

We think that customer development is not only for startups but also for researchers who would like to move their research into another level. Besides customer and internal validation we have to also check the validity of the technology itself. It can be done by patent analysis but in most cases this is not enough. Some technology might not be that unique however the change in the business model can bring the most benefits. For the validation of technology researchers can use so called Technology Evaluation Criteria [9] which contains of various questions such as: To what degree is the innovation more advantageous to existing technologies? Does it reduce cost, save time, or improve quality? To what degree is the innovation compatible with existing values, experiences, capabilities, felt needs, and organizational and cultural? To what degree is the innovation complex and difficult to adopt by the users? What degree of specialized training is required before the innovation can be adopted? What specialized equipment is needed? After the complex validation researchers can move on to the development of their products and use co-creation for that. There exist several approaches which are similar in their nature. They all consist of set of validation questions that need to be asked management, customers, researchers themselves to ensure that the technology is worth commercializing and is bringing benefits to customers. With the combination of various tools, we can ensure the value that will be co-created with researchers and customers at the same time. We hope that our tool can help in achieving that.

6 Conclusion This paper investigates the innovation process in two different organizations, one from commercial and one from university environment. It shows that companies from commercial environment including startups have realized how important it is to validate their ideas before they start actual product development. However, at the university environment, the validation of ideas is limited and sometimes even impossible. Therefore, they often fail in their innovation commercialization. It is due to the lack of validation of their research ideas with potential customers and co-creation of value with them. Thus, we proposed a set of recommendations as well as proposed mixture of validation tools that can be used in order to increase not only quality of research but moreover they can increase the commercial application of it. It is our belief that this tool can be adapted to any research institution or department within company. In order to verify our tool, further empirical studies will be needed. Acknowledgements. This project has been supported by the following project: University Science Park of the University of Zilina (ITMS: 26220220184) and Research Centre of the University of Zilina (ITMS: 26220220183) supported by the “Research & Development Operational Program funded by the European Regional Development Fund”.

Validation Tools in Research to Increase the Potential

199

References 1. Alvarez, C.: Lean Customer Development: Building Products Your Customers Will Buy. O’Reilly Media Inc., Sebastopol (2014). 240 p., ISBN 1449356613 2. Barrett, M., Velu, C.H., Kohli, R., Salge, T.O., Simoes-Brown, D.: Making the transition to collaborative innovation: issues of readiness, trust and governance. Nesta, May 2011. Business Briefing 3. Blank, S., Dorf, B.: The Startup Owner’s Manual: The Step-by-Step Guide for Building a Great Company. BookBaby, USA (2012). 608 p., ISBN 9780989200547 4. Bremser, W.G., Barsky, N.P.: Utilizing the balanced scorecard for R&D performance measurement. R&D Manag. 34(3), 229–238 (2004) 5. Coleman, D.: 42 Rules for Successful Collaboration (2nd Edition): A Practical Approach to Working with People, Processes and Technology. Happy About, USA (2013). 158 p., ISBN 1607731118 6. Gakkai, N.K.: Innovation and Management: International Comparisons. Walter de Gruyter, Berlin (1988). 371 p., ISBN 3110110075 7. García-Valderrama, T., Mulero-Mendigorri, E., Revuelta-Bordoy, D.: A balanced scorecard framework for R&D. Eur. J. Innov. Manag. 11(2), 241–281 (2008) 8. Chesbrough, H., Crowther, A.K.: Beyond high tech: early adopters of open innovation in other industries. R&D Manag. 36, 3 (2006) 9. Jain, R.K., Triandis, H.C., Weick, C.W.: Managing Research, Development, And Innovation: Managing the Unmanageable. Wiley, Hoboken (2010). 416 p., ISBN 978-0-470-40412-6 10. Kaplan, R.S., Norton, D.P.: The balanced scorecard – measures that drive performance. Harvard Bus. Rev. 70(1), 71–79 (1992) 11. Kaplan, R.S., Norton, D.P.: Using the balanced scorecard as a strategic management system. Harvard Bus. Rev. 74(1), 75–85 (1996) 12. Kaplan, R.S., Norton, D.P.: Having trouble with your strategy? Then map it. Harvard Bus. Rev. 78(5), 167–176 (2000) 13. Kerssens-van Drongelen, I.C., Cook, A.: Design principles for the development of measurement systems for research and development processes. R&D Management 27(4), 345–359 (1997) 14. Kerssens-van Drongelen, I.C., Bilderbeek, A.: R&D performance measurement: more than choosing a set of metrics. R&D Manag. 29(1), 35–46 (1999) 15. Marr, B., Schiuma, G., Neely, A.: The dynamics of value creation: mapping your intellectual performance drivers. J. Intellect. Cap. 5(2), 312–325 (2004) 16. Montoya-Weis, M.M., Calantone, R.: Determinants of new product performance: a review and meta-analysis. J. Prod. Innov. Manag. 11(5), 397–417 (1994) 17. Pearson, A.W., Nixon, W., Kerssens-van Drongelen, I.C.: R&D as a business – what are the implications for performance measurement? R&D Manag. 30(4), 355–364 (2000) 18. Rogers, E.M.: Diffusion of Innovations, 5th edn. Free Press, New York (2003). ISBN 0-7432-5823-1 19. Sachsenmeier, P., Schottenloher, M.: Challenges Between Competition and Collaboration: The Future of the European Manufacturing Industry. Springer Science & Business Media, Berlin (2013). 326 p., ISBN 3662051958 20. Zánická Hollá, K., Ristvej, J., Šimák, L.: Systematic method of risk assessment in industrial processes. In: Risk Analysis VII: Simulation and Hazard Mitigation WIT Transaction on Information and Communication Technologies, v. 43. WIT Press, Southampton (2010)

Knowledge and Organization

Global Studies About the Corporate Social Responsibility (CSR) Sahar Mansour ✉ (

)

British University in Dubai (BUiD), Dubai, United Arab Emirates [email protected]

Abstract. At present, corporate social responsibility (CSR) has become an imperative topic of modern firms’ research theories, and the enhancement of CSR practices globally is remarkable. However the association among CSR and the other factors is still ambiguous. This research reviewed the development of notion of CSR practices among the Western, Asian, Middle East and African countries. According to the recent literatures, the results of studies show that the CSR situa‐ tion in Middle East countries is very optimistic. Keywords: Corporate social responsibility (CSR) · Western studies · Asian studies · Middle east studies · African studies

1

Introduction

Corporate social responsibility (CSR) is a form of firm self-regulation incorporated into a business model. CSR policy functions as a self-regulatory tool by which a firm ensures and monitors its dynamic commitment with ethical norms, spirit of the law and the global or local standards. The notion “corporate social responsibility” became popular in the sixties of the last century and has remained a notion adopted randomly by a lot to cover moral and legal responsibility more narrowly interpreted. Business Dictionary defines CSR as “A firm’s sense of responsibility towards the environment and community (both social and ecological) in which it works. Firms express this issue through their (1) reduction processes of pollution and waste, (2) contributing social and educational programs, (3) earning sufficient returns on the used raw materials and by (4) volunteering and philanthropy actions. “Unfortunately, very few studies in the world have been carried out to examine the Asian, Middle East and African studies about the CSR. It is worth noting that the main objective of the current study is to compare between the Western, Asian, Middle East and African studies about the CSR The study also devoted to find out whether CSR has any influence on improving the competitive advantage of firms or not. For this purpose, it is necessary to examine if it is worth to carry out CSR as a strategic step toward the leading business world, and also to examine the level of CSR of firms and the extent of influence it could have on the competitive value of a firms.

© Springer International Publishing AG 2017 L. Uden et al. (Eds.): KMO 2017, CCIS 731, pp. 203–213, 2017. DOI: 10.1007/978-3-319-62698-7_18

204

2

S. Mansour

Western World Studies About CSR

[13] claimed that reading the available literature review shows that most of researches cited use US or developed European samples. [24] made a serious of study about the relationship between CSR and FP specially in Poland and generally in EU countries and he opined that the CSR level depends on the compliance of CSR factors by most of firm’s stakeholders, “The higher the degree of conformity of the CSR factors indicated by key stakeholders, the higher level of the firm’s CSR”. He found out that if Polish partners want to build mutual, long-run, effective and responsible relations, they also have to put into consideration the mutual understanding of the factors of CSR. [14] conduct a very valuable and interesting study about the relationship between CSR and FP, his purposes was to generate the most comprehensive sample, for the purpose of having the best representation of the most companies. [25] stressed the importance of having a big sample “if it is not of an acceptable size, the risk is to end up with an insignificant sample that can negatively affect the whole study”. [14] obtained the data from a sample of 322 American firms belongs to the Fortune 500 magazine and clas‐ sified by their gross revenues. The logic behind [14] sample choices is that, he was a student and so he has limited access to the data; and by choosing a population consist of the most popular U.S. firms it will be much easier to gather both financial and social information. However, the number of sample has been reduced from being 500 firms to be 322 firms, as some important information was missing. In addition to that, gath‐ ering the social data was not as easy as gathering the financial data: and for the purpose of avoiding complications in the process of data collection, [14] decided that selecting an economic important and well-known companies like the firms in the Fortune 500, will help him to get the required information without any problems. [3] analysed the relationship between CSR variables and corporate financial performance in the Baltic States of Estonia Lithuania and Latvia. The CSR variables were including (community, environment, market place and workplace). The research used the content analysis to gather the data and regression method to analyse the relationship between CSR and FP. [3] pretend that the CSR initiatives do not have any effect on FP in Baltic, however the reason behind that as he said is that the concept of CSR has not yet become familiar in the Baltic, therefore, firms are not yet ready to pay more for products or services deliv‐ ered by CSR firms. [51] also examined the relationship between CSR and firm FP. The CSR variables used was (environment, community, creditors, suppliers, customers, employees and shareholders). The author based on stakeholder theory and triple bottom line principle to extract the mentioned CSR variables, he also adopted the quantitative method to carry out the empirical study which was based on samples of 95 US listed corporations. SPSS application was used to investigate the correlation between the CSR variables and FP. The regression analysis was adopted to determine the relationship between CSR and FP. The research proved that there is a considerable positive shortrun relationship among CSR and FP for employees variable, however, there is and a considerable negative short-run relationship among CSR and FP for community vari‐ able. [35] tried to analyse the relationship between CSR and FP that the decision makers can implement to form a unique strategy that can maximise profits. If decision makers are interested in employing CSR activities, the research predicts how the FP of

Global Studies About the Corporate Social Responsibility (CSR)

205

the firms will be influenced by CSR and suggested some strategies decision makers can use to meet their goals. The research sample includes 333 corporations included in the S&P 500 for four years starting from 2001. [10] have chosen to examine the relation‐ ship between CSR and FP from a different angle for the purpose of providing an alter‐ nate viewpoint on this theme: his main question was “does CSR play a role during periods of uncertainty?”, The data were obtained in a sample of 44 firms of French listed firms over the period of the financial crises which is 2007-2009. The findings indicate a remarkable positive relationship among CSR and FP for the period of 2008 financial crises. [48] also investigate the relationship between CSR and FP by asking this question: “All else equal, do more socially responsible firms have better corporate financial performance than less socially responsible firms?”. The study revealed that there is a moderate statistically significant positive association between CSR and FP by using many control variables like industry, firm size, and year. The study used the CSR rankings of US firms in Fortune Magazine’s for 3 years in 2005 as an inde‐ pendent variable. However, for dependent variables the study used an accounting and market measures from firm SEC Form 10-K’s as reported by Standard & Poor’s. [39] claimed that the literature of CSR researches related to the aviation industry is very little, so that he has investigated the relationship between CSR and FP but in aviation industry which is UK Manchester Airport. The researcher considers that the problem‐ atic measurement of CSR is the lack of a common framework in which society-busi‐ ness relation will be involved. The author examined the relationship between CSR and FP through peer multiples (or valuation multiples) a methodology extensively used in financial study. The FP of a company was measured via its annual account reports, where data about costs, earnings, investments and growth are gathered. [36] claimed that many firms used the reputation-building context to justify social areas, and as claimed by [44] it may mediate the association between CSR and FP, this is why [31] made his research about the influence of CSR on firm’s reputation, in addition to that there is scant study about the CSR and firm’s reputation, without a conclusive find‐ ings. The data were gathered from two main databases. The CSR data were gathered from KLD database which is considering the best tool available to measure the CSR of U.S. companies, while the FP data were gathered from the Thomson Database. The author has made his research by using 40 peer-reviewed other researches, covering different academic areas (including sociology, management, economics and finance). The sample was consisting of 809 companies, and the study has implemented 7 inde‐ pendent variables including (product quality, employee relations, human rights, natural environment, diversity, corporate governance, and community), while the dependent variable was the FP approximated by ROA. The author concluded that CSR should be a distinctive strategy for the companies. Therefore, if there is a stable difference between industries with regard to CSR, there is big chance to implement CSR disclo‐ sure. In other words, for those interested in implementing CSR should remember the amount of differences represented by industry effects. [9] shed light on the current aggregation approaches and explored a new methodology based on Data Envelopment Analysis (DEA) to measure the CSR disclosure. The researcher claimed that DEA provides an efficient indicator to measure the CSR disclosure and is independent of subjective weight specifications. The data were obtained in a sample of 2,190

206

S. Mansour

companies in 3 main sectors (Service, Finance, and Manufacturing) which is extracted from KLD database in 2007.

3

Asian Western World Studies About CSR

Studies in the area of CSR initiatives started during seventies of the previous century. The earliest study was carried out in developed countries by [11] who carried out a chain of questionnaires to measure the level of CSR initiatives in the annual reports of US Fortune 500 firms. On the other hand, [42] was the first to carry out a research on CSR in an emerging country like India. [24] studied the basic factors influencing CSR of the Japanese firms operated in Poland country for the purpose of facilitating any future research in the Polish – Japanese relations. He has defined four kinds of determinants of CSR in Japan including Legal determinants, Economical & Historical determinants, social & cultural determinants and finally religious and philosophical determinants. [26] made a deep research about the relationship between CSR & FP of on 90 financial institutions from 13 countries by using the cross-sectional analysis, but he did not use certain closed countries to each other which from the same area or even the same conti‐ nent. Some of the sample countries was from Europe like UK, some other countries was from the middle east like UAE, Saudi Arabia, Qatar, Kuwait, Bahrain, Syria and Jordan, and some other countries was from east Asia like Pakistan, Malaysia, Indonesia and Bangladesh, and one country was from Africa which was Sudan. It was a strange collec‐ tion of countries but there were some common factors like they all were an Islamic banks. [50] made a research about the relationship between CSR and FP in China for the period 2008–2009. The research proved that the FP is positively liked to CSR variables and the CSR variables have a positive and fundamental effect on the corporate FP. [46] claimed that the influence of social and environmental performance on FP has not been examined in any of Indian researches, for this purpose he made a research about the CSR initiatives towards its stakeholders and explaining its triple bottom line advantages and he tried to fill the existing gap by extending previous findings on (S&P) ESG12 500 listed Indian firms from various sectors. The research addresses fundamental Indian CSR initiatives and outlines their significance in shaping the relationship between CSR and FP. The research also examined the stakeholder relationship, financial strength and competi‐ tiveness variables of socially responsible corporations using perceptual data. However the limitations of the study that it does not fully examine all the fundamental conceptual hubs of CSR as corporate governance and other issues, but only explains its significance in formulation the CSR scope. [7] examined the impact of CSR spending on the FP of Indian corporations but he used a very small sample size. [28] also attempted to explore the impact of CSR towards the core stakeholders and its implications in nonfinancial and financial performance of Indian corporations using perceptual data. [29] examined the relationship between CSR and FP of Indian corporations by using Market Value Added and Economic Value Added; however, the data were obtained in a sample of 50 Indian corporations. [5, 43] examined the relationship between CSR and corporate governance factor of Indian corporations. [47] examined the relationship between CSR and shareholder returns by comparing its Governance, Social and Environmental

Global Studies About the Corporate Social Responsibility (CSR)

207

portfolio with Market portfolio of Indian corporations. [45] suggested that corporation profits are controlled by how corporations involve in CSR. He claimed that recent study on the relationship between CSR and FP suggested many organisational and contextual variables to generate a more robust association, but few of these researches examine the role of the CSR spending. [45] included many theories and perspectives in his study including path dependence theory, asset mass efficiencies, time compression disecono‐ mies and absorptive capacity theory, and based on these theories he concluded that when a corporation engaged slowly in CSR then the FP will be improved. [45] used longitu‐ dinal data gathered from 130 corporations for twelve years starting from 1995, and he used the triple bottom line including the Environmental, Social and Governance varia‐ bles from database provided by Morgan Stanley Capital International (MSCI) which is using American companies. [45] claimed that regardless of CSR variables used, a company can choose the proper strategy to enhance its FP. [38] analysed the promotion of CSR and he adopted content analysis as the method basis and the stakeholder theory as the theoretical basis. The data were obtained in a sample of 839 Chinese listed firms in 2010; then use regression analysis to identify the relationship among Chinese firms’ CSR and the FP. The findings revealed that the CSR practice in China is still needs a lot of works to be successful. According to the study findings out, adopting CSR to employees has also remarkable positive influence, and adopting CSR to shareholders will result in remarkable positive influence on Chinese firms’ FP; however adopting of CSR to other stakeholders has no remarkable influence on Chinese firms’ FP. [49] also study the relationship between CSR and FP in China; he used a content analysis meth‐ odology is applied over 800 listed corporations on the Shanghai Stock Exchange for two years starting from 2008. The results of the analysis imply that CSR is positively affected by corporation shareholding, share ownership concentration, media exposure and firm size. Moreover, firms in environmentally based industry tend to involve more environ‐ mental responsibility data than the other firms. The study findings support the legitimacy theory in an emerging market with a strong focuses on firm’s decision makers who implement the content analysis as a technique to legitimize their corporate social behavior. [34] analysed the relationship between CSR and FP in different ways. First, he examined the wealth-protective impact of CSR by analysing its relationship with equity risk for a sample. The findings from S&P 500 firms revealed that CSR is nega‐ tively associated with corporation risk, however, CSR is positively associated and with financial risk. Second, he analysed the firm’s bond market and find out that there is a negative association between CSR and firm spreads as well as credit risk. [30] analysed the relationship between CSR and FP as well as innovation on a sample size of 241 firms for two years in 2007. The data were obtained in a sample of 25 different countries including: Asia, Mediterranean, Scandinavian, Western Europe, North America and Anglo Sax. About 50% of the firms of the sample were from North America (Canada and USA); the second region by number of firms was Western Europe. The data was analysed by regression analysis and the findings indicate that there is no role of inno‐ vation on the relationship of CSR and FP, and revealed that there is no ambiguity or statistical significance on relationships between CSR and FP. Inoue & Lee (2011) inves‐ tigated the relationship between CSR and FP but in tourism-related industries. The Author tried to disaggregate CSR into five components based on firm voluntary

208

S. Mansour

initiatives for five main stakeholder concerns: (1) diversity concerns, (2) environmental concerns, (3) community relations, (4) product quality, and (5) employee relations, and analyse how each component would influence the FP of the company within 4 industries of tourism-related including (1) restaurant, (2) hotel, (3) casino, and (4) airline. Although that all CSR components proved that it has a positive financial impact, the findings reflected that each CSR component has a different impact on both future profitability and short-run and that such financial effect diversified across the four tourism-related industries. The author gathered the data from two main sources: (1) COMPUSTAT database, and (2) KLD database, for the period from 1991 to 2007. However, the total number of companies involved in the study is 183 for the restaurant, 51 for the hotel, 59 for the casino, and 74 for the airline. The research implement a multiple regression method to analyse the impacts of 5 CSR components on both the market’s evaluation measured by Tobin’s, and profits measured by ROA. On the other hand, [19] investigated the relationship between CSR and FP but from so many different countries from different continents including 42 countries such as Singapore, Hong Kong, Japan, Australia, European countries, UK, and USA. The study covers 7 years starting on 2002 and includes about 13 thousands observations. The independent CSR variable used was corporate governance, environment and social. The findings of the study shed light on the supply and demand forces that identify CSR globally.

4

Middle East Western World Studies About CSR

In spite of the fact that the study regarding CSR disclosure covered both emerging developed countries have obviously maximized over the last 30 years, only very few of this kind of study has been carried out in the Middle East and Arab nations. [1, 2, 15, 27, 33, 37]. On the other hand, [21] carried out a research over 9 of the Middle East countries, including Syria, Jordan, Egypt, Saudi Arabia, Kuwait, Bahrain, UAE, Oman, and Qatar. The research analysed the result by using the content analysis. The sample of population was 68 firms and the research was focused on 3 determinants about the CSR dubbed; social, economic and environmental dimension. The research indicated that, the initiatives about customer, community and employee issues were the common between the Middle East countries while the initiatives about the environmental issues got the lowest score initiatives compared to social and economic. [40] examined the relationship among CSR and FP between 280 firms from different sectors which operated in UAE by using the survey approach. The research measured the company’s FP by analysing 3 determinants dubbed; firm reputation, employee commitment and financial performance. The research revealed that, there is a positive relationship among the CSR and company’s FP. [17] carried out his research only over one of the GCC country called Qatar, the researchers made a list of 44 points about CSR. The research analysed the data by using multiple regression method. The research sample was 42 annual reports of the firms listed on Qatar Security Market. The research indicated that the complexity of the firm, assets, age and size have been the most important factors that encourage the firms to disclose about CSR initiatives. [22] examined 403 firms in UAE by using a survey of 12 items about the consciousness of these firms on the disclosure of CSR. The

Global Studies About the Corporate Social Responsibility (CSR)

209

researchers developed 4 determinants about CSR named community issues, consumer protection and environmental issues. The examined firms were multinational firms which came from all over the world like Intel, DHL and Shall companies. The research pointed out that, the UAE firms are cognizant about the CSR disclosures and the level of its importance. [41] have chosen a developing Asian country which is Iran to conduct his study. He has choose Iran for many reasons (1) the big gap among expected and actual level of CSR between Iranian companies; (2) As Iranian firms are underused sample among global studies; and (3) the lack of research on relationship among CSR and FP in Iran. [23] made his study about Iran about the relationship between CSR and FP of restaurants, airlines and hotels in Iran, and he examine whether these firms has positive or negative influence on FP of the firms. However, his results proposed different results among different fields and will participate to firms’ proper decision-making for CSR initiatives.

5

African Western World Studies About CSR

[32] carried out a study over all listed public firms of the Nairobi Securities Exchange, and due to a small size of population as it was only 10 firms to be studied, he used a census survey. He claimed that he has selected the Nairobi Securities Exchange as it provides a comprehensive and accessible listing of firms in Kenya. [12] has examined the (financial and non-financial) variables of CSR of Nigerian listed Deposit Money Banks for six years on 2005. The sample was consisting of thirteen banks and the exam‐ ined variables including (organisation leverage and growth, organisational ownership, dividend, organisation size, proxied by Economic profit). The research used the content analysis and analysed the data by using multiple regression method. The results revealed that the CSR variables are significantly strongly and positively influencing the financial performance of Nigerian listed Deposit Money banks. [4] has made a very interesting study about the financial and non-financial of CSR determinants in Jordan. He used a content analysis of a sample of 60 industrial firms listed on the Amman Stock Exchange of Jordan for a period 4 years in 2006. The study results revealed that firms that are highly leveraged, maintaining growth and are expected to be large in age and size are more likely to involve CSR data. [16] made a distinguished research about the relation‐ ship between CSR and FP by providing the first research of its type carried out in Egypt as an example of a developing country by using a sample of 111 Egyptian listed firms covers 13 sectors for the period of five years starting from 2005. The results of the analysis imply that 66% of the Egyptian listed firms involved on average 10–50 CSR disclosures. Moreover, the research find out that product/customer data is used widely by Egyptian listed firms when comparing with other types of CSR data. The other finding was that profitability is the main disclosure for the CSR data in Egypt. The regressions analyses are run to determine the relationship between CSR and FP measured by ROI. [6] made a significant contribution to the CSR initiatives literature by offering the first study of its type undertaken in Libya. The CSR initiatives measured by the annual reports of Libyan firms, while the FP measured by firm size, industry type and firm age. In his research he used a mixed method of both qualitative and quantitative. The data were

210

S. Mansour

obtained in a sample of 40 Libyan firms for the period of two years in 2007 and the hypotheses was tested by using regression analysis. The qualitative results revealed a positive relationship between all factors influencing CSR used in Libyan firms. However, the quantitative results revealed that there is a positive relationship between CSR initiatives and firm’s age and industry type as independent variables. On the other hand, [20] claimed that implementing CSR is a feeder to firm’s reputation. In other words, it encourages ethical behavior of decision makers, and this has a positive influ‐ ence on corporate reputation. The author tried to refute the opponents of CSR who claimed that it is costly and incompatible with the outstanding financial objective of maximizing shareholder profits. For this purpose the author presented the case about the relationship between CSR and FP but in Nigeria’s Niger Delta region. The author adopted many theories in his study like stakeholder theory, shareholder value theory and corporate social performance theory, and he claimed that implementing CSR in the company will lead to maximise shareholder return.

6

Conclusion

It is obvious that the Western world studies about the CSR is very advanced and devel‐ oped, and it was introduced also very long time ago; however, it is a new field to be studied and examined in the Arabic, Asian and African Countries; especially that in the African countries, as we barely found some studies about the CSR field there. Imple‐ menting CSR in a society needs many factors to be available like government support, conscious management, good budget, professional entrepreneurs and long patience, which are very difficult factors to be available in all societies. CSR in Arab countries still acknowledged as a new topic, the relationship between corporate social responsi‐ bility (CSR) and the firm’s financial performance (FP) is important topic for researchers.

References 1. Abu-Baker, N., Naser, K.: Empirical evidence on corporate social disclosure (CSD) practices in Jordan. Int. J. Commer. Manage. 10(3/4), 18–34 (2000) 2. Ahmad, N.S., Gao, S.S.: Corporate environmental reporting in Libya: a study of absence. Soc. Environ. Acc. J. 25(5), 11–14 (2005) 3. Aile, S., Bausys, Z.: Corporate social responsibility and financial performance: the example of Estonia, Latvia, Lithuania. Master thesis. The Stockholm school of Economics. MSc in Business & Economics (2013) 4. Al-Shubiri, F.N., Al-Abedallat, A.Z., Abu-Orabi, M.M.: Financial and non financial determinants of CSR. J. Knowl. Manage. Econ. Inf. Technol. 2, 7 (2012) 5. Banerjee, A., Gokarn, S., Pattanayak, M., Sinha, S.K.: Corporate governance and market value: preliminary evidence from indian companies. Stand. Poor 7, 166–178 (2009) 6. Bayoud, N.S., Kavanagh, M., Slaughter, G.: Factors influencing levels of corporate social responsibility disclosure by libyan firms: a mixed study. Int. J. Econ. Finance 4(4), 13–29 (2012) 7. Bedi, H.S.: Financial Performance and Social Responsibility: Indian Scenario. Working Paper (2009)

Global Studies About the Corporate Social Responsibility (CSR)

211

8. Business dictionary (2016). http://www.businessdictionary.com/definition/governance.html. Accessed Jan 2017 9. Chen, C.: Measuring corporate social performance: an efficiency perspective. Ph.D. thesis, UCLA Institute of the Environment, University of California, Los Angeles (2010) 10. Ducassy, I.: Does corporate social responsibility pay off in times of crisis? An alternate perspective on the relationship between financial and corporate social performance. Corp. Soc. Responsib. Environ. Manage. 20, 157–167 (2013) 11. Ernst & Ernst: Social Responsibility Disclosure: 1978 Survey, Ernst and Ernst, Cleveland (1978) 12. Farouk, M.A., Hassan, S.U.: Determinants of corporate social responsibilities in the nigerian listed deposit money banks. Int. J. Econ. Bus. Finance 1(10), 342–351 (2013). ISSN: 2327-8188. http://ijebf.com/ 13. Galbreath, J., Shum, P.: Do customer satisfaction and reputation mediate the CSR–FP Link? Evidence from Australia. Aust. J. Manage. 37(2), 211–229 (2012) 14. Ghelli, C.: CSR and financial performance: an empirical evidence. MSc Finance & Strategic Management. Ph.D. thesis. Copenhagen Business School (2013). http://hdl.handle.net/ 10417/3597 15. Hanafi, R., Gray, R.: Collecting social accounting data in developing countries: a cautionary tale from Egypt. Soc. Environ. Acc. J. 25(1), 15–20 (2005) 16. Hussainey, K., Elsayed, M., Razik, M.A.: Factors affecting corporate social responsibility disclosure in Egypt (2011) 17. Hossain, M., Hammami, H.: Voluntary disclosure in the annual reports of an emerging country: the case of Qatar (2009) 18. Inoue, Y., Lee, S.: Effects of different dimensions of corporate social responsibility on corporate financial performance in tourism-related industries. Tour. Manag. 32, 790–804 (2011) 19. Ioannou, I., Serafeim, G.: What Drives Corporate Social Performance? International Evidence from Social, Environmental and Governance Scores. Harvard Business School, Working Paper no. 11-016 (2010) 20. Iwu-Egwuonwu, R.C.: Does Corporate Social Responsibility (CSR) Impact on Firm Performance? A Literature Evidence. Social Science Research Network (SSRN) (2010) 21. Kamla, R.: Critically appreciating social accounting and reporting in the Arab Middle East: a postcolonial perspective. Adv. Int. Acc. 20, 105–177 (2007) 22. Katsioloudes, M.I., Brodtkorb, T.: Corporate social responsibility: an exploratory study in the United Arab Emirates. SAM Adv. Manage. J. 72(4), 9 (2007) 23. Khanifar, H., Nazari, K., Emami, M., Soltani, H.A.: Impacts corporate social responsibility activities on company financial performance. Interdisc. J. Contemp. Res. Bus. 3(9), 545–553 (2012) 24. Krukowska, M.: Determinants of CSR in Japanese companies. Manage. Bus. Adm. Cent. Eur. 22(4(127)), 39–57 (2014) 25. Malhotra, N.K.: Marketing Research: An Applied Orientation. Pearson International Edition (2007) 26. Mallin, C., Farag, H., Ow-Yong, K.: CSR and financial performance in Islamic banks. J. Econ. Behav. Organ. 103, S21–S38 (2014) 27. Mikdashi, T., Leal, W.: Experiences on corporate social responsibility (CSR) implementation in Lebanon: a causal recursive system. Int. J. Environ. Sustain. Dev. 4(2), 181–192 (2005) 28. Mishra, S., Suar, D.: Does corporate social responsibility influence firm performance of indian companies? J. Bus. Ethics 95, 571–601 (2010)

212

S. Mansour

29. Mittal, R.K., Sinha, N., Singh, A.: Current research development – an analysis of linkage between economic value added and corporate social responsibility. Manag. Decis. 46(9), 1437–1443 (2008) 30. Montoya, M.M.: Innovation, corporate social responsibility and financial performance. Master thesis. M.Sc. in Economics. Faculty of Economics and Business Administration. Tilburg University. ANR:951467 (2011) 31. Moura-Leite, R.C.: The significance of corporate social performance of organizational effectiveness. Ph.D. thesis, Universidad de Salamanca, facultad de economia de la empresa, departmento de administracion y economia de la empresa (2010) 32. Mwangi, C.I., Jerotich, O.J.: The relationship between corporate social responsibility practices and financial performance of firms in the manufacturing, construction and allied sector of the nairobi securities exchange. Int. J. Bus. Humanit. Technol. 3(2), 81 (2013) 33. Naser, K., Al-Hussaini, A., Al-Kwari, D., Nuseibeh, R.: Determinants of corporate social disclosure in developing countries: the case of Qatar. Adv. Int. Acc. 19, 1–23 (2006) 34. Oikonomou, I.: Empirical investigations of the relationship between corporate social and financial performance. Ph.D. Thesis at Henley Business School, University of Reading (2011) 35. Palmer, H.J.: Corporate Social Responsibility and Financial Performance: Does it Pay to Be Good? CMC Senior Theses. Paper 529 (2012) 36. Porter, M.E., Kramer, M.R.: The link between competitive advantage and corporate social responsibility. Harv. Bus. Rev. 84, 1–15 (2006) 37. Pratten, J.D., Mashat, A.A.: Corporate social disclosure in Libya. Soc. Responsib. J. 5(3), 311–327 (2009) 38. Qiu, Y.: Empirical Study between CSR and Financial Performance of Chinese Listed Companies. Master Thesis in Business Administration no.: 2012:MF06 at University of Boras (2012) 39. Rapti, E., Medda F.: Corporate Social Responsibility and Financial Performance in the Airport Industry. University College London (2010) 40. Rettab, B., Brik, A.B., Mellahi, K.: A study of management perceptions of the impact of corporate social responsibility on organisational performance in emerging economies: the case of Dubai. J. Bus. Ethics 89(3), 371–390 (2009) 41. Saeidi, S.P., Sofian, S., Saeidi, P., Saeidi, S.P., Saaeidi, S.A.: How does corporate social responsibility contribute to firm financial performance? The mediating role of competitive advantage, reputation, and customer satisfaction. J. Bus. Res. 68, 341–350 (2015) 42. Singh, D.R., Ahuja, J.M.: Corporate social reporting in India. Int. J. Acc. 18(2), 151–169 (1983) 43. Singhania, M.: Corporate governance and financial performance in India: an empirical study. Soc. Responsib. Rev. 4, 44–64 (2011) 44. Surroca, J., Tribo, J.A., Waddock, S.: Corporate responsibility and financial performance: the role of intangible resource. Strateg. Manage. J. (2009). doi:10.1002/smj.828 45. Tang, Z., Hull, C.E., Rothenberg, S.: How corporate social responsibility engagement strategy moderates the CSR–financial performance relationship. J. Manage. Stud. 49(7), 1274 (2012) 46. Tyagi, R.: Impact of CSR on Financial performance and competitiveness of business: a study of Indian firms. Ph.D. thesis of Department of Management study of Technology Roorkee. 247-667 (India) (2012) 47. Vasal, V.: Corporate social responsibility & shareholder returns – evidence from the Indian capital market. Indian J. Ind. Relat. 44, 376–385 (2009). (Special Issue on Corporate Social Responsibility Guest Editor – Mrityunjay Athreya)

Global Studies About the Corporate Social Responsibility (CSR)

213

48. Vollono, R.: Doing well by doing good: the empirical relationship between corporate social responsibility and financial performance. Master thesis, Faculty of School of Arts and Sciences of Georgetown University (2010) 49. Yao, S., Wang, J., Song, L.: Determinants of CSR disclosure by Chinese firms. Discussion Paper no. 72 at University of Nottingham (2011) 50. Yin, W.H.: CSR and Firm performance: evidence from China. Master thesis in Philosophy in Business (Accountancy) at Lingnan University (2012) 51. Zhang, X., Gu, P.: On the relationship between CSR and financial performance: an empirical study of US firms. Master thesis within Business Administration. Jönköational Business School, Jönköping University (2012)

Enhancing Work Engagement Towards Performance Improvement Chris W.L. Chu1(&), Reuben Mondejar2, Akira Kamoshida3, and Zahir Ahamed3 1

The Surrey Business School, University of Surrey, Guildford, UK [email protected] 2 City University of Hong Kong, Kowloon Tong, Hong Kong [email protected] 3 Yokohama City University, Yokohama, Japan [email protected], [email protected]

Abstract. Underpinned by role expansionist perspective, this study examined why and how family-to-work enhancement is related to work engagement leading to an improvement in task and contextual performance. Controlling for family-to-work conflict and job autonomy, the results revealed that consistent with our predictions, work engagement mediated the relationships for family-to-work enhancement with task and contextual performance. Moderated mediation analyses further revealed that work engagement mediated the relationships for (a) task performance for only those supervisors who were supportive to their subordinates’ work-family issues; and (b) contextual performance, regardless of supervisor’s level of support. Results highlighted the importance of supportive supervisors or family-friendly work environment when examining the relationships between enhancement, engagement, and job performance. Keywords: Work engagement
Family-to-Work enhancement performance
Contextual performance




Task

Family-to-work enhancement occurs when family resources can be used to benefit work domain. Recent studies have consistently shown that social support from the family (i.e. family-to-work enhancement) could generate better work performance (Friedman and Greenhaus 2000; Frone 2003; Frone et al. 1997; Graves et al. 2007). The field, however, may require more research into the theories postulating which mediators of relationships between family-to-work enhancements and work outcomes are significant or relevant. Underpinned by expansionist perspective, this study conceptualized family-to-work enhancement as a resource gain and examined (i) the mediating influence of work engagement in the relationship of family-to-work enhancement and the two dimensions of job performance (i.e. task and contextual performance), and (ii) the moderating influence of supervisor support on the relationship of work engagement and the two dimensions of job performance in a sample of employed parents in Ghana. Accordingly, our study sought to fill in the missing gaps in the extant literature by proposing and examining how family-to-work enhancement relates to the two © Springer International Publishing AG 2017 L. Uden et al. (Eds.): KMO 2017, CCIS 731, pp. 214–227, 2017. DOI: 10.1007/978-3-319-62698-7_19

Enhancing Work Engagement Towards Performance Improvement

215

dimensions of job performance, task and contextual performance, through work-engagement after controlling for family-to-work conflict and job autonomy. By pursuing these objectives, our study contributes to the extant literature in three ways. First, most of the existing studies predominantly focus on the influence from work domain (e.g. job characteristics) on work engagement (Christian et al. 2011; Hakanen et al. 2006; Liao et al. 2012; Maun et al. 2007; Rich et al. 2010; Salanova et al. 2005; Schaufeli and Bakker 2004). Examining the antecedents of work engagement from a non-work domain not only enhance our existing knowledge to the extent which employee’s work engagement can be improved by non-work context, but also expand the implication of motivation theoretical perspective in the field (Kahn 1990; Rich et al. 2010). Second, consistent with conservation of resources (COR) theory (Hofboll 1989), our study provides insights on how one’s family involvement can create additional resources to enhance his/her work performance through the mediating influence of engagement. This study provides support for work engagement as a mediator of the relations between distal factor and work outcomes, in which an individual’s engagement behaviors could be enhanced by factors (which in this particular study is individual’s positive family experience), other than individual and organization factors (cf. Macey and Schneider 2008; Rich and Lepine 2010). These results contribute to researchers’ understanding of the expansionary paradigm of role involvement on work-family interface (Marks 1977; Ruderman et al. 2002; Sieber 1974) as well as the extended implication of job characteristics theory (Hackman and Oldham 1976). Lastly, our focus on examining task and contextual performance simultaneously in this study as the outcomes of engagement not only addresses an unanswered issue raised by Christian and colleagues (2011), but also provides probably the first empirical finding to support or reaffirm that the concept of engagement should be related to both in-role and extra-role behaviors. This is particularly important given research evidence suggesting that engagement is a unique concept that it conceptually differs from traditional job attitudes which are usually viewed to enhance one’s wellbeing such as organizational commitment and job satisfaction (cf. Christian et al. 2011).

1 Theoretical Framework and Literature Review The conceptual heritage of work-family enhancement can be traced to multiple roles theory originally suggested by Sieber (1974) and Marks (1977), in which it posits that individuals taking up several life roles can accrue benefits (e.g. role privileges and status security) which lead to greater role gratification rather than role stress. Committed individuals find abundant energy in fulfilling those roles as opposed to stress when the commitment is lacking thereby suggesting that strength and energy may be generated in individuals talking up multiple roles (Marks 1977). Kahn (1990: 700) defined work engagement as “the simultaneous employment and expression of a person’s ‘preferred self’ in task behaviors that promote connections to work and to others, personal presence (physical, cognitive, and emotional) and active, full performances”. To account for the mediating influence of work-engagement between the linkage of family-to-work enhancement and the two dimensions of job performance (i.e. task and

216

C.W.L. Chu et al.

contextual performance), we posit that the experience of family-to-work enhancement could improve one’s positive emotional response which in turn enhances one’s work engagement (Rothbard 2001). We finally posit that supervisor support constitutes an important physical, emotional or psychological resource in the work environment which could intensify the positive influence of work engagement on job performance. Thomas and Ganster (1995) defined supervisor support as the actions, attitudes or behaviors of a supervisor towards the employee which may affect the balance between work and family responsibilities. We argue that a high level of supervisor support creates a supportive work climate that enriches employee’s work experiences, which in turn fosters favorable job performance. 1.1

The Mediating Influence of Work Engagement

As explained above, family-to-work enhancement is expected to give more work resources to an individual. Hakanen et al. (2006: 497) defined work resources as assets that “(1) reduce job demands and the associated physiological and psychological costs, (2) are functional in achieving work goals, and (2) stimulate personal growth, learning, and development”. The idea to posit that family-to-work enhancement enhances one’s engagement at work is based in part on both role expansionist perspective (Marks 1977; Sieber 1974) and conservation of resources (Hofboll 1989). The expansionist perspective suggests that family-to-work enhancement will enable one to generate resources from one’s family life which can be used in one’s work life. When one receives more work resources, one is likely to “retain, protect, and build resources” (Hobfoll 1989: 516) in work domain leading to better work engagement. This argument can be understood in two ways: (i) resources compel individuals to engage in work domain and they reinforce continued engagement because individuals seek to acquire and maintain valued resources (Hobfoll 1998), and (ii) resources which set family-to-work enhancement into motion by influencing individuals’ work engagement are likely to be greater for individuals in a resource-rich environment than for comparable individuals with comparable levels of engagement in a resource-poor environment (Hobfoll 1998). Therefore, it is believed that family-to-work enhancement is related to one’s work engagement. Research has shown that one can generate resources from family domain to benefit work domain, thereby creating family-to-work enhancement (Rothbard 2001; Ruderman et al. 2002). For example, dealing with family issues make employed women proactively plan for their schedule at work, thereby promoting family-to-work enhancement through learning and improving their planning skills acquired from the family domain (Ruderman et al. 2002); one’s family experiences could generate positive feelings which can be brought over and make an impact on one’s work experiences (Graves et al. 2007; Grzywacz and Marks 2000). Therefore, while obtaining more work resources through family-to-work enhancement, according to Hofboll’s (1989, 1998) conservation of resources theory, one should engage more in the work domain to prevent the loss of these work resources. Supporting our contention, Demerouti et al. (2001) and Salanova et al. (2005) found that the availability of work resources can predict one’s work engagement.

Enhancing Work Engagement Towards Performance Improvement

217

We also posit that work engagement is related to task and contextual performance. First, we posit that work engagement is related to task performance. Task performance is defined as an in-role performance which reflects how an individual perform his/her assigned duties at the workplace (Borman and Motowidlo 1997; Christian et al. 2011). Because the engaged individuals exhibit high levels of energy, persistence and focus at the workplace, they are likely to spend more time to deal with work-related issues (Macey and Schneider 2008), leading to higher levels of task performance. In support of our contention, Aryee et al. (2012) found that work engagement enhances task performance indirectly through innovative behavior. Second, we posit that work engagement is related to contextual performance. Witt and colleagues (2002) defined contextual performance as “outcomes of behaviors that are needed to support the social fabric of the organization” (Witt et al. 2002: 911). These behaviors included individual’s willingness to “follow rules, persist, volunteer, help, and cooperate” (Witt et al. 2002: 912) through which it helps to accomplish organizational goals as well as improve overall firm performance. This construct has been noted (Van Scotter and Motowidlo 1996) to comprise the dimensions of interpersonal facilitation (cooperation, consideration, and helpful acts that assist co-worker’ performance) and job dedication (self-disciplined, motivated acts such as working hard, taking initiative). Kahn (1990) suggested that the engaged individuals are likely to go beyond their job description to facilitate other coworkers or organization (cf. Borman and Motowidlo 1993). For this reason, we expect that the engaged individuals who are in general more energetic and more willing to invest discretionary effort at their workplace, leading to an increase of contextual performance. In support of our contention, Halbesleben et al. (2009) found that work engagement is positively related to OCB. Specifically, based on role expansionist perspective and COR theory, we expect that family-to-work enhancement will provide more resources in the form of work-related resources to enhance work engagement, which in turn, improves job performance. Hypothesis 1. Family-to-work enhancement will be positively related to work engagement. Hypothesis 2a. Work engagement will be positively related to task performance. Hypothesis 2b. Work engagement will be positively related to contextual performance. Hypothesis 3a. Work engagement will mediate the relationship between family-towork enhancement and task performance. Hypothesis 3b. Work engagement will mediate the relationship between family-towork enhancement and contextual performance.

1.2

Moderating Influence of Supervisor Support

According to COR theory, we conceptualized supervisor support as visible sources of work resources available to employees to healthy balance between their work and family responsibilities. This kind of support (e.g. accommodating employee’s flexible work schedule) allows employees to have more control or to be more flexible in

218

C.W.L. Chu et al.

juggling in the responsibilities of both work and family roles. Therefore, we believe that a supportive supervisor will foster an environment which allows the engaged individuals to feel less stressed to deal with their work and family responsibilities, so that they will be more focused or behave more positively at work leading to higher levels of job performance. This argument can be explained with reference to job characteristics theory (Hackman and Oldham 1976) which has been modified recently by Humphrey et al. (2007). They incorporated motivational, social, and contextual characteristics in the job characteristics model to explain how work environment from these three dimensions motivates employees to be more engaged at work (Humphrey et al. 2007). Consistent with this view, we argue that engaged employees will have more job autonomy (motivational characteristics) or assistance and advice from supervisor (social characteristics) to deal with their work tasks in a high supervisor support environment. These employees not only have more work-related resources, but also are more willing to invest energy and time at their work roles. Thus, we believe that engaged employees in a high supervisor support environment will perform better at their job tasks (i.e. task performance) and they will also be more willing to assist others at work (i.e. contextual performance). In support of the preceding argument, research has demonstrated social support to be related to work engagement (Kahn 1990; Macey and Schneider 2008) which indicates its potential role in amplifying the influence of work engagement on the two dimensions of job performance examined in this study (i.e. task and contextual performance). On the basis of the above discussion, we suggest the following hypothesis: Hypothesis 4a. Supervisor support will moderate the relationship between work engagement and task performance such that high levels of supervisor support will expand the positive effects of work engagement on task performance. Hypothesis 4b. Supervisor support will moderate the relationship between work engagement and contextual performance such that high levels of supervisor support will expand the positive effects of work engagement on contextual performance. Hypothesis 5a. The indirect relationships of family-to-work enhancement with task performance will be stronger when supervisor exhibits high (vs. low) supportive behavior. Hypothesis 5b. The indirect relationships of family-to-work enhancement with contextual performance will be stronger when supervisor exhibits high (vs. low) supportive behavior.

2 Methods 2.1

Sample and Procedure

Three hospitals in Ghana participated in the study. Of the 350 questionnaires distributed, 206 were returned. However, after deletion of cases with respondents who were not parents, the final sample was 169 representing a response rate of 48.3%. Of the 169 respondents, 50.9% were female, reported working an average of 44 h (s.d. = 8.76) a week, and an average of 19 h (s.d. = 23.91) spent on domestic

Enhancing Work Engagement Towards Performance Improvement

219

responsibilities a week. In terms of educational attainment, 33.7% had an undergraduate degree, 33.1% had taken senior secondary school or below, and 28.4% had a postgraduate degree or professional qualification. The average age reported by respondents in this study is 41.

2.2

Measures

All the measures used in this study are the validated scale in the literature. The reliabilities of these scales are above 7.

3 Data Analysis We first ascertained the distinctiveness of the variables in the study by conducting a confirmatory factor analysis (CFA) using LISREL 8.80 (Jöreskog and Sörbom 2006). We compared the fit of our hypothesized model to nested alternative models and a one-factor model. Next, we examined the following conditions: (a) significant mediating influence of work engagement (i.e. significant indirect effect); (b) significant moderating influence of supervisor support; and (c) different conditional indirect effect of family-to-work enhancement on task and contextual performance via work engagement in which the indirect effect differs in strength across low and high levels of supervisor support.

3.1

Results

Table 1 presents the results of the CFA that examined the distinctiveness of the study variables. As shown in the table, the fit indices revealed that our hypothesized 5-factor model fit the data better than each of the alternative models suggesting support for the distinctiveness of our variables. Descriptive statistics and zero-order correlations among the study variables are presented in Table 2. Family-to-work enhancement was positively related to work engagement (r = .21, p < .01) while work engagement was positively related to task performance (r = .50, p < .01) and contextual performance (r = .54, p < .01). Table 3 presents the results of the regression analysis that examined the mediating influence of work engagement on the relationship between family-to-work enhancement and the two dimensions of job performance (i.e. task and contextual performance). As shown in the table (Model 1), family-to-work enhancement was related to work engagement (b = .18, p < .01) satisfying the first condition for mediation (Kenny et al. 1998). The regression results in Model 2 and Model 3 indicated that work engagement was related to task performance (b = .67, p < .001) and contextual performance (b = .49, p < .001) respectively. When considering the mediating influence of work engagement, family-to-work enhancement was no longer related to task performance (b = .10, p = n.s.) in Model 2 but continued to be related to contextual performance albeit at a reduced magnitude (b = .14, p < .05) in Model 3. Following Kenny et al. (1998), the pattern of these results suggested that work engagement fully

220

C.W.L. Chu et al. Table 1. Confirmatory Factor Analysis (CFA) results of measurement model

Model Hypothesized 5-factor model 4-factor model A (combining task performance and contextual performance) 4-factor model B (combining work engagement and task performance) 4-factor model C (combining work engagement and contextual performance) 3-factor model (combining work engagement, task performance and contextual performance) 2-factor model (combining self reports) 1-factor model

△v2

△df

____

____

470.44***

1601.52 (813)

v2 (df) 1233.23 (809) 1703.67 (813)

RMSEA 0.06

SRMR 0.07

NNFI 0.92

CFI 0.93

4

0.08

0.09

0.88

0.89

368.29***

4

0.08

0.08

0.88

0.89

1451.81 (813)

218.58***

4

0.07

0.08

0.90

0.91

1805.53 (816)

572.3***

7

0.09

0.09

0.86

0.87

2671.68 (818)

1438.45***

9

0.12

0.11

0.78

0.79

2711.94 (819)

1478.71***

10

0.12

0.11

0.77

0.78

mediated the relationship between family-to-work enhancement and task performance but partially mediated the relationship between family-to-work enhancement and contextual performance. We subsequently used Sobel’s (1982) test for indirect effects (MacKinnon et al. 2002). Results showed that the indirect effects of family-to-work enhancement on task performance (Sobel = 2.58, p < .01) and of family-to-work enhancement on contextual performance (Sobel = 2.64, p < .01) are both significant. Taken together, our data provided support for Hypotheses 1, 2a, 2b, 3a and 3b. The results of moderated regression analysis that examined Hypothesis 4a and 4b are presented in Tables 4 and 5. As shown in Table 4, the two-way interaction of work-engagement and supervisor support significantly predicted task performance

Enhancing Work Engagement Towards Performance Improvement

221

Table 2. Means, standard deviations and zero-order correlations (N = 169) Variable

Mean SD

1

1. Gender 2. Age 3. Tenure 4. Family-to-work conflict 5. Job autonomy 6. Family-to-work enhancement 7. Supervisor support 8. Work engagement 9. Task performance 10 Contextual performance

____

____

____

2

3

41.25 9.65 −.13 ____ 14.45 10.54 −.04 .84** ____ 2.05 .78 −.04 −.08 −.10

4

5

7

8

9

.95 .64

−.07 −.02 −.02 .15

____ −.08 .14 .18* −.17* .09

2.80

.92

−.07 .10

.06

3.79 3.39 3.33

.53 .78 .50

.00 .03 .04

−.07 −.04 −.05 .01 .08 −.03

.07

____

.23** .10 .17* .11 .04

____

.21** .19* ____ .16* .46** .50** ____ .29** .15# .54** .41**

Table 3. Results of regression analysis for mediation Variable:

Controls: Gender Age Tenure Family-to-work conflict Job autonomy R2 Direct effects: Family-to-work enhancement △R2 Mediating effects: Family-to-work enhancement Work engagement △R2 Overall R2 Overall model F *p < .05; **p < .01; ***p
0 denote the monetary investment in information security to protect the given information set, measured in the same units used to measure the potential loss L. The purpose of the investment z is to lower the probability that the information set will be breached. Let S(z, v) denote the probability that an information set with vulnerability v will be breached, conditional on the realization of a threat and given that the firm has made an information security investment of z to protect that information. The expected benefits of an investment in information security, denoted as EBIS, are equal to the reduction in the firm’s expected loss attributable to the extra security: EBIS(z) = [v − S(z, v)]L = 𝜆[vt − S(z, v)t]

Matsuura [38] noted that the information security investment z can reduce the threat probability and that the reduction depends only on the investment z and the current level of threat probability t . So let T(z, t) denote the probability that a threat occurring, given that the firm has made an investment of z. So, in his extended model: EBIS(z) = 𝜆[vt − S(z, v)T(z, t)]

(3)

Equation (3) can be used as a basis for quantitative comparison of risks of various alternatives. Obviously, the monetary loss 𝜆 is independent of the choice of type of sourcing. Then the problem of alternative comparison is reduced to the comparison of the product of probabilities vj tj. Here tj - the threat probability and vj - vulnerability for alternative j. Necessary data can be obtained from monitoring systems or based on expert evaluations. Lower value of vj tj corresponds to the more attractive alternative. Therefore, to go to the maximization problem, we consider the value (1 − vj tj ).

4

Example

Thus, to compare alternatives of sourcing of IT services, we will use AHP with following criteria: • The reciprocal cost of ownership in form of 1∕NPVj, measured in inverted monetary units (i.e., 1/dollars); • the reciprocal expected average time of change 1∕𝜏j, measured in inverted time units (i.e., 1/days); ( ) • and the product of the probabilities of threats and vulnerabilities in form 1 − vj tj ,

0 ≤ vj tj ≤ 1.

282

Y. Zelenkov

Consider the example of the use of these criteria. Suppose that some company consider three alternatives for the development of its IT services: • Continued use of its own IT infrastructure (Alternative 1); • Migration of all IT services to the public cloud (Alternative 2); • Migration of only non-critical IT services to a public cloud (Alternative 3). Quantitative values of criteria calculated for all alternatives are shown in Table 1. Table 1. Values of alternatives. Criterion

NPVj (million dollars)

𝜏j (days) vj tj (probability)

Alternative1 Alternative2 Alternative3

0,5 0,2 0,4

3 2 3

0,20 0,30 0,22

An important step of AHP is to determine the priorities of the criteria. Fundamental scale that reflects the relative strength of preferences and feelings of company [5] is used for it. Suppose that the specialists of company made a pairwise comparison of the importance of criteria and rated them as shown in Table 2. The value in the cell of the table corresponds to relative importance of criterion in the row regarding criterion in the column. Note, if criterion i has number assigned to it when compared with criterion j, then j has the reciprocal value when compared with i. Table 2. Matrix of pairwise comparisons. NPVj

𝜏j

vj tj

Weight

NPVj

1

2

1/3

0,294

𝜏j

1/2

1

1/2

0,176

vj tj

3

2

1

0,529

To obtain the criteria weights, eigenvector that corresponds to largest eigenvalue of matrix of pairwise comparisons should be calculated. Its values are presented in right column of Table 2. The normalized reciprocal values of the alternatives for each criterion are presented in Table 3. The same table shows the integral value (fitness function - fj) of each alter‐ ∑n native, which is obtained as a weighted sum of the partial evaluations fj = i=1 wi xij, here wi - weight of criterion, xij - value of the j-th alternative for i -th criterion, n - number of criteria. Table 3. Results of alternatives comparison. Criterion

1∕NPVj

1∕𝜏j

(1 − vj tj ) Integral value

Weight Alternative 1 Alternative 2 Alternative 3

0,294 0,211 0,526 0,263

0,176 0,286 0,429 0,286

0,529 0,388 0,259 0,353

0,318 0,367 0,314

Decision Making in Cloud Computing

283

Data in Table 3 shows that in accordance with given company’s priorities Alternative 2 is the most effective.

5

Conclusion

The main goal of paper is to propose simple model that can be used in practice. Three criteria (costs of ownerships, intangible benefits that associated with speed of reaction to change and security risks) that have been proposed here are enough simple and all necessary data can be obtained from accounting system, contract conditions, statistics and expert opinions. AHP method that is used for comparison of different sourcing alternatives does not require complex calculations. All of this leads to the conclusion that the proposed method can be used in practice.

References 1. Mell, P., Grance, T.: The NIST Definition of Cloud Computing (Draft): Recommendations of the National Institute of Standards and Technology. http://csrc.nist.gov/publications/ nistpubs/800-145/SP800-145.pdf 2. Marston, S., Li, Z., Bandyopadhyay, S., Zhang, J., Ghalsasi, A.: Cloud computing; the business perspective. Decis. Support Syst. 51(1), 176–189 (2011) 3. Armbrust, M., Fox, A., Griffith, R., Joseph, A.D., Katz, R., Konwinski, A.: A view of cloud computing. Commun. ACM 53(4), 50–58 (2010) 4. Karunakaran, S., Krishnaswamy, V., Sundarraj, R.P.: Business view of cloud: decisions, models and opportunities – a classification and review of research. Manag. Res. Rev. 38(6), 582–604 (2015) 5. Venters, W., Whitley, E.A.: A critical review of cloud computing: researching desires and realities. J. Inf. Technol. 27(3), 179–197 (2012) 6. Saaty, T.L.: Fundamentals of Decision Making and Priority Theory with the Analytic Hierarchy Process. RWS Publications, Pittsburgh (2002) 7. Schneider, S., Sunyaev, A.: Determinant factors of cloud-sourcing decisions: reflecting on the IT outsourcing literature in the era of cloud computing. J. Inf. Technol. 31(1), 1–31 (2016) 8. Yang, H., Tate, M.: A descriptive literature review and classification of cloud computing research. Commun. Assoc. Inf. Syst. 31(1), 2 (2012) 9. Benlian, A., Hess, T.: Opportunities and risks of software-as-a-service: findings from a survey of IT executives. Decis. Support Syst. 52(1), 232–246 (2011) 10. Tak, B.C., Urgaonkar, B., Sivasubramaniam, A.: To move or not to move: the economics of cloud computing. In: Proceedings of the 3rd USENIX Conference on Hot Topics in Cloud Computing, p. 5. USENIX Association (2011) 11. Naldi, M., Mastroeni, L.: Economic decision criteria for the migration to cloud storage. Eur. J. Inf. Syst. 25(1), 16–28 (2016) 12. Khajeh-Hosseini, A., Greenwood, D., Smith, J.W., Sommerville, I.: The cloud adoption toolkit: supporting cloud adoption decisions in the enterprise. Softw. Pract. Exp. 42(4), 447– 465 (2012) 13. Williams, B.: The Economics of Cloud Computing. Cisco Press, San Jose (2011) 14. Misra, S.C., Mondal, A.: Identification of a company’s suitability for the adoption of cloud computing and modelling its corresponding return on investment. Math. Comput. Model. 53(3), 504–521 (2011)

284

Y. Zelenkov

15. Marešová, P.: Cost benefit analysis approach for cloud computing. In: Sulaiman, H.A., Othman, M.A., Othman, M.F.I., Rahim, Y.A., Pee, N.C. (eds.) Advanced Computer and Communication Engineering Technology. LNEE, vol. 362, pp. 913–923. Springer, Cham (2016). doi:10.1007/978-3-319-24584-3_77 16. Garg, S.K., Versteeg, S., Buyya, R.: A framework for ranking of cloud computing services. Future Gener. Comput. Syst. 29(4), 1012–1023 (2013) 17. Sundarraj, R.P., Venkatraman, S.: On integrating an IS success model and multicriteria preference analysis into a system for cloud-computing investment decisions. In: Kamiński, B., Kersten, G.E., Szapiro, T. (eds.) GDN 2015. LNBIP, vol. 218, pp. 357–368. Springer, Cham (2015). doi:10.1007/978-3-319-19515-5_28 18. Delone, W.H., McLean, E.R.: The Delone and Mclean model of information systems success: a ten-year update. J. Manag. Inf. Syst. 19(4), 9–30 (2003) 19. Takabi, H., Joshi, J.B., Ahn, G.J.: Security and privacy challenges in cloud computing environments. IEEE Secur. Priv. 6, 24–31 (2010) 20. Catteddu, D., Hogben, G.: Cloud Computing: Benefits, Risks and Recommendations for Information Security, ENISA (2009). www.enisa.europa.eu/act/rm/files/deliverables/cloudcomputing-risk-assessment/at_download/fullReport 21. Subashini, S., Kavitha, V.: A survey on security issues in service delivery models of cloud computing. J. Netw. Comput. Appl. 34(1), 1–11 (2011) 22. Hashizume, K., Rosado, D.G., Fernández-Medina, E., Fernandez, E.B.: An analysis of security issues for cloud computing. J. Internet Serv. Appl. 4(1), 1–13 (2013) 23. Angeles, S.: 8 Reasons to Fear Cloud Computing. Business News Daily (2013). http:// www.businessnewsdaily.com/5215-dangers-cloud-computing.html 24. Martens, B., Teuteberg, F.: Decision-Making in cloud computing environments: a cost and risk based approach. Inf. Syst. Front. 14(4), 871–893 (2012) 25. Kantarcioglu, M., Bensoussan, A., Hoe, S.: Impact of security risks on cloud computing adoption. In: Proceedings of the 49th Annual Allerton Conference on Communication, Control, and Computing, pp. 670–674. IEEE (2011) 26. Saripalli, P., Pingali, G.: MADMAC: multiple attribute decision methodology for adoption of clouds. In: 2011 IEEE International Conference on Cloud Computing, pp. 316–323. IEEE (2011) 27. Savage, S.L.: The Flaw of Averages: Why We Underestimate Risk in the Face of Uncertainty. Wiley, New York (2009) 28. Sarker, S., Munson, C., Chakraborty, S.: Assessing the relative contribution of the facets of agility to distributed systems development success: an analytic hierarchy process approach. Eur. J. Inf. Syst. 18(4), 285–299 (2009) 29. Jadhav, A., Sonar, R.: Evaluating and selecting software packages: a review. Inf. Softw. Technol. 51(3), 555–563 (2009) 30. Benlian, A.: Is traditional, open-source, or on-demand first choice? Developing an AHP-based framework for the comparison of different software models in office suites selection. Eur. J. Inf. Syst. 20(5), 542–559 (2011) 31. Bonham, S.S.: Actionable Strategies Through Integrated Performance, Process, Project, and Risk Management. Artech House, Boston (2008) 32. Zelenkov, Y.: Business and IT alignment in turbulent business environment. In: Abramowicz, W. (ed.) BIS 2015. LNBIP, vol. 228, pp. 101–112. Springer, Cham (2015). doi: 10.1007/978-3-319-26762-3_10 33. Ciborra, C.: The Labyrinths of Information: Challenging the Wisdom of System. Oxford University Press, Oxford (2002)

Decision Making in Cloud Computing

285

34. Maurer, C., Goodhue, D.: A Theoretical model of the enterprise system agility life cycle. In: AMCIS 2010 Proceedings, paper 231 (2010) 35. Luftman, J., Kempaiah, R.: An update on business-IT alignment: “A Line” has been drawn. MIS Quarterly Exec. 6(3), 165–177 (2007) 36. Murer, S., Bonati, B., Furrer, F.J.: Managed Evolution: A Strategy for Very Large Information Systems. Springer, Heidelberg (2011) 37. Gordon, L.A., Loeb, M.P.: The economics of information security investment. ACM Trans. Inf. Syst. Secur. 5(4), 438–457 (2002) 38. Matsuura, K.: Productivity space of information security in an extension of the Gordon-Loeb’s investment model. In: Johnson, M.E. (ed.) Managing Information Risk and the Economics of Security, pp. 99–119. Springer, New York (2009)

Study on Network Online Monitoring Based on Information Integrated Decision System Fan Yang ✉ and Zhenghong Dong (

)

Academy of Equipment, No. 1 Bayi Road, Huairou District, Beijing 101416, China [email protected], [email protected]

Abstract. Information integrated decision system (IIDS) integrates multiple sub-system developed by many facilities, including almost hundred kinds of soft‐ ware, and provides with various services, such as email, short messages, drawing and sharing, etc., which also supports two types of network, WLAN and Wi-Fi. Because of the underlayer protocols are different, user standards are not unified, and many errors are happened during the stages of setup, configuration, and oper‐ ation, which seriously affect the usage. Because the errors are various, which maybe happened in different operation phases, stages, TCP/IP communication protocol layers, and sub-system software. It is necessary to design a network online monitoring tool for IIDS. In order to solve the above problems, this paper studies on network online monitoring based on IIDS, which provides strong theory and technology supports for the operating and communicating of IIDS. Keywords: Online monitoring system · Network diagnosis · Information flow collection

1

Introduction

Modern high technique work relies on high efficient and real time information services that are working in order, the system should be running stably and reliably, no matter for the servers in headquarters, or the PDAs at mobile termination. IIDS integrates multiple sub-system developed by many facilities, including almost hundred kinds of software, and provides with various services, such as email, short messages, drawing and sharing. This requires the system to work in a stable and efficient way. However, IIDS has the following problems: When parent unit assigns a mission, the transmission status of the mission operation information is not clear, and lack of monitoring and feedback for the status of the whole system information operation, which makes the operators cannot understand if the mission has been sent to subordinate units, and also have no idea about the implemen‐ tation status of subordinate units, so they cannot use IIDS to carry out mission assign‐ ment, distribution, planning, and operation, etc. System requirement analysis is shown as Fig. 1. From the viewpoints of mission operation, system of system, information, and physics, the mappings between mission assignment, mission planning, network information system, information stream and mission results evaluation have many unclear relations, information transmission status © Springer International Publishing AG 2017 L. Uden et al. (Eds.): KMO 2017, CCIS 731, pp. 286–294, 2017. DOI: 10.1007/978-3-319-62698-7_24

Study on Network Online Monitoring

287

is unclear, which lacks of the tool of monitoring and feedback for network information operation status, and makes that the relation between mission flow relation and mission operation relation is unclear, so it is hard to carry out mission results evaluation.

Fig. 1. System requirement analysis

According to the above requirements, it is necessary to develop network online monitoring study focus on the communication problems of IIDS.

2

Related Work

Monitoring on network status is to monitor network flow and network data, and provides support for the resources distribution, capacity planning, service quality analysis, failure monitoring and isolation, security management of network [1]. Currently, the ways to monitor network status do not limit to using machines that can directly connect to WLAN, but also can use wireless sensor network [2] to collect data as well as commu‐ nication. This method is often applied to applications that cannot directly connect to WLAN, such as railway network. [1] proposed a design on IP network monitoring system, which is based on NetFlow monitoring tool, and has the advantages of low cost and easy to implement. However, this method only suits to the statistic of real-time network flow, but not for long time statistical analysis, which is often used to deal with abnormal network flow. Because of the quantity of network and communication machines is large, and those machines are laid to different places, online monitoring system based on distributed

288

F. Yang and Z. Dong

network is developed. Reference [3] constructed a machine long-distance failure diag‐ nosis integrated system, which implemented monitoring and data collection for moni‐ toring machines by intelligent collection machines, and provided unified monitoring, failure specialist diagnosis, long distance information management service, and imple‐ ment the largest resources sharing of status monitoring and failure diagnosis. [5] adopted multi-thread probe information monitoring technique based on ManagedAgent model, and Agent technique based on SNMP to collect and transfer loaded information and mission status information, which has the characteristics of low load, small disturbance, strong ageing, and high extensibility. In order to monitor network status, it is necessary to analysis network protocol, and obtain the applied network protocol, so that to collect and analysis network data. For example, ICMP (Internet control message protocol) [4], is an error connect oriented protocol, which is used to transfer error report control message. It is mainly used to transfer controlling messages between principal machine and router, including reporting errors, exchanging limitation control and status information, etc. The design of network status monitoring tools has to be based on specific protocol. [6] proposed a method using ICMP protocol to monitor network transport layer, which can help user to explain the network operation status and find out errors in the transport layer immediately. This paper relies on distributed data monitoring platform, which requires to collect system status data from multiple nodes from different places, gathers those collected data through data transmission network, and efficiently and accurately to analysis global status information.

3

Online Monitoring Program Design

3.1 Key Technique: Distributed Status Monitoring Technique Focus on the characteristics of cooperated using complicate information from different places, this paper proposes local distributed status monitoring technique, which relies on global operation status and error information collected from data transmission network nodes, and improves allsideness and accuracy of network diagnosis. Each distributed monitoring node can be setup to any IIDS node that requires moni‐ toring. This way can independently analysis and display, as well as provide information to parent unit, which can be freely setup and applied no matter from the aspect of distributed framework of system, distribution and collection of data, or from the aspect of synthesis and analysis of information. The special function of node dispatching and switching can make any distributed monitoring node to setup its layer, and sending messages to parent nodes, automatically revise the topology of each IIDS. 3.2 Study Objective This paper carries out technique research on network failure online monitoring method based on IIDS. The problem of IIDS is as follows, the transmission status of the mission operation information is not clear, and lack of monitoring and feedback for the whole

Study on Network Online Monitoring

289

system information operation status, which makes the operators cannot understand if the mission has been sent to subordinate units, and also have no idea about the imple‐ mentation status of subordinate units, so they cannot use IIDS to carry out mission assignment, distribution, planning, and operation, etc. The general study framework of network online monitoring tool of IIDS is shown as Fig. 2. The main research objective is to solve the above problem and construct an online monitoring system to achieve the transparency of mission operation in each unit, and accuracy of technique maintenance. In order to improve the ability of mission plan‐ ning, decision, and operation based on information system, and provide strong theory and technology supports for the running and communicating of IIDS.

Fig. 2. General study framework

3.3 Online Monitoring System Design Online monitoring system is to monitor and test IIDS and its machine operation status. Real-time non-stopping monitoring can make sure the system continuously and stably running, so that operators can make correct decision and prediction, and find out prob‐ lems in time. The main contents of monitoring includes, general operation status of system, operation status of the interface of each sub-system and each machine. 3.3.1 System General Structure In order to make full use of the functions of IIDS, and solve the problems, such as communication, information sharing, information security, and information monitoring, etc., this paper proposes to construct an online monitoring tool to monitor the operation states and system errors of IIDS. The main function is to monitor external information exchanging status. According to the missions, to monitor the flow, state, type, time, and

290

F. Yang and Z. Dong

path of information flow, and summarize important information, such as the exchanging, forwarding, and location of information flow. This tool can also provide key data for operators to design mission flow, and working nodes. The online monitoring system adopts independent monitoring deployment. When the IP address resources are plenty, we can adopt the deployment way of independent monitoring. And the advantage is that the monitoring system would not affect IIDS, which can deduce the compatibility problem caused by system coupling.

Fig. 3. System general structure

The general topology is shown as Fig. 3. Online monitoring system would be deployed to each computer in the IIDS, the two sets of system are independent with each other to use the hardware platform, and share with the same data transmission network

Study on Network Online Monitoring

291

under the condition of not changing the network of IIDS. Online monitoring system monitors the information in the data transmission network, summarizes and analyzes this data, and sends the analyzed state information back to parent unit. The monitoring system in each level can only display the state information of its parent unit and its subordinate units that are directly connected with it. All the state information can be summarized in unit of the highest level, and all states of the whole IIDS can be checked in the unit with the highest level. 3.3.2 System Software Design Online monitoring system mainly includes, data online monitoring unit, data classifi‐ cation unit, data storage unit. Its structure is shown as Fig. 4. Data transmission network Monitoring all information in the network Data online monitoring unit Data classification unit

Mission information

IIDS status information

Information flow status information

Data storage unit

Real time data storage base

Fig. 4. Online monitoring system structure design

Its main working flow is as follows, based on the data in data transmission network, and relying on the types of information, first, to carry out online monitoring, second, to classify three types of information, which are mission, status of IIDS, state of informa‐ tion flow, and third, to put those different types of data to data storage base. The three types of information are explained as follows, (1) Mission information, is the information in IIDS to transfer mission information. System monitors data pack in different computers and communication ports, according to the difference of external interactive services. After analysis of data packs, important information can be retrieved, such as service type, sending IP, receiving IP, sending time, and receiving time, and save time to the real-time data storage base. (2) IIDS status information, is the external information exchanging service operating state information and data transmission state information in the data transmission network in each servers or clients. The online monitoring system has been deployed

292

F. Yang and Z. Dong

to each server and client in the IIDS, the external exchanging service operating state and data transmission state information can be summarized and sent to the server of parent unit. In the end, the general information of IIDS state can be generated and save to real time data storage base. (3) Information flow state information, is the information flow state information in the IIDS, which contains the sending data exchanging information between the IIDSs, time information of information flow, and transmission path information of infor‐ mation flow.

4

System Prototype Implementation

In order to guarantee the adaptability and reliability of online monitoring system in practice, this paper also implements a prototype of the proposed tool, which can monitor IIDS system status and information flow state in real time. Figure 5 shows IIDS system status, which mainly displays the external exchanging service operating state of IIDS, and the real time dynamic diagram of data transmission network connection state. IIDS system status diagram relies on automatically imported contacts and parameter configuration unit, based on real time data storage base, builds up a diagram that takes the current level node as the center, and contains its parent unit and all subordinate units.

Parent unit Status: Normal

Status: system turn off

Position 8

Position 9

Subordinate unit

Subordinate unit

Subordinate unit

Position 10

Position 7

Position 6

Position 5

Position 4

Position 3

Position 2

Status: link not connected, status unknow

Position 1

Status: service turn off

Data link normal

Data link not connected

Subordinate unit

Data link blocked

Fig. 5. System state of IIDS (Color figure online)

Figure 5 shows the operation state of a certain unit, the green color represents that all servers and clients operating normally, red color represents that this machine gone

Study on Network Online Monitoring

293

wrong. Yellow color represents that some required services of server or client gone wrong. Grey represents that the data link of subordinate unit gone wrong, which cannot retrieve the state information of this unit. (1) Fig. 6 shows information state diagram to display the exchanging of mission in real-time during IIDS. The representation of information state replies on the information flow data in the real time storage database, which shows the information exchanging, and transmission state of related computer. Parent unit Code：DY-1 State：Normal Code：DXX-1 Normal

Status: Normal

Code：DY-2 State: Info has been sent, not received

Status: system turn off

Position 3

Position 3

Subordinate unit

Subordinate unit

Subordinate unit

Position 3

Position 3

Position 3

Position 5

Position 4

Position 3

Position 2

Status: link not connected, status unknow

Position 1

Status: service turn off

Data link normal

Data link not connected

Subordinate unit

Data link blocked

Fig. 6. Information flow state diagram

Figure 6 displays the transmission status of all information flow, each flash point in the figure represents an information flow, the color is used to distinguish the information state. Click the flash point can show up the detail information of this information flow.

5

Conclusion

The online monitoring system monitors and storage the mission, information flow, and system operation state information in IIDS through Gigabit Ethernet. And solves many serious problems that are occurred during the joint test, interconnection of IIDS, including the information transmission state is not clear, and lack of monitoring and feedback of general information operation states.

294

F. Yang and Z. Dong

References 1. He, F., Jin, N.: Design and implementation of the IP network state monitor system based on netflow. Commun. Technol. 40(08), 34–36 (2007) 2. Xie, J., Wang, Y., Wu, M.: Contact network online monitoring technology based on internet of things. In: The 5th Equipment Communication Conference on China Railway Electrification Technology, pp. 220–224 (2013) 3. Hao, H.-J., Cheng, G.-H., Wang, M.-L.: Research of remote fault diagnosis expert system based on cloud computing. In: 2015 China High-Performance Calculation Annual Conference (2015) 4. Du, S.: Developing of network connecting status monitoring tools based on ICMP protocol. In: Computer Programming Skills and Maintenance, pp. 72–73 (2015) 5. Wang, D., Hu, M., Zhi, H., Liu, X.: Research On the distributed lightweight payload monitoring based On the large scale topology information collecting system. In: 2008 China Computer Network Security Emergency Annual Conference, pp. 291–296 (2008) 6. Wang, M.: Study on network status monitoring based on ICMP protocol. Sci. Technol. Inf. 31, 865–866 (2009)

Marine Geochemical Information Management Strategies and Semantic Mediation Tenglong Hong1, Xiaohong Wang2(&), Jianliang Xu1(&), Shijuan Yan3, and Chengfei Hou3 1

3

Ocean University of China, No. 238, Songling Road, Lao Shan District, Qingdao, China [email protected] 2 Shandong Provincial Key Laboratory of Computer Networks, Shandong Computer Science Center (National Supercomputer Center in Jinan), Jinan, China [email protected] The First Institute of Oceanography, State Oceanic Administration, No. 6 Xianxialing Road, Qingdao, China

Abstract. Oceanography is a comprehensive, interdisciplinary, cooperative global ocean science. Evolution of marine investigation technologies brings various and massive marine data, so its format, syntax and semantic are varied. Such heterogeneity is an obstacle to knowledge integration, transmission and sharing. A framework of E-oceanography is proposed for dealing with marine information with the background of big data era. Within this framework, semantic mediation system is developed for data interoperation in marine community. The greatest benefit of the semantic management system is to allow a growing number of marine knowledge to be accessed without semantic obstacle and interdisciplinary constrain. Keywords: Metadata


Oceanography
Ontology
Semantic mediation

1 Introduction Science and engineering increasingly rely on a rapidly evolving infrastructure of digital information and computing resources. Both the potential power and the challenges of cyber infrastructure lies in the integration of the many relevant and mostly disparate resources to provide a platform that can on the long term empower the modern scientific research endeavour [1]. Oceanography is no exception, oceanography study transfers from a single object to a diversified, interdisciplinary, inter-regional marine data discovery. Similar information have been illustrated in different manners and in diverse terminologies by scientists. This environment leads to many information management challenges as following. One of the major problems is that individual information source is constructed in various representation models. Multiple data sources are organized with different metadata standards, even in similar information description, there are rarely same metadata standards. If a global unified schema approach is directly used, other © Springer International Publishing AG 2017 L. Uden et al. (Eds.): KMO 2017, CCIS 731, pp. 295–306, 2017. DOI: 10.1007/978-3-319-62698-7_25

296

T. Hong et al.

representations may need transformation and modification. However, during the process of transformation and modification, some information on each conceptual structure might evitable be lost or added. Another important problem is that the same or similar concepts may be contained in heterogeneous information sources but expressed in different terminologies. In order to obtain sustainable interoperability, semantic mappings should be built between different schemas. Since semantic mappings excessively rely on domain knowledge and information context, there has been a great deal of research into semantic mediation. In this paper, we propose a framework of E-oceanography to achieve information management for overall lifecycle, additionally we design a semantic management system based on ontology to perform semantic mediation and terminology management. The remainder of the paper is organized as follows. In Sect. 2, successful instances of marine data management are discussed. In Sect. 3, the framework of E-oceanography we proposed is described in detail. Furthermore, semantic management system are discussed to achieve data interoperability. In Sect. 4, an experiment for marine geochemical data access is used to prove efficiency of our method. Section 5 is the conclusion and our future work.

2 Related Work For oceanographic data management of information lifecycle, there are a number of successful instances in advanced country. Their success is ascribed to the growing information infrastructure techniques for data integration, communication, and exchanging under the support of international projects shown as follows.

2.1

A Growing Information Infrastructure Technique for Data Exchange

With the advent of computer networking, information infrastructure, such as OGC (Open Geospatial Consortium [2]), SeaDataNet infrastructure [3], OPeNDAP (Open-source Project for a network Data Access Protocol [4]) as a component of E-oceanography to accommodate data exchange in data intensive science. One of the most successful instances in European is SeaDataNet, whose infrastructure links 90 national oceanographic data centers and marine data centers from 35 countries riparian to all European seas. The data center manage large sets of marine and ocean data, originating from their own institutes and from other parties in their country, in a variety of data management systems and configurations. SeaDataNet has developed an efficient distributed Marine Data Management Infrastructure for the management of large and diverse sets of data deriving from in-situ and remote observation of the seas and oceans. Professional data center, active in data collection, constitute a Pan-European network providing on-line integrated databases of standardized quality. The on-line access to in-situ data, meta-data and products is provided through a unique portal interconnecting the interoperable node platforms constituted by the SeaDataNet data center [3].

Marine Geochemical Information Management Strategies

2.2

297

Community-Specific Organizations for Data Integration and Access

The MMI (Marine Metadata Interoperability [5]) project, GALEON IE (Geo-interface for Atmosphere, Land, Earth, and Ocean netCDF interoperability Experiment [6]) and EarthChem (Geochemical Database for the Earth [7]) are classical data integration and access projects in oceanography. MMI was first funded by the National Science Foundation in 2004 [5]. The goal of MMI is to support collaborative research in the marine science domain, by simplifying the incredibly complex world of metadata into specific, straightforward guidance. MMI encourages scientists and data managers at all levels to apply good metadata practices from the beginning of a project, by providing the best guidance and resources for data management, and developing advanced metadata tools and services needed by the community. Today it continues to provide guidance, vocabularies and semantic services, with regular updates on events and news of interest to the community. There are 33 ontologies available in MMI project, such as Environment Ontology, SWEET Ontology, SWEET Units Ontology, Earth System Grid Ontology, Spire Ecological Concepts Ontology, Biodiversity Resource Information Ontology, Hydrology Units Ontology and so on [8]. EarthChem [7] which is maintained by Lamont Earth Observatory, is a communitydriven effort to facilitate the preservation, discovery, access and visualization of data generated in the geosciences, with particular emphasis on geochemical, geochronological, and petrological data. EarthChem was founded as a consortium of the igneous rock database in 2003 with the goal to nurture synergies among the databases, minimize duplication of effort, and share tools and approaches. Data systems hold a variety of measured values and descriptive metadata with the purpose of preservation, discovery, and reuse of geochemical datasets. Synthesis databases such as PetDB [9], SedDB [10], NAVDAT [11], and the Deep Lithosphere Dataset [12] provide access to large, thematically focused compilations of geochemical data, and allow users to generate customized subsets of the data, integrating analyses from any number of publications and datasets. The EarthChem Portal offers a “one-stop-shop” for geochemistry study of seafloor rocks and provides full range of “big data” support.

3 Strategies and Technologies New mode of oceanography research solves the problem of information management under the background of data-intensive science discovery. We call the new mode E-oceanography. The essence of E-oceanography is the interdisciplinary global oceanography which emphasizes data normalization, integration, optimizing search, archival, and dissemination features that encourages innovation in accessibility and interoperability [13]. E-oceanography can ensure long-term access of the scientific results.

3.1

A Framework of E-oceanography

We presents a framework of E-oceanography including data normalization, integration, quality control and provenance technologies, as a systematization of current

298

T. Hong et al.

Fig. 1. Framework of E-oceanography for information lifecycle

technology, is to accommodate data collections, transferring, integration, visualization and analysis within information lifecycle (Fig. 1 (a)) inspired by Michah Altman [13]. The framework of E-oceanography consists of management strategies, storage, communication and computing component, relies on the data normalization, integration, and data access technologies as show in Fig. 1(b). E-oceanography is viewed as a new paradigm for data-intensive marine scientific discovery. Both the potential power and the challenges of E-oceanography are developing a platform to integrate amount of relevant or disparate resource which shared by many institute, scientist. Many technical, disciplinary, organizational, political, and semantic barriers still need to be overcome before infrastructure will reach its full potential and fulfill its promises.

3.2

Semantic Mediation System of Marine Geochemical Data

Although a semantic mediation framework has been proposed by MMI in 2011. The framework of MMI promotes the data interoperation for marine community. But it seems as if there are some deficiencies such as the vocabulary construction is deeply relied on manual and to be time consuming and error prone. Additionally, semantic mappings have not been applied widely by the public as there are 244 vocabularies but merely 10 pairs of mappings between them [5]. So in this paper, we propose a semantic mediation system for marine geochemical data. The system is consist of Marine Sample Ontology, Metadata Extraction, Metadata Mapping and Data Querying as shown in Fig. 2. The details of the semantic mediation are introduced in the following text. The semantic mediation system provides a Marine Sample Ontology (MSO) and the function of metadata extraction and metadata mapping for marine geochemical datasets both in the spreadsheet and relational database.

Marine Geochemical Information Management Strategies

299

Fig. 2. Semantic mediation system for marine geochemical data

(1) Marine Sample Ontology We build Marine Sample Ontology (MSO) inspired by PROV Ontology [14] to provide a common understanding of the fields,. MSO plays a role in sharing and reusing the specific marine geochemical knowledge by identifying the marine geochemistry fields of different databases. The current full version of Marine Sample Ontology contains 154 classes and 35 properties. These attributes provide a rich representation of the oceanographic knowledge model so as to supports data query. (2) Metadata Extraction In order to extract the metadata from data providers’ datasets in spreadsheet, one need to upload their spreadsheet using upload button. Then a popup window is provided for data providers to input the start and end row number of column header as shown in Fig. 3. After that, the system extracts metadata from the column header. The system also provides the function for modification merge cells in the column header through dragging the cells. For instance, we can drag the “PGE (Platinum Group Elements)” cell to cover its subclasses which listed next row, including chemical elements “Ru”, “Rh”, “Os”, “Ir”, and “Pt” (as shown in Fig. 3). Meanwhile we develop a website to extract the schema of relational database (as shown in Fig. 4). After inputting the user name, password, and connect string, all the tables and related fields are displayed in the popup window. One can select relative table and fields for creation metadata mappings. (3) Metadata Mapping Websites of metadata mapping are developed for datasets both in spreadsheet and relational database. Metadata of datasets in the spreadsheet are listed in the left side of the website and hierarchies of marine sample ontology are listed in the right side (as shown in Figs. 5 and 6). The system provide the button “AutoMapper” to establish the semantic mappings rely on the “is-a” relationship in the marine sample ontology and identical spelling. The new terms which have not been included in the marine sample ontology still need human intervention. The semantic relation button ‘=’ helps semantic providers create equivalence semantic relations between terms of providers’ and marine sample ontology.

300

T. Hong et al.

Fig. 3. Popup window of Metadata Extraction

Fig. 4. Website of Metadata Extraction

(4) Data Query A data query website (as shown in Fig. 7) is developed based on the metadata mappings and marine sample ontology. The process of data query is shown as following. • Convert the user’s query condition to SPARQL query. • Generate new queries based on the inference ability of Jena inference mechanism and semantic mappings originated from metadata mapping (as shown in Fig. 8).

Marine Geochemical Information Management Strategies

Fig. 5. Website of Metadata Mapping for spreadsheet

Fig. 6. Website of Metadata Mapping for RDB

301

302

T. Hong et al.

Fig. 7. Website of data query

• Transfer SPARQL queries to SQL queries using D2RQ rely on a file originated from generate-mapping tool which is provided by D2RQ. • Use parallel technique to extract the data from diversity database (as shown in Fig. 9). If data user want to query rock’s iron content, SPARQL query is shown as following:

Semantic mappings are shown as following:

Marine Geochemical Information Management Strategies

The processes of query:

Fig. 8. The operation principle of data query

303

304

T. Hong et al.

Fig. 9. Use parallel technique to extract the data from diversity database

4 Experiment of Semantic Mediation In order to test the ability of semantic mediation, we run an experiment, we create three databases firstly. The first one is referred to the ICP-AES data model of the ODP (Ocean Drilling Program) database. The structures of the other two databases are consistent with Ridge PetDB and GEOROC. Next, we randomly select some geochemical test datasets and load them into the above three databases. Finally, we run the SPARQL query through the data query website. If users are interested in the FeT contents of rocks that originate from South West Indian Ocean area, our SPARQL query will also return the related data from above three databases. SPARQL Query:

Marine Geochemical Information Management Strategies

305

The response results including not only the Fe, but also the subclasses of Fe, such as “FeT”, “FeO”, and “Fe2O3” (shown in the following) Returned data from three database:

5 Conclusion This paper presents a framework of E-oceanography. What is more, a semantic mediation system of marine geochemical data is proposed for data integration and sharing. This is one of the earliest studies to propose the strategies both for E-oceanography and semantic mediation in the global interdisciplinary oceanography. Experimental results on some of existing marine geochemical datasets show that the approach provides one of possible solutions for the heterogeneity problem of diverse representation models, so as to increase the interoperability among different information sources.

306

T. Hong et al.

Acknowledgments. This work is supported by the National Nature Science Foundation of China (Grant No. 61672475), Chunhui Research Grant from the Ministry of National Education (Grant No. Z2016102) and the 12th Five-Year Plan of investigation and development for International marine resource - Construction of China Ocean Sample Repository and marine sample management (under grant number DY125-25-01).

References 1. Lehnert, K. Ratajeski, K., Walker, D.: Using Online geochemical database for research and teaching. short course tutorial (2008). http://earthchem.org/sites/earthchem.org/files/ Shortcourse-Tutorial_2008.pdf 2. Open Geospatial Consortium. http://www.opengeospatial.org/. Accessed 1 Jan 2017 3. SeaDataNet infrastructure. http://www.seadatanet.org/Overview. Accessed 1 Jan 2017 4. OPeNDAP. http://www.opendap.org/. Accessed 1 Jan 2017 5. MMI (Marine Metadata Initiative). https://marinemetadata.org/. Accessed 1 Jan 2017 6. Geo-interface for Atmosphere, Land, Earth, and Ocean netCDF interoperability Experiment. http://www.opengeospatial.org/projects/initiatives/galeonie 7. EarthChem (Geochemical Database for the Earth). http://www.earthchem.org/ 8. Yun, H.: Ontology Modeling of Marine Ecology from Device-Function Viewpoint. diss. Ocean University of China (2012). doi:10.7666/d.y2158805 9. Petrological Database. http://www.earthchem.org/petdb/. Accessed 1 Jan 2017 10. SedDB. http://www.earthchem.org/seddb. Accessed 1 Jan 2017 11. NAVDAT. http://www.navdat.org/. Accessed 1 Jan 2017 12. Deep Lithosphere Dataset. http://www.earthchem.org/deeplith. Accessed 1 Jan 2017 13. What is e-science and how should it be managed. http://www.scilogs.com/scientif-ic_and_ medical_libraries/what-is-e-science-and-how-should-it-be-managed/. Accessed 1 Jan 2017 14. Garijo, D.: PROV-O: The PROV Ontology Tutorial (2013). http://oa.upm.es/21511/1/ DC2013.pdf

Implementation of a Text Analysis Tool: Exploring Requirements, Success Factors and Model Fit Giorgio Ghezzi, Stephan Schlögl ✉ , and Reinhard Bernsteiner (

)

Department Management, Communication and IT, MCI Management Center Innsbruck, 6020 Innsbruck, Austria [email protected]

Abstract. This paper reports on lessons learned from implementing a text anal‐ ysis tool in an industrial setting. We conducted two rounds of focus group inter‐ views — one pre- and a second one post-implementation — and extended our analysis by a survey undertaken one month after the tool had gone live. This methodology let us explore and compare the suitability of three different tech‐ nology acceptance models. Findings show that the Technology Acceptance Model (TAM) fits as a general mathematical approach describing our tool’s acceptance, whereas the Hospitality Metaphor (HM) produces slightly more precise analytical results, explaining its adoption from a more holistic point of view. Finally, we found that the hybrid approach emphasized by the Unified Theory of Acceptance and Use of Technology (UTAUT) showed the most reliable and trustful results, as it combines both human and business/technology aspects. Keywords: System implementation · Text analysis · TAM · HM · UTAUT

1

Introduction

Business organizations constantly search for new ways of gaining advantage over their competitors [2, 13]. This concerns the entire business value chain ranging all the way from procurement to production and sales. While in the past businesses often aimed at boosting production without considering an exact demand prediction [17], today’s ongoing propagation of Enterprise Resource Planning (ERP), Production Planning (PP), Business Intelligence (BI) and other, similar analytical systems, allows for more efficient ways of steering business decisions [18]. All too often, however, these systems fail to meet the high expectations assigned to their implementation [14]. Moreover, a great number of implementations are late, over budget or simply not successful [1]. Based on a real case, our work thus aims at better understanding the implementation process of such a tool. We explore the pre-implementation as well as the post-implementation stage and highlight several challenges that had to be tackled and lessons that had to be learned. To guide to our research, we focused on the identification of key requirements (preimplementation stage) and success factors (post-implementation stage), and how to best explicate these aspects through different technology acceptance models.

© Springer International Publishing AG 2017 L. Uden et al. (Eds.): KMO 2017, CCIS 731, pp. 307–317, 2017. DOI: 10.1007/978-3-319-62698-7_26

308

G. Ghezzi et al.

While we do realize that similar studies conducted in different settings may produce deviating results, we strongly believe that many of the insights reported in this paper are conferrable to other cases and thus that our analysis is of significant relevance to the information systems community.

2

Study Setting and Research Question

Our team was asked to accompany the implementation of a Text Analysis System (TAS) based on SAP HANA; the goal of the proposed system being the support of the compa‐ ny’s quality management department in automating and improving their current complaint handling process. Pooling, combining, and linking vast amounts of data to produce relevant insights may be considered a key success factor in such a complaint handling process. Generic applications supporting this task are, however, barely avail‐ able and so the company considered building a proprietary solution. Understanding the perspective of the staff members whose work is directly linked to the existing data-flow and whose daily routines may thus be affected by a changing tool landscape, was considered a significant cornerstone supporting the implementation process – one whose influences may be even greater than those expected from various technical decisions. Consequently, the analysis we present in this paper treats the technical implementation as a secondary aspect and rather puts its focus on the people who are affected. To guide this exploration process, we followed previous studies of technology implementation and acceptance (cf. [3, 8, 19]), being particularly inspired by the Technology Acceptance Model (TAM) [6], the Hospitality Metaphor (HM) [4], and the Unified Theory of Acceptance and Use of Technology (UTAUT) [21]. The goal was to understand the changes such a new tool would introduce into daily routines and to highlight both posi‐ tive and negative aspects of acceptance. Based on the assumption that the correct imple‐ mentation of the system would lead to a relatively quick and straight forward adoption, we believed that changes should be easily identified [8]. Only those aspects that require additional learning would need more in-depth follow up analyses for them to be better understood [9]. Consequently, our work may be defined as an initial before-after eval‐ uation exploring the following research question: What are apparent challenges concerning the implementation of an industrial text analysis system and how do existing acceptance models such as TAM, HM and UTAUT compare in identifying them?

3

Task Setting and Problem Space

The technical task was to implement a text analysis tool based on SAP HANA. A dataset including customer complaints (i.e. customers who identified a defect with their product after its purchase) as well as complaints filed by affiliated firms (i.e. authorized shops and other resellers) served as a starting point. From there the goal was to create a dash‐ board application capable of providing structured information on the current complaint situation all the way down to the product level. It should highlight words, feelings and common problems associated with distinct products. Different types of defects should be given different codes so that in the future a more holistic problem analysis would be

Implementation of a Text Analysis Tool

309

supported. While one may argue that the market would offer better tools than SAP HANA to deal with this type of task (particularly tools specialized in text and sentiment analysis such as SAS’s Analytics and Oracle’s Social Cloud) the company chose SAP for three reasons. First, they had been using SAP for many years, and hitherto were never disappointed by its performance (note: we as authors neither support nor reject this reason, we simply report on what had been presented to us). Second, a follow up project should consider adopting the resulting system for company internal data analysis, and third, the company did not feel comfortable interacting with other, 3rd party solutions providers; i.e. their trust had been with SAP for many years. Consequently, a first proto‐ type was implemented using the ‘Voice of Customer’ (VOC) feature. The pre-defined text analysis dictionaries offered by this feature were extended by a custom dictionary which allowed for linking the alphanumeric code of a product with its name, country code and responsible business sales unit. Data to test the prototype came from internal sources. A typical example for how such a data stream would be triggered starts with a customer who identifies a defect with a bought product and files a complaint at the point of sales. The complaint is then manually added to the system adhering to the following structure: COMPLAINTID, COMPLAINEEINFO, COMPLAINTDESCRIPTION. The SAP HANA VOC feature processes the complaint and consequently updates the dashboard applica‐ tion, which staff members can then use to display and analyze all relevant issues. While this workflow seems relatively straight forward it had to deal with two types of chal‐ lenges i.e. (1) the technical challenge of text mining, and (2) the rather social challenge of accepting this new technical tool. In the following we particularly focus on the latter of these two challenges.

4

Social Challenge: Technology Acceptance

In search for some guidance on understanding the social challenge of accepting new technologies we have decided to use three of the existing theories reported by Venkatesh et al. [19] and explore how well their constructs align with the interview-based research methodology we used to understand the challenges and success factors of implementing the company’s TAS. The Technology Acceptance Model (TAM) was chosen because of its meticulousness [10] and its strong empirical support [16]. The Hospitality Meta‐ phor (HM), on the other hand, should provide a powerful theoretical framework for investigating technology adoption in a messy, realistic and highly emotive environment [5]. And, finally, the Unified Theory of Acceptance and Use of Technology (UTAUT) should integrate different models to offer a more holistic understanding of technology acceptance. The following sections describe these three theories in some more detail. 4.1 Technology Acceptance Model (TAM) The Technology Acceptance Model (TAM) consists of four core constructs, i.e. the external variables Perceived Ease of Use (PEOU) and Perceived Usefulness (PU) explain the Attitude Toward Using (ATU), which consequently defines the Behavioural Intention to Use (BIU) a system [6, 10]. Per previous work (cf. [6, 10, 19]) PU is used

310

G. Ghezzi et al.

both as an independent and a dependent variable, as it derives from PEOU and, in turn, predicts both ATU and BIU. TAM’s rational is based on the assumption that technology acceptance is mediated by a user’s attitudes and beliefs, where beliefs are understood as the degree of instrumentality that is tied to an action while attitude is considered to be purely affective [7]. Consequently, one may see beliefs as being related to a person’s subjective judgment of whether performing a given behavior results in a specific conse‐ quence, whereas a person’s subjective attitude (either positive or negative) affects the performance of said behavior [10]. It can thus be argued that, according to TAM, the way an individual accepts a technology depends on his/her attitude towards its usage, which furthermore is defined by its perceived usefulness and ease of use. 4.2 Hospitality Metaphor (HM) Contrasting the clear variable constructs employed by TAM, the Hospitality Metaphor (HM) is seen as a rather theoretical lens that helps understand innovation processes and technology adoption [4]. In fact, some researchers claim that HM’s emphasis on emotional aspects of users’ everyday dealing and struggling with technology makes the metaphor a perfect tool for exploring acceptance [5, 12]. HM is based on the concept that a technology may be perceived as an alien, which is exemplifying and embodying its alien affordances and culture, and that a successful implementation can only be achieved if the ‘host’ organization can extend complaisance and is able to assimilate, absorb and fit the alien’s culture where it offers leverage and benefits (e.g. in new working procedures) [4]. Thus, HM is focusing on how and whether people use tech‐ nology, give it meaning and make sense of its innovation, which in this case may be interpreted as a ‘guest requiring hospitality’ [12]. According to Coleman, once and organization adopts this perspective, it is able to bridge the cultural boundaries between guests and host. This implies that guests can behave as if in their own environment and therefore hosts may need to relinquish control over the environment. In other words, hosting requires duties and efforts, i.e. it usually requires the modification of everyday routines and practices [5]. 4.3 Unified Theory of Acceptance and Use of Technology (UTAUT) The final perspective we consider relevant for our analysis is manifested by the Unified Theory of Acceptance and Use of Technology (UTAUT). The theory was established through reviewing and subsequently consolidating many of the previous acceptance models, providing a more holistic approach to exploring acceptance by including different perspectives. Consequently, UTAUT is believed to be more stable and robust than its predecessors. According to [19], the theory is based on four key elements; i.e. performance expectancy, effort expectancy, social influence and facilitating conditions. The first three directly determine the Behavioural Intention (BI), while the last element is a direct determinant of the usage behavior. In addition, Venkatesh and colleagues found that, gender, experience, age and voluntariness of use are details posited to moderate the impact of the four key elements. Other constructs that have shown an influence include the attitude towards using technology, self-efficacy, and anxiety. All

Implementation of a Text Analysis Tool

311

in all, UTAUT has been one of the most employed acceptance research models of the last decade [15].

5

Study Methodology

In order to explore challenges, important requirements, and relevant success factors of implementing an industrial TAS we conducted two rounds of focus group interviews; one before and one after the implementation of the system. Our interview sample was restricted to staff members who were directly affected by the introduction of the software, including representatives of the implementation team, the current BI team as well as the QM team, amounting to a total of five people (Note: talking to five people meant that we would talk to most employees who were directly affected by the introduction of the TAS). Interviews were fully transcribed and coded, using a deductive coding scheme as guidance (extracted from [8]) and inductively produced codes to fill in the gaps. The focus of the pre-implementation interviews (1st round) was on identifying user require‐ ments for the tool. The second round of interviews (i.e. after the tool implementation) focused on the challenges and success factors of the implementation process as well as on not or only barely fulfilled user expectations. Next, so as to link findings and existing theories, we compared our results with previously published studies on success factors for implementing business intelligence systems [20] and its challenges rooted in organ‐ izational change [8]. Finally, we wanted to understand to what extend existing models on technology acceptance (i.e. TAM, HM, and UTAUT) can explain our findings. Given that these models mostly employ quantitative survey data we sent a questionnaire to our participants one month after the tool went live. Although our small sample size does not suffice to draw any conclusion whatsoever on model related issues, it does help cate‐ gorize the type of feedback TAM, HM and UTAUT provide with respect to the imple‐ mentation of such a system.

6

Discussion of Results

Following we report on the results of the above describe research agenda. To provide some guidance, the discussion is divided into three different sections. The first section summarizes the three key requirements regarding the system implementation, which we could identify based on the pre-implementation focus group interviews. The second section focuses on the critical success factors we were able to extract from the postimplementation interviews, and the last section reflects on how our data fits the above described technology acceptance models (i.e. TAM, HM, and UTAUT). 6.1 Three Key Requirements The most important factor pointed out during the focus group interviews concerned the level of automation the TAS should provide. All the interviewees stated that it was fundamental for them to keep control over the process and to not lose their decisionmaking authority, e.g. “We don’t ask you to build something that is overtaking our

312

G. Ghezzi et al.

position here in the company. We still want to have control over the complaints. We want help from the system, that’s it. […] a human must remain in control at all time”, stated a user in the pre-implementation interview. Keeping the users informed at all times, i.e. keeping them in the loop during the development and implementation of the system, the company tried to fulfil this requirement as much as possible. A second important aspect regarding the expectations concerned the amount of effort a user should put into learning the TAS. Users asked to create something easy to learn, given the limited time they would have for training. Consequently, we asked them after the implementation about their learning effort and despite the employed training method (i.e. self-learning, learning through trials and guidelines) they reported that it took them only a few days to use the system without any problems. Here, the fact that the imple‐ mentation happened on a platform they were already accustomed with (i.e. SAP HANA), probably helped smoothen the adoption process. Finally, the third major issue highlighted by the pre-implementation focus groups concerned the system’s core functionality. Here one of the participants stated: “What we really want is to avoid is a mistake happening twice. […] I want to have a dashboard showing how many times we had that complaint, all the products related to that complaint, all the complaints of that product and all the complaints of that customer. We want to have the chance to investigate complaints in a new and more efficient way […] from here we can then start creating a database of possible solutions”. The company aimed at tackling this request to the point that it became its key use case guiding system implementation. 6.2 Three Critical Success Factors Guided by Yeoh and Koronios’s [20] Critical Success Factors (CSF) model the following reflects upon three critical aspects affecting the TAS implementation. First, focusing on the organizational dimension, users stated that it was important to have someone pushing for the system’s implementation and usage. Most of the technical people involved in the actual implementation were busy developing and consequently had no time to advocate and promote the new system. Thus, it was up to the team manager to show commitment and involve key staff members. This sort of commitment was perceived as an important key factor, since the resulting TAS was to be used by different types of staff members (respective end users) and so they had to be involved and pushed to participate in various implementation procedures (i.e. use case descriptions, test runs, feedback loops, etc.). Second, exploring the process perspective, the existence of a well-balanced team is often considered a critical success factor, which was also highlighted by our participants. To that end it was important for the company to have people from different fields on board, so that users would not only deal with the IT department. The different back‐ grounds also helped to learn about realistic TAS use cases and affiliated tasks. A second aspect of this perspective relates to the iterative development and communication strategy. That is, clearly defined communication channels supported the implementa‐ tion’s focus, and helped save both time and money.

Implementation of a Text Analysis Tool

313

Finally, a user-oriented change management process seemed critical [20]. Here it was clear that the TAS was going to change daily routines and workloads. Consequently, the company aimed at involving end users as much as possible in order to shape the TAS according to their expectations, needs and desires. On top of that, users also expected that the TAS would change the workload and procedures prevalent in other teams, e.g. a user firmly stated: “I don’t know exactly how this will influence the company in detail but, apart from our team, there will be a serious effect on the Sales Management, on the Quality Inspection and definitely on the Process Management team related to the Quality Department”. 6.3 Three Acceptance Models Trying to relate the results of our focus group interviews and their subsequent survey study to Davis’ TAM we looked at the data from a more structured point of view; i.e. we used a short questionnaire to collect model specific data. All our participants considered themselves as experts, and they stated that the TAS had already become an essential tool for their daily routines (although they had only been using the system for one month). According to the TAM we may argue that the acceptance of a certain technology depends on the Behavioural Intention to Use (BIU) which, in turn, is strictly linked to the user’s Attitude Towards its Usage (ATU). ATU, as we can see in the model, depends on its Perceived Usefulness (PU) and its Perceived Ease of Use (PEOU). As we have seen before, our participants considered the TAS implementation a success as they rated the final system to be useful and able to improve the quality of their work. Except for some doubts with respect to its level of independence and the consequent reliability of produced results, participants were confident that the system had become an essential component of their everyday work procedures. This optimism and trust toward the TAS, and the perceived usability provided through a rather user-centered development approach, seem to have been the main drivers of adoption, although other external variables may have been positive influencers as well. Personal differences or organizational interventions, for example, might have modified or altered a user’s PU and PEOU. Thus, while our results support the influence of PU and PEOU on ATU and BIU, they also highlight that some other aspects, which lie outside these core TAM constructs, should be considered. To explore the system implementation from an HM perspective we investigated whether the TAS was perceived to evoke a threat to the users’ everyday working proce‐ dures. Ciborra stated in his research that a successful implementation can only be achieved if the ‘host’ organization is able to extend complaisance and assimilates, absorbs and fits the alien’s culture where it offers leverage and benefits [3]. Both preand post-implementation interviews showed that none of our participants perceived the TAS to be an alien that can harm their work. The users felt comfortable with the system from the very beginning so, that once it was in place they even started extending its functions rather than just reading its guidelines and executing pre-defined commands. However, the ability to be in control of the ‘guest’ seemed to be a particularly important factor for this positive adoption, as during the pre-implementation interviews we were often confronted with distinct fears of participants related to the potential power the

314

G. Ghezzi et al.

system would eventually have. This insight is very much in line with the HM theory as it shows that, even if the ‘guest’ is welcome, the ‘host’ wants to keep control over the situation. In the post-implementation interviews, participants stated that they were perfectly aligned with the TAS simply because the system itself does not act or behave without previous authorization. Rather, it shows data in a different way, which enhances the quality of their work without endangering their position. To this end, an important aspect of HM is viewed in the concept of ‘hostility’, which may lead to two different types of behavior; i.e. resistance with respect to new technologies (when a host does not allow a guest to enter the environment and thus, even knowing its characteristics, is not interested in sharing its habitat), and hostile rejection (when the host perceives the guest as an enemy and therefore will treat it as something that can harm and endanger its current situation). Here, our study results particularly point to the easiness with which the TAS has been accepted as a part of an existing working environment. None of the staff members we talked to felt endangered because of the TAS and therefore none of them showed hostility towards this new system. In summary, we thus found that by treating the situation more like a complex organism rather than a static set of influencing constructs, HM was able to offer a sound description of the given implementation chal‐ lenges. Finally, based on the existence of both qualitative focus group data and a small number of completed questionnaires, it was possible to also explore the suitability of the UTAUT model as a potential means to explain the given implementation case. Here, the data showed that participants expected strong performance from the TAS. That is, they clearly hoped from the system to significantly improve their prevalent job perform‐ ance. Thus, similar to what we were able to see when looking at the data through the TAM perspective, also the UTAUT lens identified the given trust in the expected outcome as the most important determinant for the system’s acceptance. On the other hand, the expected effort required to assure the system’s success was rated rather low. That is, in the pre-implementation focus group interviews, participants already stated that they wanted from the company a system which would not require too much effort in learning. From the very beginning of the implementation process this demand was expected to be met and eventually positively confirmed through the post-implementation interviews. Thus, it seems that the user-centered development process, which continu‐ ously kept end users informed, also helped in keeping up this instalment of trust towards the usability of the final product. The questionnaire data confirmed this assumption, showing that participants had a positive feeling regarding the system’s ease of use and did not experience any intricacies operating it. As for other social influences, the small sample size kept us from identifying any important external variables which would exhibit a strong effect on the given implementation scenario, although we are very aware that with systems that affect the working procedures of more people such social deter‐ minants would most definitely play an important role. However, in the given case the social environment rather acted as a facilitator. This is, from the very beginning partic‐ ipants felt positive about the project’s outcome, which made them consider the complete IT department, the distinct implementation team as well as the higher management, who originally initiated the implementation of the system, a facilitating condition assuring their expectations and consequently their acceptance. According to Venkatesh et al. [19]

Implementation of a Text Analysis Tool

315

this facilitating condition can be split into three different components. First we speak of perceived behavioral control (encompassing self-efficacy and technological/resource facilitating conditions), which was achieved by keeping the end users in control of the system, giving them the resources to learn and use the system (e.g. written guidelines) and, finally supporting its complete adoption by utilizing an underlying system platform which already serves as their daily starting point at work (i.e. SAP HANA). Next, we find the core facilitating conditions themselves (i.e. environmental factors which support a successful implementation process), which in our case were represented by the close contact to the implementation team throughout the entire development process. Finally, the last facilitating condition may be found in the reign of compatibility. That is, given that the TAS was built on top of the already employed SAP HANA plat‐ form, a rather seamless integration with other systems was easily achieved. This aspect was also clearly perceived and notably mentioned by our participants. In summary, we thus may argue that an overall positive intention towards the system (i.e. high outcome expectations and low effort expectations) paired with the existence of additional facili‐ tating conditions led to a successful implementation. Examining the situation from this perspective we may thus argue that the UTAUT model very well explains the given implementation process.

7

Concluding Remarks and Future Work

The goal of the work reported in this paper was to analytically describe key requirements and success factors of implementing a new text analysis tool in an industrial setting. An additional goal was to apply the core concepts described by three different technology acceptance models (i.e. TAM, HM, and UTAUT) and evaluate their fit for the given use case. The tool, i.e. an industrial text analysis system, has been successfully implemented and we believe that the above analysis helps identify the core requirements and success factors that contributed to this achievement. As for the acceptance models, we have seen that they exhibited similarities but also showed different strengths in explaining our data. The TAM model may be considered a robust and powerful instrument for studying and predicting the users’ attitude towards using the system. It was easy to identify factors that contributed to the model’s two main determinants Perceived Usefulness (PU) and Perceived Ease of Use (PEOU) and consequently to show their influence on Attitude Toward Use (ATU). A completely different angle was provided by the HM, which predominantly encompasses emotional aspects. The model treated the new system as a ‘guest’ that aims at entering an existing environment. Here we could see that our users were ready to acknowledge the system but wanted to maintain control over it, high‐ lighting a certain fear of novelty. They argued that, while being generally in accordance with the doings of the system, they require time to build up the necessary trust towards its actions. This human characteristic is often perceived as an insurmountable aspect of technology implementation processes. Finally, the UTAUT model was able to describe the users’ expectations assigned to the new tool. Similar to what was already shown by TAM we found high performance expectancies combined with low effort expectancies. In addition, we were able to identify facilitating conditions, such as the roles of involved

316

G. Ghezzi et al.

parties, as important factors influencing the overall acceptance and consequent adoption of the system. Regarding future work, we plan to conduct similar studies within other companies in order to verify our results. These studies should also add an additional departmental perspective to the data, as we believe that there may be differences between different company departments, particularly when it comes to employees’ age and technology affinity (two parameters which were rather constant in our study setting). Finally, we want to find adequate (potentially better) ways of combining different technology acceptance models and analysis perspectives so as to eventually provide a blue print for a more holistic analysis method capable of describing the implementation and adoption of industrial software systems.

References 1. Abugabah, A., Sanzogni, L.: Enterprise resource planning (ERP) system in higher education: a literature review and implications. Int. J. Hum. Soc. Sci. 5(6), 395–399 (2010) 2. Castaneda, D., Toulson, P.: The value of human resources measurement in intellectual capital and knowledge sharing. Electron. J. Knowl. Manag. 11(3), 226–234 (2013) 3. Ciborra, C.: Hospitality and IT. In: Proceedings of the 22nd Information Systems Research Seminar in Scandinavia, Käkölä, Finland, pp. 176–182 (1999) 4. Ciborra, C.: The Labyrinths of Information: Challenging the Wisdom of System. Oxford Press, New York (2002) 5. Coleman, E.: Hospitality and mood: Ciborra’s hospitality metaphor as a theoretical lens for IS innovation. In: Southern African Institute for Computer Scientist and Information Technologists Annual Conference, Centurion, South Africa, pp. 176–182 (2014) 6. Davis, F.D., Bagozzi, R.P., Warshaw, P.R.: User acceptance of computer technology: a comparison of two theoretical models. Manage. Sci. 35, 982–1003 (1989) 7. Erasmus, E., Rothmann, S., Van Eeden, C.: A structural model of technology acceptance. SA J. Ind. Psychol. 41(1), 1–12 (2015) 8. Fetzner, M.A.D.M., Freitas, H.: Business intelligence (BI) implementation from the perspective of individual change. JISTEM J. Inf. Syst. Technol. Manag. 8(1), 25–50 (2011) 9. Hani, N., Abai, Z., Deraman, A.: Incorporating business intelligence and analytics into performance management for the public sector. In: Proceedings of the 5th International Conference on Electrical Engineering and Informatics, Bali, Indonesia, pp. 484–489 (2015) 10. Lee, Y., Kozar, K.A., Larsen, K.R.T.: The technology acceptance model: Past, present, and future. Commun. Assoc. Inf. Syst. 12(1), 752–780 (2003) 11. Lim, E.-P., Chen, H., Chen, G.: Business intelligence and analytics: research directions. ACM Trans. Manag. Inf. Syst. 3(4), 1–10 (2013) 12. Saccol, A.Z., Reinhard, N.: The hospitality metaphor as a theoretical lens for understanding the ICT adoption process. J. Inf. Technol. 21(3), 154–164 (2006) 13. Schurink, W., Beyleveld, A.: The social construction of systems applications products (SAP) R/3 by employees of a South African chemical industries company. SA J. Hum. Resour. Manag. 3(1), 22–33 (2005) 14. Soh, C., Kien, S.S., Tay-Yap, J.: Enterprise resource planning: cultural fits and misfit: is ERP a universal solution? Commun. ACM 43(4), 47–51 (2000) 15. Taiwo, A.A., Downe, A.G.: The theory of user acceptance and use of technology (UTAUT): a meta-analytic review of empirical findings. J. Theor. Appl. Inf. Technol. 49(1), 48–58 (2013)

Implementation of a Text Analysis Tool

317

16. Taylor, S., Todd, P.A.: Understanding information technology usage: a test of competing models. Inf. Syst. Res. 6(2), 144–176 (2001) 17. Turner, D., Chung, S.H.: Technological factors relevant to continuity on ERP for E-business platform: Integration, modularity, and flexibility. J. Internet Commer. 4(4), 119–132 (2005) 18. Umble, E.J., Haft, R.R., Umble, M.M.: Enterprise resource planning: implementation procedures and critical success factors. Eur. J. Oper. Res. 146, 241–257 (2003) 19. Venkatesh, V., Morris, M.G., Davis, G.B., Davis, F.D.: User acceptance of information technology: toward a unified view. MIS Q. 27(3), 425–478 (2003) 20. Yeoh, W., Koronios, A.: Critical success factors for business intelligence systems. J. Comput. Inf. Syst. 50(3), 23–32 (2010)

Value Chain and Supply Chain

A Workflow-Driven Web Inventory Management System for Reprocessing Businesses Huanmei Wu1 ✉ , Jian Zhang2, Sunanda Mukherjee1, and Miaolei Deng3 (

)

1

2

School of Informatics and Computing, Indiana University Purdue University Indianapolis (IUPUI), Indianapolis, IN, USA [email protected] School of Computer and Engineering, Shenzhen Polytechnic, Shenzhen, Guangdong, China 3 School of Information Sciences and Engineering, Henan University of Technology, Zhengzhou, Henan, China

Abstract. This paper describes the design and implementation of a workflowdriven web application to manage the inventory and business operations for the recycling and reprocessing businesses. On the backend, the relational database is built with flexibilities and extensibilities for future organization expansion. The user management is customized to the company needs with tiered role-based data access control. For the front end, the graphical user-friendly interface is tailored for non-IT users to efficiently update and search inventory items, track incoming and outgoing orders, and manage their business workflow. For the middleware, the business logics, operational workflow managements, the data processing, the system optimizations, and web-page connections are integrated into the design and development for the various business operations. The system has been imple‐ mented using MySQL and PHP with the XAMPP the web service package. Addi‐ tional dashboards, graphic user interfaces, and data analytical functions have also been implemented. This system can be easily transformed to other database management systems and other middleware tools. The overall database design can be conveniently adopted by other businesses where inventory management is a challenge. Keywords: Web application · Inventory management system · Data analytics

1

Introduction

It is well known that the information technology and database management systems have a great impact on the supply chain businesses [1]. There are many data management systems developed for inventory management and production planning, such as for the automotive service industry [2–4] and for disaster management [5]. Most of these systems provide the essential functions to monitor the product inventory and facilitate the resource purchasing and orderings. However, many existing commercially available systems are generally designed for the common supply chain and inventory manage‐ ments. Some special supply chain industries have unique inventory procedures or

© Springer International Publishing AG 2017 L. Uden et al. (Eds.): KMO 2017, CCIS 731, pp. 321–335, 2017. DOI: 10.1007/978-3-319-62698-7_27

322

H. Wu et al.

distinctive business workflows, which cannot be satisfied with the existing inventory management systems without major customization of the commercial systems. Our project team was approached by such a recycling company owner. They have investigated and consulted many vendors of inventory management systems and had not found one that will satisfy their special needs for resource managements and produc‐ tion monitoring. For the privacy concern, the business name is anonymized in this manuscript. Since their business needs on inventory management are common to many reprocessing companies, this paper will introduce the specific considerations, general design, and potential applications of a web inventory system, which is driven by the production workflow, customized for the diverse user groups, and restricted access privileges based on the user roles. Upon the request of the recycling company, we have designed and developed a web application for the inventory management and business workflow monitoring. The system design has addressed the special demands for their organizational resource supervision and incorporated the distinctive reprocessing workflow. In addition, the system has been optimized for future business growths and expansions and enhanced with additional data mining and reporting functions. The rest of the paper is arranged as follows: Sect. 2 will introduce the special chal‐ lenges of the project. Section 3 will focus on the database design and user management. The overall system implementations and developments are covered in Sect. 5, with some friendly application-based user interface associated with data mining and reporting functions user interfaces. The last section is a quick summary and discussions of future development directions.

2

Special Challenges for the Reprocessing System Design and Information Management

Through several onsite visits of the recycling workspace, careful observation of the production workflow, and thorough discussions of numerous levels of potential system users, we have discovered the special requirements and unique production features for the reprocessing company. These make it challenging for the inventory management system and have not been systematically addressed with existing commercial inventory system providers. First, an inventory management system needs to address the numerous opera‐ tional workflows, especially the reprocessing procedure of a recycling business. Thus, in the beginning, it is important to have a systematic design of the business workflow and optimize it into the relational database design. This task requires a comprehensive understanding of the business operations and involved entities (including employees, products, resources, and others). Figure 1 illustrates a simpli‐ fied business workflow we designed for a reprocessing company from the purchasing of the recycling materials to the selling of the reproduced final products. By design, categorizing and normalizing the entities and the relationships, the coherent business workflow is disseminated into three major transactions: (i) the purchasing transac‐ tions for the incoming recycling materials, (ii) the engineering transactions that will

A Workflow-Driven Web Inventory Management System

323

reprocess the recycling materials to the reprocessed products, and (iii) the selling transactions for the outgoing reprocessed products.

Fig. 1. The business workflow for inventory items throughout of the reprocessing company.

Each of these transactions involves many automated sub-transactions and different entities. Existing commercial inventory management systems do not usually cover the engineering processes in such a sophisticated way. For instance, some of the reproc‐ essing procedures involve three separate major instruments, located in different areas of the warehouse. The tracking and transporting of the intermediate products, the analysis of the major instrument usages, and the operator performance analysis on these major instruments, are big concerns of the organization leaders. Unfortunately, they did not discover an easy solution from the commercial inventory systems. Second, in addition to the three major transactions and the sub-procedures, there are additional interconnections among different stages of the transactions. For example, when an incoming purchase order is submitted, the company needs to track the inbound shipping status from the specific carrier, to schedule the unloading by specific operators, and to plan the storage spaces for the incoming materials. During unloading and storing, the inventory database will update the related information for the incoming materials, such as the appropriate quantities and storage information. Similarly, for any outgoing selling, the subsequent tasks include schedule the shipping carrier, load the products onto a specific truck, contact the buyers for status updates, release the storage spaces, and update the inventory database for the processed product quantities and other related information.

324

H. Wu et al.

Third, the engineering transactions is a special feature for the reprocessing busi‐ nesses. Usually, the unprocessed recycling materials are made from multiple parts of materials with different qualities. The reprocessing procedures will separate these parts based on the chemical and physical properties of the materials, such as colors, densities, sizes, and other features. These separated parts are grouped into different categories. The intermediate fragments of the same categories will undergo multiple reprocessing steps again to make it into packages of processed products, which will be the base units for selling. During each reprocessing step, the inventory management system will update the processed product information, along with the storage to temporally store the prod‐ ucts. Meanwhile, the inventory information of the unprocessed materials and the ware‐ house storages will also be updated in real-time. Fourth, one additional desired function of the recycling company owner is the effi‐ cient management of their employees, especially for the operators with special skills, such as skills to operate the major instruments. Another additional function is to monitor the usages of the major instruments in the engineering processes. The major instruments are not only expensive but also essential for the business operations. Thus, skilled oper‐ ators with the full utilization of the instruments are key financial factors for the success of the recycling companies. Another major challenge is the implementation of intelligent data mining functions in the inventory system. For instance, the company leadership team would like the inventory system to provide big data analysis for decision support on what recycling materials making more profit for the company or what reprocessed product packages are in great needs. Another example is to use the inventory system to perform environ‐ ment hazard analysis to satisfy the federal regulation. These big data analysis functions have not fully implemented in existing commercial system yet. There are additional challenges coming with the requests to have a friendly user interface for various user specific functions and privileges in using the systems. For example, the operators for loading and unloading need only to access the loading/ unloading operational pages to update the quantities and warehouse storage of the incoming recycling materials. For the operators of the major instruments, they need to log in with additional security checks and verify the authorization on operating the instruments. In addition, they need to access the reprocessing pages to update the infor‐ mation of input materials and output products from the reprocessing procedure. They also need to take the binary pictures of both the unprocessed materials and processed products and upload into the inventory system for future quality assurance. These advanced data mining and knowledge discovery capabilities are not supported by regular inventory management systems. Out design and implement of the workflow driven inventory system has considered these special requirements.

3

System Design and Implementation Methods

This section will briefly present our solution for the database design with considerations of all the challenges mentioned above, along with the user management based on the user roles and fine-grained data access control.

A Workflow-Driven Web Inventory Management System

325

Database Schema Design: Follows the logical relational database design, with consid‐ eration of the specific industrial workflow, the overall simplified entity relationship diagram (ERD) of the inventory database is illustrated in Fig. 2. A few major entries are introduced below.

Fig. 2. Entity relationship diagram for MySQL database

• The InboundLineItem table: stores the detailed information about the incoming unprocessed recycling materials. • The ProcessLineItem table: stores the detailed information about the processed product. • The Work_in_Process table: records the engineering processing procedure, which transforms the unprocessed materials to the processed product, with the skilled oper‐ ators and equipment information. • The UserPrivilege table: for the authorization and authentication process based on the potential user roles and access rights, to be described next in details. The details of the item features and storages information are associated with both the unprocessed materials and processed products. They are designed into different tables to facilitate the data mining functions and reporting tools. The buyers for the processed product can be domestic and international business partners. The outbound shipping information will be different depending on the locations of the buyers. For example, for international shipping, additional custom reports for both the departure and arrival countries will be needed, such as the CCIC reports along with additional agent information for international shipping from USA to China. User Security Management and Access Controls: The project revolutionized the cyber security in database systems with strict authorization and authentication

326

H. Wu et al.

approaches, yet user-friendly. It also provided the state-of-art biometric based cyber security techniques and advanced fine-controlled multi-tier hierarchical data access approaches. The project provided higher data sharing security, more portability, and better accountability. The fundamental model is the database-tier, consisting of Rela‐ tional DBMS that manages CRUD (Create, Read, Update and Delete) functionalities. The users have different roles, such as owners, loading/unloading, processing, sales, and others. Each role will have different privileges to access different data sets, i.e., different tables, different columns of a table, or only customized views. This additional informa‐ tion is kept in the additional user tables, which can be dynamically updated for the role/ data access. In addition to the traditional user security control, such as user password and pass‐ word encryptions, advanced security features are also adopted into the system. First, security is enhanced by separating database users and application (web page access) users. Most of the user can only access the web application but not the database directly. Second, fine-grained access control is implemented so that users can only access the web page which is related to their job responsibilities. For example, an operator for loading/unloading can not access the reprocessing and instrument operation pages. On the other way, the instrument operator can not access the loading/unloading web pages. Third, the authorization applies the biometrics (such as fingerprints) for touch log into the necessary web page. System Implementation: We have implemented and tested the system with the XAMPP web server package. The XAMPP stands for cross-platform (X), Apache (A), MySQL (M), PHP (P) and Perl (P). It makes transitioning from a local test server to a live server smoothly and easy as well. The homepage design follows the business work‐ flow, which links the inventory management, system setup, administrative tasks, and report generation. The web page flow follows the advanced technology with page file-oriented web application design [6]. The designs of the database and the front-end user interfaces have been iteratively refined through discussion with our industrial partners. The interactive web pages are user-friendly towards non-IT users. For example, the pages for adding and updating/deletion of item purchase, item selling, and work-in-progress order, are designed with the consideration of the roles of the logged-in users. In addition, the database query engine provides complex query capabilities and various report functions which facilitate efficient data retrieval.

4

Auxiliary Functions to Improve Organization Operations

Our system design has addressed the special concerns of the recycling business. The system implementation made it easier for resourceful data retrieval. The database design made it more flexible to have additional auxiliary functions, which can improve the organization performance in different perspectives. In this section, the customized reports and task-specific data analytics will be introduced

A Workflow-Driven Web Inventory Management System

327

Customer Oriented Dashboard Design: One important task of the system design is to make the inventory management system available to the organizational users, who have no or limited background in database or information technology. Thus, userfriendly design and implementation of different dashboards are in great needs for the widely use of the system. Sample dashboards are illustrated in Fig. 3 for a user with administrative privileges, who can perform more advanced tasks of the database and system. When the user is signed in, the first dashboard (Fig. 3a) will be available. It illustrated the workflow of all the three major transactions. The user can click on any procedure of the workflow to start navigating the system.

(a) Home page dashboard

(b) Dashboard for various inventory

(c) Dashboard for specific Reports

(d) Dashboard for system/DBA only

Fig. 3. User-friendly dashboards for organizational use.

The left panel of each dashboard is the navigation controller, which including the Home, Inventory, Reports, and System four major modules for the current implementa‐ tion. Each of these modules will direct an authorized user to a new dashboard with different functions. • The default Home page (Fig. 3a) is for performing CRUD operations on the inventory system by following the organization business workflows.

328

H. Wu et al.

• The Inventory dashboard (Fig. 3b) is to search and retrieve inventory information from the system, such as information about the unprocessed materials, the processed products, and their stocking within different durations. • The Report dashboard (Fig. 3c) will generate various reports. Samples include the employee performance (such as the skilled instrument operator reports), the supply chain (such as supplier or buyer reports), the resource utilizations (such as the major instrument running reports), and the business operations (such as the shipping reports for inland sales, oversea sales, and receiving reports). • The System Dashboard (Fig. 3d) is power as it can change the database structure. Thus, available only to users with a system or DBA privilege. From this dashboard, a user can update the database, such as the employee records, access privileges, and other core information of the system. The user security and fine-grained access control will guide the availabilities of the dashboards. Not all the dashboard functions are available to all the users. Each user can only access the functions according to his/her roles defined, controlled by the system security policies as explained in last sections. For example, the System Dashboard (Fig. 3d) is available only to users who have a system or DBA privilege. User-friendly Interactive Graphic User Interface (GUI): From the dashboard, clicking on any of the detailed functions will navigate to a user-friendly interactive GUI. With the GUI, a user can specify different parameters to retrieve information, update the database, or generate the corresponding reports. Our inventory system has made many different functions easy to operate with the GUI, with respect to inventory management and system setup. For example, the insertion of new records, update or deletion of an existing record, or retrieve specific purchase is possible with the GUI for users with no database/IT background. Figures 4, 5 and 6 gives a few sample web page insert/list/update an incoming purchase order.

Fig. 4. Web page of Add an Inbound Purchase with proper authorization.

Upon the arrival and unloading of a new incoming purchase order, the operator for loading/unloading will have a GUI for them to “Add an Inbound Purchase” (Fig. 4). A user can input the details for a purchase. Upon saving, the system will execute an insert statement and create a new record of the inbound purchase in the back-end data‐ base. These database operations are transparent to the user. After that, this new record

A Workflow-Driven Web Inventory Management System

329

Fig. 5. Web page to list the previous Inbound Purchase for update.

Fig. 6. Detailed page information for updating an inbound purchase.

can be updated by listing the most recent purchase orders (Fig. 5) for an authorized user to insert/update/delete a purchased items from the specific purchase order. For “Update an Inbound Purchase” page (Fig. 6), the overall purchase order information is listed on the top panel. The detailed purchase items are tabulated in the lower panel. A user can then selects the corresponding item to be modified. The record can be then updated or removed from the database. Please note that the delete operation will require additional security check and authentication. The flow for these pages is automated according to the business workflow. For example, upon clicking on “Save” of Fig. 4, the underlying database operations will be carried out in the back-end. Once succeeded, the specific information of these purchase order will appear as in Fig. 6. The user does not need to do additional searches or other operations on the database. In addition to the automated flow of these pages, most of the fields are optimized with controlled information, such as predefined values or drop-down list, which make it easy to search and update. It will also reduce potential insertion and updating errors. Furthermore, interactive dialogues are designed to confirm and validate the operations. Automatically Generated Reports: One of the major challenges for the company is to have an inventory database system, which can help to generate a variety of reports,

330

H. Wu et al.

tailored for various purposes, such as organizational reports operator performance and instrument usages for the engineering transactions, warehouse section managements for efficient moving of materials, asset managements, and sales performances. The diffi‐ culties are these specific requirements are only necessary for the recycling businesses. Thus, most of them are not available from commercial software systems for general inventory purposes.

(a) Sample inland shipping report

(b) Sample oversea shipping report

Fig. 7. Customized Reports for (a) inland shipping and (b) oversee shipping. The highlighted areas are customized for specific requests.

Figure 7a and b provide sample outbound shipping reports, which are strongly needed for one recycling company to get this process automated and tracked. The two shipping reports, one for inland shipping and one for oversea shipping, have some common fields for selling products, most in the top panel of the reports. The yellow

A Workflow-Driven Web Inventory Management System

331

shaded fields with dashed circles are specific fields, either for inland or oversea shipping report. For example, as highlighted in Fig. 7a, the trailer number is only available for inland shipping and used for tracking the sale status. There are several specific infor‐ mation for oversea shipping circled in Fig. 7b. Thus, depending on the destination of the shipping, the shipping reports will be generated automatically and provide the corre‐ sponding information as needed. In addition, fundamental statistical information about the shipped product will be calculated and formatted in the reports, such as grouping the products based on catego‐ ries, calculating the sub-total and total weights, and the number of items. Advanced Data Analytics: On top of the business operations, advanced data analytical functions have been implemented, which provide convenient tools to the organization leaders for analyzing the daily operations and performance assessments. We have also implemented query optimization, data mining functions, and inventory control. A few key featured data analytical functions are briefly introduced below. • Employee performances: Effective employee management plays an important role in the organization operation. Our inventory management system will provide customized summary and statistics of the employee performance to the managers through the dashboards. In addition, the performance of the employees can be analyzed in a time-series or cross-compared with other employees for the similar job responsibilities. For example, for the loading and unloading operators, the hours they worked, the numbers of incidents for misplacing materials, and other metrics, are important performance metrics. It is a great desire that the managers can have these metrics, along with their trends, in a visual presentation. It will increase the effective super‐ vision of their employees. A sample performance analysis for one new loading employee is illustrated in Fig. 8. There are two metrics. One is the overall work efficiency (where 10 is the best) and the other is the number of work-related incidents per week. From these results, the managers can overserve the improvement of the working efficiency and the decrease of the incidents for the employee in the last eight weeks. Comparing the performance of this employee with those of other employees, the manager can easily have evidences-based performance evaluations of the employees in the company. • Purchase and Sale Analytics: The purchases and sales are the vital information for the organization. Timely analytics on the purchases and sales will provide essential guidance for the decision makers and future operational strategies. Our inventory management system offers additional analytics on categorized materials. It not only provides statistical summaries but also mining the information and providing knowl‐ edge learned from the data. For example, through statistical analysis, the productions from two types of unprocessed materials are always stored in the warehouse for more than half a year, while the productions from another six incoming materials are always on high demands and sold out most of the time. This information guides the decision makers in the organization on the priorities and strategies for their purchases and sales. It helps the growth of the organization.

332

H. Wu et al.

• Mining Major Instrument Utilizations and Simulation Modeling: For this recycling organization, there are several expensive instruments, which process the incoming materials and transform the intermediate products to the final goods for sale. Efficient utilization and management of these instruments are a big concern for the organiza‐ tion. Based on the inventory system, extensive data mining is performed to investi‐ gate their utilization, including the hours they are running per day, the operators who maneuver them, the waiting times for the intermediate products, and the maintenance needed. A simulation model is being built based on this information. The best prac‐ tices to use these major instruments can be guided by the simulation model.

Fig. 8. Sample analysis and visualize results for the employee performance analysis.

5

Discussions and Future Work

Efficient inventory management is an important area for software engineering. The system design for customized inventory system with consideration of the special organ‐ ization needs is an essential part of the successful operation. This project illustrates the successful example by cultivating software engineering and project driven information management design to practical development of specific applications. The underlying principles can be extended to other organization types easily. The system design and implementation process will also be a good learning experience for future workforce training, especially in organizational information system design and knowledge management [7–9]. For the information system design in industry, the focus will be on developing a quality product that is reliable and profitable. The design process involves creative thinking, application of modern technology, and economic consideration. A good engi‐ neering design not only ensures outstanding performance but also offers simplicity in manufacturing and facilitates the production. It is important to integrate the manufac‐ turing and assembling phases in the design. A design experience would not be completed without actually building the product and testing it to ascertain if it meets the design specifications [10].

A Workflow-Driven Web Inventory Management System

333

There are many potential opportunities and projects to be carried out on the recycling inventory management system. The following will briefly introduce two of these future projects. (1) Knowledge discovery for Environmental Control: For the recycling companies, environmental control is an important issue. There are many potential environ‐ mental hazards in the reproduction process, which should be considered by the organization leaders. However, there are no such tools available to them yet. Knowledge discovery guided by the federal environmental control policies on the recycling information system will help to improve the environmental hazard controls [11–13]. • Warehouse Storage Managements: For the recycling company, storage management is an important issue. Transferring unprocessed materials and processed products are major operations inside the warehouse. They are the major influential factors of the production efficacy. Managing the warehouse storages to reduce the transferring distances and to minimize the times for transferring and instrument waiting is essen‐ tial for the success of the company. Ongoing data analytics and simulation modeling will be carried out based on the types of materials, the methods of transportations, the major instrument locations, and the reprocessing procedures. A predictive model will be built and various process controls will be simulated for the best strategies to manage the sections and storages in the warehouse [14, 15]. • Supply Chain Management with Data Science: There are many opportunities for supply chain management (SCM), associated with data science, predictive analytics and big data [16]. We will compare our design with the suggestions for SCM to improve the recycling inventory information system, including the data quality problem in the knowledge discovery and predictive analysis [17] system optimization with the challenges [18], and formal models for sustainable SCM [19]. These tasks will be considered in the next phase of the system. • Adapting the systems for other industries: As mentioned before, the system is designed with extensibilities and flexibilities. It can be adapted to other industries where SCM is a challenge. Currently, the system can work for recycling business, such as solid waste management, and paper cycling [20–22]. It can also be customized to other reprocessing industrials, such as medical devices, fuel reprocessing, and used oil collections [23, 24].

6

Conclusion

We have successfully designed and developed a web-based database application to work with a recycling company. The application efficiently stores, organizes and retrieves data to and from an inventory management system. Customized dashboards have been developed with user-friendly graphic user interfaces. Initial data analytics and knowl‐ edge discovery have been performed which provide valuable information for the organ‐ ization decision makers.

334

H. Wu et al.

This framework was designed by following the business workflow, with additional auxiliary functions. The web page design is of great help for the efficiency and effec‐ tiveness in the recycling-supply chain management. Future work will optimize the system with additional universal modules for seamless transitions among different organizations and incorporate more business intelligence for big data analysis and knowledge discovery. For example, we can extend the current implementation to heter‐ ogeneous database systems [25]. In addition, we will optimize the system with compar‐ ison of the state-of-the-art SCM models. Acknowledgments. This work is partially supported by Science and Technology Development Project of Henan Province (No. 152102410033).

References 1. Peterson, R., Silver, E.A.: Decision Systems for Inventory Management and Production Planning, p. 799. Wiley, New York (1979) 2. Gill, S., Bhullar, S., Singh, N.P., Khullar, P.: A review paper on various approaches of spare parts inventory management system. In: 3rd National Conference, p. 304, February 2016 3. Li, S.G., Kuo, X.: The inventory management system for automobile spare parts in a central warehouse. Expert Syst. Appl. 34(2), 1144–1153 (2008) 4. Bahrom, M.Z., Roslin, E.N., Razak, S.N.A., Rahman, M.A.A.: A conceptual model of inventory management system using an EOQ technique–a case study in automotive service industry (2015) 5. Ozguven, E.E., Ozbay, K.: A secure and efficient inventory management system for disasters. Transp. Res. Part C Emerg. Technol. 29, 171–196 (2013) 6. Zhang, J., Wu, H.: Page file-oriented web application design. In: The Proceedings of 2016 8th International Conference on IT in Medicine and Education (ITME 2016), pp. 1–6 (2016) 7. Baker, D.C.: An innovative project-driven senior-level antennas course. IEEE Trans. Educ. 40(3), 190–194 (1997) 8. Zhang, Y., Chen, S.Y.: Cultivating program of software talents based on project-driven. Comput. Educ. 9, 21–23 (2010) 9. Ohlsson, L., Johansson, C.: A practice driven approach to software engineering education. IEEE Trans. Educ. 38(3), 291–295 (1995) 10. Liu, J.: Application of ‘Project-driven’ teaching method in teaching of SQL SERVER database technology. J. Bohai Univ. (Natural Science Education) 30(4), 366–370 (2009) 11. Xing, G.H., Chan, J.K.Y., Leung, A.O.W., Wu, S.C., Wong, M.H.: Environmental impact and human exposure to PCBs in Guiyu, an electronic waste recycling site in China. Environ. Int. 35(1), 76–82 (2009) 12. Byström, S., Lönnstedt, L.: Paper recycling: environmental and economic impact. Resour. Conserv. Recycl. 21(2), 109–127 (1997) 13. Arena, U., Ardolino, F., Di Gregorio, F.: Technological, environmental and social aspects of a recycling process of post-consumer absorbent hygiene products. J. Clean. Prod. 127, 289– 301 (2016) 14. Kara, S.S., Onut, S.: A stochastic optimization approach for paper recycling reverse logistics network design under uncertainty. Int. J. Environ. Sci. Technol. 7(4), 717–730 (2010) 15. Louis, G., Shih, J.S.: A flexible inventory model for municipal solid waste recycling. SocioEcon. Plan. Sci. 41(1), 61–89 (2007)

A Workflow-Driven Web Inventory Management System

335

16. Waller, M.A., Fawcett, S.E.: Data science, predictive analytics, and big data: a revolution that will transform supply chain design and management. J. Bus. Logist. 34(2), 77–84 (2013) 17. Hazen, B.T., Boone, C.A., Ezell, J.D., Jones-Farmer, L.A.: Data quality for data science, predictive analytics, and big data in supply chain management: an introduction to the problem and suggestions for research and applications. Int. J. Prod. Econ. 154, 72–80 (2014) 18. Garcia, D.J., You, F.: Supply chain design and optimization: challenges and opportunities. Comput. Chem. Eng. 81, 153–170 (2015) 19. Brandenburg, M., Govindan, K., Sarkis, J., Seuring, S.: Quantitative models for sustainable supply chain management: developments and directions. Eur. J. Oper. Res. 233(2), 299–312 (2014) 20. Pariatamby, A., Tanaka, M.: Municipal solid waste management in Asia and the Pacific Islands. Environmental Science. Springer, Singapore (2014) 21. Rahmaninia, M., Khosravani, A.M.I.R.: Improving the paper recycling process of old corrugated container wastes. Cellul. Chem. Technol. 49(2), 203–208 (2015) 22. Cheung, W.M., Pachisia, V., Hasan, R.: Facilitating waste paper recycling and remanufacturing via a cost modelling approach (2014) 23. Macaluso, A.: Characterization of biofilms on medical device materials with application to reusable surgical instruments (2014) 24. Rivier, C., Roudil, D., Rigaux, C., Camès, B., Adnet, J.M., Eysseric, C., Organista, M., et al.: Validation of analytical methods for nuclear spent fuel reprocessing. Prog. Nucl. Energy 72, 115–118 (2014) 25. Wu, H., Ambavane, A., Mukherjee, S., Mao, S.: A coherent healthcare system with RDBMS, NoSQL and GIS databases. In: The Proceeding of the 2017 32nd International Conference on Computers and Their Applications (CATA 2017), pp. 1–6 (2017)

Towards Information Governance of Data Value Chains: Balancing the Value and Risks of Data Within a Financial Services Company Haifangming Yu and Jonathan Foster ✉ (

)

Information School, University of Sheffield, Regents Court, 211 Portobello Street, Sheffield S1 4DP, UK {hyu21,j.j.foster}@sheffield.ac.uk

Abstract. Data is emerging as a key asset of value to organizations. Unlike the traditional concept of a business value chain, or an information value chain, the concept of a data value chain has less currency and is still under-researched. This article reports on the findings of a survey of employees of a financial services company who use a range of data to support their financial analyses, and invest‐ ment decisions. The purpose of the survey was: to test out the idea of the data value chain as an abstract model useful for organizing the different discrete processes involved in data gathering, data analysis, and decision-making; and to further identify issues, and suggest improvements. While data and its analysis is clearly a tool for supporting the delivery of financial services, there are also a number of risks to its value being realized, most prominently data quality, along with some reservations as to the relative advantages of data-driven over intuitive decision-making. The findings also raise further data and information governance concerns. If implemented these programmes can aid in the realization of value from data, while also mitigating the risks of value not being realized. Keywords: Data value chain · Financial services · Data governance · Information governance

1

Introduction

In an age of information and big data, the capture, processing analysis and use of data presents companies with both opportunities and challenges. Within this context, it can be argued that the implementation of a systematic data value chain becomes a prereq‐ uisite for companies to be able to achieve their business goals. From this perspective, the systematic capture, processing, analysis and use of data can be viewed as a programme and a technology for the gathering and processing of data from an organi‐ zation’s internal and external environments for, among other purposes, decisionmaking, analysis of customer intelligence, the monitoring of organizational perform‐ ance, and organizational forecasting [10, 13]. The data value chain encompasses a sequence of processes including data capture, storage, distribution, analysis and use. By implementing a data value chain, an organization can take data-driven as well as intuitive decisions, and more generally increase the use that is made of the data that it captures. © Springer International Publishing AG 2017 L. Uden et al. (Eds.): KMO 2017, CCIS 731, pp. 336–346, 2017. DOI: 10.1007/978-3-319-62698-7_28

Towards Information Governance of Data Value Chains

337

As well as consisting of a set of organizational processes, Kasim, Hung and Li [5] also point out how data value chains can encompass inter-organizational and networked processes. On the one hand, data driven decision-making presents organizations with the opportunity to increasing profits, by aiding the targeting of returns from certain investments and by aiding processes of risk analysis. On the other hand, as one process within a set of interdependent processes, data-driven decision-making also carries with it some of the potential risks associated with the other elements of the data value chain e.g. quality. Therefore, a better understanding of the data value chain, its operating standards and governance, can help to increase the value of data, while also mitigating against the potential risks. The article reports the results of a survey of the employees of a Chinese branch of a financial services company. The aim of the research was to evaluate how the data value chain works within the company, and to identify what factors can influence the outcome of data-driven decision-making. More specifically, how does the data value chain benefit the company? What are employees’ opinions of and attitudes towards the data value chain? What factors influence the data processing of data along the data value chain? Do companies currently have a preference for data-driven decision-making over intui‐ tive decision-making? The structure of the paper is as follows. Section 2 provides a brief review of literature on the concept and structure of a data value chain. Section 3 outlines the survey methodology, including sampling approach. Section 4 presents the findings. Section 5 discusses the results in light of previous literature. The conclusion provides recommendations for future research into the processes of the data value chain.

2

Data Value Chain

The concept and practice of the data value chain is the focus of this research. The concept plays an important role in organizations that rely on the capture and processing of data from their external and internal environments, and its onward processing in a systematic way. Here we provide a number of current definitions of the data value chain and its elements. Miller and Mork [8] suggest that the data value chain can be decomposed into three parts: data discovery, data integration and data exploitation. Kasim’s [5] approach is more granulated with the data value chain including data collection, data management, data integration, data analysis, data simulation and data visualization. With trends towards the utilization of data for growth and well-being increasing, the OECD [9] has proposed a number of enabling factors including digitization, open data, a fast and open internet and the internet of things (IoT) in the area of data gathering; analytics (algo‐ rithms), cloud computing, and specialist data skills in the area of data analysis; and machine learning, automated decision making and simulations and other data experi‐ ments in decision-making. Along this data value chain, data are collected, stored and maintained, used, repaired, and finally destroyed. In general, it is rare for a single model of the entire data value chain to be used in organizations [9]. Therefore, management of each individual element of the data value chain is considered a cornerstone of the effec‐ tive working of an organization’s data value chain [3] with communication, analysis and evaluation considered important aspects of this management. Ofner [10] holds the same

338

H. Yu and J. Foster

opinion. Indeed, the details on how the concept of a data value chain actually works in an organization is difficult to deduce from the literature. For example while Kasim [5] and Miller and Mork [8] refer to the structure of the data value chain, they do not mention how the capture and gathering of data is converted and realized into value that can benefit the organization. While positing the idea of ‘data as an asset’, and therefore the value of data as an entity and a resource, Khatri and Brown [6] also direct attention to the accountability aspects of managing data, and the allocation of decision rights; rather than directing attention to the enabling aspects of converting data into value. Foster [2] points to some of the initial models being used to realize value from data.

3

Methods

A survey of employees of a financial services company was conducted. The survey method was considered appropriate as a deductive method of inquiry for testing out the idea of the data value chain, as a useful abstract model for organizing the different discrete processes involving in data gathering, data analysis, and decision-making. The survey is organized into four parts. The initial part gathers demographic data on partic‐ ipants; the second section contains four questions related to data gathering the first process of the data value chain; the third section contains three questions related to data analysis, the second process of the data value chain; and the fourth section contains seven questions related to the final stage of the data value chain, decision-making. Participants in the survey were all members of an international financial services company based in Shenzhen. In order to avoid biasing a specific interest group, a randomized sample of 100 employees from the financial services company was compiled; with the expectation of 70 responses. The survey was posted to the online website Wenjuanxing and, in actuality, 82 responses were received. Descriptive analyses of the data were conducted initially; while evidence for any correlations between two variables was generated as part of further regression analyses.

4

Findings

Demographics. 82 employees of the Financial Services Company participated in the research. Figure 1 presents the distribution of the employees who responded, by age. As can be seen the overwhelming majority of those who are participating in data-driven decision-making are of a younger age, with 91.46% of the employees who are leveraging the data value chain being aged between 18 and 40; and with only 8.54% of employees aged between 40 and 60 similarly doing so.

Towards Information Governance of Data Value Chains

339

Fig. 1. Distribution of respondents by age

Figure 2 shows the distribution of employees by highest educational qualification held. The responses clearly demonstrate the educated nature of data users within the company with 63.41% of respondents holding a Bachelors degree, a further 31.71% a Masters degree, and 2.44% a PhD degree. In sum, 97.56% of the data users within the company are in possession of a Bachelors degree or above.

Fig. 2. Distribution of respondents by highest education degree

Figure 3 provides a distribution of respondents by job title. The general trend is that employees at a lower rank in the hierarchy will tend to be the greater users of data, with employees of a higher rank tending to be the lesser users of data.

340

H. Yu and J. Foster

Fig. 3. Distribution of respondents by position

For example, Associate positions account for 65.85% of employees, and Managing Directors positions 1.22%. The exception to this is Director positions that account for 10.98% of data users. Figure 4 presents information on the allocation of responsibility for the different phases of the data value chain. In terms of distribution of responsibility the greatest number, and vast majority, of users are involved in the data analysis phase of the data value chain (57.32%), with 24.39% involved in data gathering, and the least number involved in data-driven decision-making (18.29%). Respondents were in clear agreement that data and its analysis was a useful tool in supporting their work.

Fig. 4. Distribution of respondents by phase of data value chain

In response to the statement “I think big data analysis helps my work”, 89.03% of the respondents were in either agreement or strong agreement with the statement; with only 10.98% of respondents either neutral about or disagreeing whether data analysis is helpful in supporting their work. Nevertheless respondents were also aware that there exist a number of risk factors that could affect the accuracy of the data used (multiple choice). Figure 5 illustrates these problematic factors. The most frequented cited factor is data quality (78.05%), followed

Towards Information Governance of Data Value Chains

341

by a clutch of other factors: data lacking variation (41.46%), human factors (39.02%), and data processing errors (30.02%).

Fig. 5. Distribution of risk factors

Data gathering phase: Fig. 6 illusrates the sources from which employees gather their data. These are predominantly from the firm’s website (79%), but also from quantitative (11%) and qualitative (7%) surveys, with only 3% coming from other sources e.g. professional data supply websites. The overwhelming majority of employees (91%) will then systematically store this data for future use. Data analysis phase: Fig. 7 identifies the data analysis tool(s) that employees use to analyse data (multiple choice).

Fig. 6. Data by source type

342

H. Yu and J. Foster

Fig. 7. Data analysis tools by frequency of use

The tools most frequently used for data analysis are: Excel (82.93%), Matlab (23.17%), SPSS (20.73%), Stata (13.41%), and R (3.66%). Decision making phase: Respondents were overwhelming favourable towards data-driven decision-making; with 87.8% of employees either favourable or very favourable towards the practice, with only 12.2% of employees exhibiting only an occasionally favourable or unfavourable attitude towards the practice. Figure 8 illustrates the perceived consistency between the results of data analysis and intuitive predictions as to financial performance. Here the figures are more circumspect. While 71.95% of employees perceive a consistency and therefore a useful complementarity between the two types of decision-making, nearly a third of employees (28.05%) were neutral on this issue. Figure 9 illustrates the number of people who will be typically involved in making decisions on the basis of the data analyzed.

Fig. 8. Perceived consistency between big data analysis results and personal predictions

Towards Information Governance of Data Value Chains

343

Fig. 9. Decision-making process by number of employees

The typical number of decision-makers being 3 employees (39.02%), although the number of decision-makers can range from 1 decision-maker to (15.85%) to more than 4 decision-makers (15.85%). What can be said is that the collaborative use of data, and available of data for joint interpretation and decision-making appears to be the norm; with 84.15% of decisions involving 2 or more people. Two further bi-variate analyses of the data were conducted. Older and therefore typically more experienced employees tended to be responsible for the latter phases of the data value chain. Figure 10 illustrates the relationship between age and responsibility for the different phases of the data value chain. There is a tendency for the younger more inexperienced employees to be involved in data gathering, for slightly older and middle-ranking employees to be involved in data analysis; while senior more experienced employees are 100% responsible for decision-making. After further testing there is some weak evidence that this association is statistically significant (p-value 0.0593). Given the importance attached to data quality as a barrier to accurate decision-making, this rela‐ tionship is worth further investigation. The relationship between the number of employees participating in data-driven decision-making, and employee’s attitudes towards the accuracy of the resulting decision was also tested. Although there is some evidence that in some situations, an increased number of employees can lead to an inac‐ curate decision (Fig. 11), there is very little if any evidence that this is necessarily the case (p-value 0.089). A combination of a competitive financial institution, along with a low-trust culture may provide an explanation for this finding, when multiple people are employed. In addition further data would need to be collected to see whether, in this instance, any decision-making crossed the hierarchical lines that are typical of Chinese business culture.

344

H. Yu and J. Foster

Fig. 10. Relationship between age and responsibility

Fig. 11. Decision-making process by number of employees

5

Discussion

In this section we briefly discuss the findings in light of the previous literature. The value of data is largely determined by its accuracy [4, 7], while timeliness is also considered to be one of the key attributes of data quality [1, 11, 14]. While the key benefits to the Financial Services Company include (a) a logical and reliable data-driven decisionmaking system that can inform the design and selling of financial products that meet customers’ preferences and reduce product risk, thereby leading to improvements in financial returns (b) increases the effectiveness of the collective and individual making of data-driven decisions. Therefore, it is significant that the key risk factor, cited by 78.05% of respondents was data quality. It is clear that data quality is the key issue that

Towards Information Governance of Data Value Chains

345

can reduce the likelihood that value will be created from data. From a data resource perspective it will be important to assess the degree of quality attached to each stream of data. There is some weak evidence that younger, more experienced, employees are responsible for data gathering. If substantiated by further evidence, it will also be impor‐ tant to attend to this issue when allocating decision-making rights and accountability for the domain of data quality, and how this is a key criterion on evaluating the worth of data gathering [6]. The hierarchical structure typical of Chinese business culture may also be a relevant factor, worthy of further investigation in itself, and whether this is generalizable beyond the Chinese context. Given good quality data, the value of data analysis for organizing and interpreting the data is not in question, with the majority of employees in this company also holding a positive view of its worth in supporting financial services. Nevertheless, the perceived gap for some between intuitive and evidence-based data-driven decision-making is worthy of further investigation.

6

Conclusion and Future Research

Data and its analysis are clearly assets valued by those who responded, as being a tool for supported the delivery of financial services. However, what appears to be the case is that the accumulation of this value is framed in a piecemeal fashion. What is required is much greater cumulative management and oversight of the data value chain; and treating the data value chain as a set of inter-linked processes, rather than independent processes. In this regard previous work on the management ([3, 4, 7, 10]) and governance [6] of the data value chain would appear to be a guide to future practice. Nevertheless Khatri and Brown [6] tend to view data as an exercise in potential value and accounta‐ bility, whereas the emerging field of information governance (e.g. [2, 12]) tends to examine the factors that can shape the realization of the value of data, and its conversion into information. These include not only procedural practices associated with managing the data value chain, but also structural practices that assign decision-making and accountability rights; and relational practices that draw attention to the need for commu‐ nication and co-ordination along the data value chain. In doing so the organization will enhance its capacity for enabling value to be realized from data e.g. allocating decisionmaking rights; while also mitigating the risks to that value being created e.g. data quality.

References 1. Blake, R., Mangiameli, P.: The effects and interactions of data quality and problem complexity on classification. Assoc. Comput. Mach. Comput. Surv. 41(3), 1–52 (2011) 2. Foster, J.: Towards an understanding of data work in context: issues of economy, governance and ethics. Libr. Hi-Tech 34(2), 182–196 (2016) 3. Gartner: Governance of master data starts with the master data life cycle. Stamford CT, Gartner Research (2008) 4. Jorge, M., Ismael, C., Bibiano, R.: A data quality in use model for big data. Future Gener. Comput. Syst. 63, 1–8 (2015)

346

H. Yu and J. Foster

5. Kasim, H., Hung, T., Li, X.: Data value chain as a service framework: for enabling data handling, data security and data analysis in the cloud. In: International Conference on Parallel and Distributed Systems, vol. 18, pp. 804–809 (2012) 6. Khatri, V., Brown, C.V.: Designing data governance. Commun. ACM 53(1), 148–152 (2010) 7. Liu, J., Li, J., Wu, J.: Rethinking big data: a review on the data quality and usage issues. J. Photogram. Remote Sens. 115, 1–9 (2015) 8. Miller, H., Mork, P.: From data to decisions: a value chain for big data. IEEE Comput. Soc. 15, 57–59 (2013) 9. OECD: Data-driven innovation: big data for growth and well-being (2015). http:// www.oecd.org/sti/data-driven-innovation-9789264229358-en.htm Accessed 30 April 2017 10. Ofner, M., Straub, K., Otto, B., Oesterle, H.: Management of the master data lifecycle: a framework for analysis. J. Enterp. Inf. Manage. 26(4), 472–491 (2013) 11. Pipino, L.L., Lee, Y.W., Wang, R.Y.: Data quality assessment. Commun. ACM 45(4), 211– 218 (2002) 12. Tallon, P.P., Ramirez, R.V., Short, J.E.: The information artifact in IT governance: toward a theory of information governance. J. Manage. Inf. Syst. 30(3), 145–181 (2013) 13. Vera-Baquero, A., Colomo-Palacios, R., Molloy, O.: Real-time business activity monitoring and analysis of process performance on big-data domains. Telematics Inf. 33, 793–807 (2016) 14. Wand, Y., Wang, R.Y.: Anchoring data quality dimensions in ontological foundations. Commun. ACM 39(11), 86–95 (1996)

A Dempster Shafer Theory and Fuzzy-Based Integrated Framework for Supply Chain Risk Assessment Yancheng Shi(&), Zhenjiang Zhang, and Kun Wang Beijing Key Laboratory of Communication and Information System, Beijing Municipal Commission of Education, Beijing Jiaotong University, Beijing, China [email protected]

Abstract. This paper presents an integrated framework for supply chain risk assessment. The framework consists of some main components: risk identification, D-S calculation, fuzzy inference, risk analysis and risk evaluation. The risk identification comprises three parts, literature review, expert opinion interview, and questionnaire there are all used to identify the risk categories and their reasons and hazards. D-S calculation utilizes Dempster-Shafer Evidence Theory to fuse the potential risk’s information which are identified by the experts’ knowledge, historical data, literature review and questionnaire. The fuzzy inference part aims to solve how to identify the risk’s impact when there are no explicit data. The risk analysis part use the data from D-S calculation and fuzzy inference to define the main bodies of risk, it’s total probability, impact, and the final score of this risk-event. The risk evaluation component integrates all resources from the risk analysis part and gets a final supply chain score based on the assignment weight which are decided by the experts. A case study from a computer manufacturing environment is considered. Through the analysis of the supply chain, integrating the probability, hazard, and weight of the risk events and calculating a final score, managers can have a comprehensive understanding of the risks in the supply chain, and make some reasonable adjustment to avoid risks and reduce error rate for the purpose of maximizing their profits. Keywords: Risk identification aggregation
Fuzzy logic




D-S calculation




Risk analysis




Risk

1 Introduction Supply chain was sought after by business and academia from its birth date for the information sharing, cohesion of the core competitiveness of enterprises, rapid response to the market demand, effective allocation and optimization of resources, reducing of the unnecessary circulation, reducing of the costs, improvement of customer satisfaction and improvement of competitiveness of global economic integration. SupplyChain risk which uses the vulnerability of supply-chain systems is a potential threat, and it can bring enterprise losses, damage to the supply chain system. How to measure and manage supply-chain risk has become an important field of it’s research. © Springer International Publishing AG 2017 L. Uden et al. (Eds.): KMO 2017, CCIS 731, pp. 347–361, 2017. DOI: 10.1007/978-3-319-62698-7_29

348

Y. Shi et al.

The key drivers for supply chain profitability are: responsiveness, efficiency, and reliability. To maintain their profitability, supply-chains must be able to respond quickly to external and internal risk events, and keep their businesses efficient and dynamic. This paper define risk as two big categories, external risk and internal risk. As for external risk, there are some specific branches like natural risk, market risk, political risk and so on. Another big class is internal risk, which includes all procedures, information, resources and players such as suppliers, manufacturers, intermediaries, third-party service providers, logistics activities, merchandising and sales activities, finance and information technology. These risks are common in each system, every risk has their own elements with the probability defined by the experts, literature review or questionnaire in process of risk identification. Considering the supply-chain risk elements have great uncertainty, it is hard to make accurate estimates of risk based on historical data or information and only rely on experts or decision makers based on their own experience and knowledge to make subjective estimates of risk. But this kind of subjective estimate is not accurate, it’s the uncertain information, if the expert opinion has big deviation, the assessment of the whole system may totally wrong. The level of uncertainty depends on the amount and type of information available for estimating risk likelihood and impact. All studies were screened through 4 steps: Risk identification is a critical step for the success of whole system’s risk management. Risk identification is the process of classifying the risk affair, defining the risk’s element, documenting, and gathering the risk information form experts and questionnaire. Risk identification in this paper through expert opinion with an open questionnaire, It is trying to select experts from different kind of companies and organizations with different owner ships and field of work to cover all points of view in the industry. In this step, risks which have certain special structure are classified and defined explicitly, it’s subsidiary information also be collected. This structure consists of three components: risk event, the elements of the risk (we believe each elements of the risk are independent), each element has a probability, and it’s hazard index based on the fuzzy set. From the survey part, the weights of each risk also be collected. Step 2: D-S calculation. Because of the uncertainty of the information from the experts, this step aims to fuse the data suitably. Dempster Shafer theory is used to fuse the probability of each element of the risk, and also acquire a confidence interval of each risk’s element. When a whole system is established, before the risk analysis step, basic information must be input due to the characteristic of each risk. D-S calculation process provides the final data from the multiple input system to the risk analysis part. Step 3: Risk analysis. In this step, according to the confidence interval of each element, some risk elements which interval value below the threshold value defined by the authority personnel should be sifted out. Through the specific algorithm, those probabilities of each risk’s element are calculating into a integration probability which are used to help the managers make the right decisions. The focus of this step is to figure out the risk score of the event, using the data from step 2, and step 3. Based on the different computational scheme decided by the management, a overall score of the supply chain can be calculated to be the input data of the risk evaluation process. Step 4: Risk evaluation. In this step, corresponding risk rank can be revealed to the management due to the final score of the whole system

A Dempster Shafer Theory and Fuzzy-Based Integrated Framework

349

from the previous step. Different risk levels may provides the information and some suggestion for supply chain risk management. The rest of this paper is organized as follows: Sect. 2 presents the proposed framework for supply chain risk management. Section 3 discusses the application of the proposed framework for risk assessment in a computer manufacturing case. Finally, conclusions and future work are presented in Sect. 4.

2 Proposed Framework The proposed framework (see Fig. 1) combines human involvement and mathematical analysis methods for whole system risk assessment. For each supply-chain, risk events are clearly defined and the basic elements of the supply-chain belong to specific risk affair. The main risk affair’s elements and their probability are identified based on experts’ knowledge, historical data, document literature. A survey in the risk identification part is developed to identify the basic risk affair and their elements in the whole system, the investigation of this part has generality. What kind of basic risk and the event’s elements are well-defined in this survey. Another survey is used to obtain estimates for the other parameters of the risk in the start of the information fusion part. The parameters consist of the probability, hazard index and the weight of each risk affair’s element. The quantity of this part depends on the number of experts. The estimates for risk parameters are used as inputs to the D-S calculation part, because of the uncertainty of each parameter investigated by different experts, some fusion rules are used in here for the purpose obtaining a integrated probability of each risk affair’s element.

Fig. 1. A proposed framework for supply chain risk management.

When the probability fusion from different experts is completed, there also has a confidence interval of each risk’s element. Not all elements which are defined in the first step are necessary in every specific supply chain. Therefore, next step aim to delete the corresponding element by setting the threshold. Through screening and normalization, the complete information of the supply-chain risk has been collected and

350

Y. Shi et al.

processed. Previously we assumed elements are independent of each other, so the rules for element’s synthesis are formulated in the risk analysis part. To this end, the probability of each event in the whole system has been derived, and by interacting with the previously defined hazard index matrix, there have output score of each risk affair. Different risk events have different weights, finally get a total score of supply-chain risk. The weights are obtained from the supply-chain’s experts in the company.

2.1

Risk Identification

In this paper, a risk may be involved in the whole system are defined for two categories: Internal risks and external risks. Internal risks include the basic elements of the supply-chain. External risks include some social, environmental, and market factors. Each risk has a specific structure: event of risk, risk elements. The difference between risk element and it’s event is that risk element is the driver for the event of risk. For example, a customer risk’s event can be caused by order cancellation, returns, customer liquidation, demand variability. Such structure is showed in Fig. 2. The example above shows the elements of the event. The risk element has two parameters: probability and hazard index. For example, the elements of order cancellation has probability and hazard index which are defined by the experts. A typical supply-chain consists of supplier(s), manufacturer(s), customer(s), and transportation. Any part of a supply-chain always be corresponding to the above basic types (see Fig. 3).

order cancellation

hazard index

returns hazard index

Weight

Customer Risk

customer liquidation demand variability hazard index

Fig. 2. Structure of supply chain risk

Fig. 3. Supply Chain Risk

A Dempster Shafer Theory and Fuzzy-Based Integrated Framework

351

As for the internal risks, there are several categories: For each supplier, the corresponding risk is called “Supplier Risk”; for customers or customer regions, risk is called “Customer Risk”; for transportation, it is called “Transportation Risk”; for each manufacturer the risk is called “Manufacturer Risk”. Except the Customers risk and it’s element, other risk’s structure (risk affair’s element) should be well-defined in the risk identification step. Possible elements of other risk affairs are as follows: Supplier risks consist of the elements like: Capacity constraints, supplier bankruptcy, quality issues; Commodity risk consists of price change, technology risk, quality risks; Manufacturer risks consist of Poor planning and scheduling, lack of standardization, process variability, forecasting errors, contract management, payment errors, Technical limitations, technology change, innovation risk, Design changes, quality issues. Natural risk consists of natural disaster, force majeure risk, disease and so on. Political risks consists of Social, political turmoil, laws and regulations change, exchange rate change, inflation. Market risks consist of industry volatility, competing risks. The above process is completed in the first part. It is assumed that each element is independent of each other. Second survey part is used to input the probability and hazard index of each element from it’s event of risk. This quantity of survey is determined by the number of experts. Each expert makes judgments based on his own experience, including the probability and hazard index of each element. These data will be used as input for the next step. About the hazard index, this paper makes a evaluation set V = {low, less secondary, medium, significant, high} = {0.1, 0.3, 0.5, 0.7, 0.9}, experts can decide the hazard index score based on this set. For example, a expert thinks the element of the risk has a great harm, he may decide the hazard index score of the risk affair’s element as 0.8. On the contrary, when the expert believes the element is unimportant, the hazard index score may set as 0.2. Thus we can get an hazard matrix from all experts, it collects all hazard index score of one risk affair’s elements. The rows of the matrix are equal to the number of the elements from a event. The columns of the matrix is equal to the number of experts. Different number in a columns represents a expert’s evaluation for different element’s hazard index score form one risk affair. 2

r11 R¼4M rn1

r12 M rn2

r13 M rn3

r14 M rn4

3 r15 M5 rn5

The survey of the first part is general and universal, this step can be omitted when the system is not used for the first time. The survey of the next part is special and different, experts need to input the information based on the characteristics of the whole system differently when the supply chain is different. 2.2

D-S Calculation

The uncertainty of information from the different experts is the greatest problem in the previous step. Probability and hazard index from each risk affair’s element are decided by experts own experience. The main idea in this step is fusing the probability from the experts which towards to one risk’s event. Dempster Shafer theory is used to accomplish this goal. The beginning of this section first has a brief review on dempster Shafer

352

Y. Shi et al.

theory. The theory of belief functions initiated from Dempster’s work in understanding and perfecting Gisher’s approach to probability inference, and was then mathematically formalized by Shafer toward a general theory of reasoning based on evidence. Belief functions theory is a popular method to deal with uncertainty and imprecision with a theoretically attractive evidential reasoning framework. Dempster-Shafer theory introduces the notion of assigning beliefs and plausibility to possible measurement hypotheses along with the required combination rule to fuse them. It can be considered as a generalization to the Bayesian theory that deals with probability mass functions. Mathematically speaking, consider X to represent all possible states of a system, in this paper, X represents all the elements from one event including a universal set which express the meaning of uncertain. Shafer theory assigns belief mass m to each element of X, which represent the probability receive from one expert regarding the opinion on each element. Function m has two properties as follows:

1:

2:

mð/Þ ¼ 0 X E2X

mðE Þ ¼ 1

This step believes each element is independent of each other, we discuss the connection between elements in next section. Value m is decided by the expert represents the probability of each element. When an expert evaluation is completed, all the elements including a universal set from one risk affair have the m value. Evaluation from different experts is fused using the Dempster’s rule of combination. Consider two sources of information with belief mass functions m1 and m2, respectively. The joint belief mass function m1,2 is computed as follows: m1;2 ðE Þ ¼ ðm1  m2 ÞðEÞ ¼

1 X m ðBÞ m2 ðC Þ B \ C¼E6¼/ 1 1K

m1;2 ð/Þ ¼ 0 where K represents the amount of conflict between the sources and is given by: X m1 ðBÞ m2 ðC Þ K¼ B \ C¼/

Thus we can get a fused evaluation from two experts. According to this rule, we can achieve any number of expert’s evaluation fusion. bel(E) is called belief of E and pl(E) is called plausibility of E they are defined as below: X belðEÞ ¼ mðBÞ X BE plðEÞ ¼ mðBÞ B \ E6¼/

A Dempster Shafer Theory and Fuzzy-Based Integrated Framework

353

An interval constituted by the value [bel(E), pl(E)] is so called confidence interval. It is used to remove corresponding elements through the setting threshold in next section. If the median of confidence interval less than the threshold setting by the experts, it is believed that the element of the risk affair is a small probability factor, contributing almost nothing to the event, then remove it from collection of risk affair. Based on this, the final probability of each element from a event could be deduced. The integrated data as the input for the next part.

2.3

Risk Analysis

After the step 1 and step 2, three valid information has been formed: the specific risk events involved in the whole system, the integrated probability of each element from a risk event, the hazard index matrix had the weight of each risk affair. This step aims to solve three problems: 1. It is assumed that the elements of risk event are independent of each other in step 2. Considering the combination of different elements, this part figure out the final probability of each event and the data is used to be the input for the risk evaluation step; 2. Calculate the risk affair score using the data from integrated probability and hazard index matrix; 3. Calculate the whole supply chain risk score via every risk’s event score in the overall structure and their weight. In the use of probability fusion from different experts, it is assumed that the elements of risk affair are independent of each other. However, in reality the elements of one risk affair may have some interrelated relations. In this paper, the relations divided into two part: AND and OR (Fig. 4). The AND rule indicates the elements are parallel in a risk affair, every element is necessary to compose the whole risk affair. The OR rule indicates some elements only occur just one of it. For example, in customer risk, the two elements customer liquidation and demand variability, just one of them can be occurred. This AND and OR structure can be multi-level in a supply chain’s risk affair.

Risk Event

And

And

Or

Fig. 4. A AND and OR structure

Based on the combination rule (AND or OR) among the risk elements, the aggregated risk’s event probability is calculated as: 1. For AND rule (there are only And structure) in the left formula; For OR rule (there are only OR structure) in the right formula:

354

Y. Shi et al.

Pk ¼

I Y

Pk ¼ 1 

pi

i¼1

I Y

ð1  pi Þ

i¼1

where Pk is the probability of occurrence of risk’s event k and Pi is the probability of occurrence of risk element i. If the estimates involve both OR and AND rules together as shown in Fig. 4, the aggregated risk probabilities will be calculated based on the two formulas above. The probability of risk affair k is calculated as: Pk ¼

M Y

" pj 1 

j¼1

I Y

# ð1  p i Þ

i¼1

This formula can be seen as a combination of the first two formulas, the part of AND rule and the part of OR rule as a whole respectively, and then to be the input for the superior And rule. The superior similarly can also be the Or rule, the related formulas will also change based on the first two formulas. Specific problems should be analyzed differently. The final probability of occurrence of risk affair can show the risks probability distribution in a overall structure to the experts and management. The next goal is to obtain the score of the risk affair. Through the step 2, the fused probability of each element from a risk affair and the hazard index matrix decide by different expert which towards a risk’s event have already obtained. For example, a event has n elements. The m1 m2…mn is the fused probability (threshold filtering and normalization) from D-S calculation. The table as below. The final probabilities of different events of the risk need to be normalized. Risk event Probability

Element 1 m1

Element 1 m2

Element 1 m3

Element … …

Element n mn

A one row, n column probabilistic matrix P is defined as below P ¼ ½m1; m2; m3; m4; . . .mn n is the number of elements of a risk affair which has been screened by the threshold setting by expert. In the risk identification part, a hazard index matrix gathered by all experts has been initially generated. Some elements are generic in the risk affair, determined by general experience. But in a specific overall structure, the element’s probability collected by the experts is tiny, which means the element of the event that almost certainly didn’t happen. Through the threshold selection in the beginning of this section, the hazard index matrix will change accordingly. Here define the new hazard index matrix as below: where n is the number of the expert in the survey part, m is the number of the elements from a risk affair after screening. Next we define a initial score function B, calculating a preliminary score towards a risk’s event. The define as below: 2

r11 6 r21 R¼6 4 r... rm1

r12 r21 r... rm2















3 r1n r11 7 7 r... 5 rmn

A Dempster Shafer Theory and Fuzzy-Based Integrated Framework

2

r11 6 r21 B ¼ P  R ¼ ½m1; m2; m3; m4; . . .mn  6 4 r... rm1

r12 r21 r... rm2

355

3


r1n


r11 7 7


r... 5


rmn

The value of B is a normalized score. In the first section, a fuzzy evaluation set is defined. The evaluation set V according to the degree of harm = {low, less secondary, medium, significant, high} = {0.1, 0.3, 0.5, 0.7, 0.9}, experts can decide the hazard index score based on this set. The final score of a risk affair is defined as below (E is the importance weights towards the experts): S ¼ B  ET The value of B is also a normalized score between 0 and 1. If the final score of a risk affair is close to the value 1, it represents the event has significant risk based on the two parameters, probability and hazard. On the contrary, if the final score of a risk affair is close to the value 0, it represents the event is unimportant. In the same way, other risk events in the overall structure can be calculated. The table as below: Risk event Score Weight

Event 1 S1 W1

Event 2 S2 W2

Event … … …

Event n Sn Wn

The weights are normalized and are collected by the expert in the first section. The total risk score of the overall structure is calculated as: D¼

Xn i¼1

Si Wi ¼ S1  W1 þ S2  W2 þ


Sn  Wn

The total risk score D reflects the whole overall structure risk index, according to different scores, supply-chain risk will be divided into several levels. Different risk levels will give some reasonable advice to the management in the next step.

2.4

Risk Evaluation

According to different scores, overall structure risk will be divided into several levels: low, medium, high. Based on the different level, there are several suggestions. It is generally accepted, When the risk value is greater than 0.7, the risk level is high; When the risk value is between 0.3 and 0.7, the risk is medium; When the risk value is less than 0.3, the risk is low. Interval values are as follows:

356

Y. Shi et al.

D value D  0:3 0:3\D  0:7

Risk level Low Medium

D [ 0:7

High

Action required No further risk mitigation is required Risk mitigation required to reduce the risk level to low is optional, monitoring is required Risk mitigation required to reduce the risk level to medium or low. If the risk is not mitigated, monitoring and making a contingency plan is required

A supply-chain get a risk level through the above classification. The accuracy of the level can be defined into more detailed classification (No impact-no impact on the company, Small impact-small loss, Medium impact-cause short-term difficulties, serious impact-cause long-term difficulties, Disastrous impact-business interruption) as needed. The action required of each risk level should close to the specific business requirements based on the expert’s experience in different fields. Decision makers should take effective measures to prevent the occurrence of risk according to the level of warning signals. The data obtained from the previous step which consists of the information about the probability, score, weight from each event in a supply chain. The table as below: Risk event Probability Score Weight

Event 1 m1 S1 W1

Event 2 m2 S2 W2

Event … m… S…. W…

Event n mn Sn Wn

These data reflects the details of the supply-chain’s risk more intuitively and concretely. Supply-chain’s risk assessment is an important part of risk management. Because of the complexity and uncertainty of supply-chain risk, enterprises need to deal with risks and adjust business strategy immediately so as to reduce the losses and for guaranteeing the supply and the business with the continuous and the balanced.

3 Risk Assessment in Household Appliance Industry Supply Chain The proposed framework was applied in a household appliance manufacturing environment that is mostly based on the different parts of components from different suppliers in different geographical locations (Fig. 5). Many risk’s events can occur and affect the overall structure. Examples of risks include internal risk: supplier risks, customer risks, manufacture risk, transportation risks, commodity risks, management risk and others. And external risk: natural risk, political risk, market risk and others. Each risk event has many elements individually and elements are independent of each other. The order cancellation in customer risk, for example, has a large business impact on the marketing plan because the change of customer’s order can strongly influence the in a business. The table above is

A Dempster Shafer Theory and Fuzzy-Based Integrated Framework retailers Supplier

machining

Manufac turer

Supplier

Supplier

Raw material

Supplier

Co ns um er

Marketp lace

machining

machining

357

Material supply

wholesale corporaon

Department store

Fig. 5. A Supply chain structure in household appliance industry and a table of risk classification

established in the risk identification part’s first survey based on the expert’s experience, questionnaire and literature review. The information in table is possessed of stronger applicability and generality in current industry chain. For example, when another electronic product supply-chain needs to be evaluated, those information above can still be used with a little modification. According to the specific overall structure (see Fig. 6.), the second survey needs to establish the probability of each element from every risk event, and the hazard index towards each element. The establishment of this information is based on each expert. The customer risks, for example, has four elements: order cancellation, returns, customer liquidation, demand variability. Each element has a probability and a hazard index from different expert (Probabilistic data need to be normalized-the probability sum is 1). In addition, these data are dynamic, changing, most difficult to describe, and showed great ambiguity. Those data need to be judged according to the specific supply-chain by the expert. The table of information from one expert list below: Manufacture risk Probability Hazard index

Design changes 0.3 0.71

Quality issues 0.2 0.52

Technical limitations 0.25 0.31

Process variability 0.05 0.44

Forecast errors 0.2 0.12

(Probability set also includes a complete set consists of all risk events for the purpose of expressing the opinion of unclear from experts. In order to show more intuitive, the complete set are not written in the previous table.) The hazard index above reference to the evaluation set V = {low, less secondary, medium, significant, high} = {0.1, 0.3, 0.5, 0.7, 0.9}. There are 4 experts in this case. Another four experts’ assessment of probability is: {0.25, 0.25, 0.1, 0.1, 0.3}, {0.2, 0.25, 0.4, 0.05, 0.1}, {0.35, 0.1, 0.2, 0.15, 0.2}, {0.1, 0.3, 0.3, 0.1, 0.2}. The number of experts is determined according to the actual needs Another four experts’ assessment of hazard index score is:{0.62, 0.52, 0.61, 0.45, 0.31}, {0.47, 0.42, 0.28, 0.68, 0.21}, {0.42, 0.25, 0.75, 0.62, 0.51}, {0.52, 0.64, 0.18, 0.25, 0.37}. The data of different experts are not the

358

Y. Shi et al.

Risk classification

Internal risk

Risk event capacity constraints

supplier bankruptcy

quality issues

Customer risks

order cancellation

returns

customer liquidation

Manufacture risk

design changes

quality issues

technical limitations

Proximity to airports

quality of roads

logistics provider problems

Transportation risks Commodity risks Management risk

External risk

Risk element

Supplier risks

Price change

technology risk

quality risks

job security control

financial control

regulatory obligations

Natural risk

natural disaster

force majeure risk

Disease

Political risk

laws and regulations change

social, political turmoil

Exchange rate changes

Market risk

industry Volatility

competing risks

demand variability process variability

forecasting errors

Inflation

Fig. 6. A Supply chain structure in household appliance industry and a table of risk classification

same, after D-S data fusion, the integration probabilities of each element from manufacture risk and the hazard index matrix R which contain all impact index from every expert towards the risk event are established. Manufacture risk Probability

Design changes 0.34

Quality issues 0.24

Technical limitations 0.18

Process variability 0.09

Forecast errors 0.15

Hazard index matrix R is used to record the hazard assessment of each expert for different risk events. The risk event score formula from the previous chapter defines the final score according to the above two sets of data. 2

0:71 6 0:52 6 R¼6 6 0:31 4 0:44 0:12

0:62 0:52 0:61 0:45 0:31

0:47 0:42 0:28 0:68 0:21

0:42 0:25 0:75 0:62 0:51

3 0:52 0:64 7 7 0:18 7 7 0:25 5 0:37

B ¼ P  R ¼ ½0:48; 0:53; 0:40; 0:47; 0:44 S ¼ B  E T ¼ 0:469 The value B is the initial score of manufacture risk, and the value S is the final score of manufacture risk which is used to be a input of the whole supply-chain’s risk assessment. The set E is the wrights of each expert in the evaluation procedures. In this household appliance industry overall structure, the set E is {0.4, 3, 0.2, 0.1, 0.1} based on the importance of each expert in this industry. For example, the first value of set E may represent an authoritative expert in household appliance industry and the second value may reflect the corporate adviser’s weight. At the same way, the final score of Supplier risk, Customer risks, Transportation risk, Commodity risk, Management risk, Natural risk, Political risk, Market risk are calculated respectively as below: {0.781,

A Dempster Shafer Theory and Fuzzy-Based Integrated Framework

359

0.523, 0.627, 0.261, 0.392, 0.097, 0.124, 0.217}. Every risk event including the internal risk and the external risk has a risk score through the process above. Than use the weights that defined in the risk identification section to calculate the final score of the whole supply chain including the risk events involved. At the same time every risk event in the overall structure has a total probability based on the AND and OR rules in the previous chapter, and it intuitively reflects the probability of the occurrence of the event. (The probabilities of these events are normalized.)

Internal risk

Weights 0.8

External risk

0.2

Risk events Supplier risk Customer risks Transportation risk Manufacture risk Commodity risk Management risk Natural risk Political risk Market risk

Weights 0.3 0.2 0.05 0.2 0.1 0.15 0.1 0.4 0.5

Using the weighted summation formula of the previous chapter, the final risk score D is calculated by the weights in the table above as: 0.8[0.469 * 0.2 + 0.781 * 0.3 + 0.523 * 0.2 + 0.627 * 0.05 + 0.261 * 0.1 + 0.392 * 0.15] + .02[0.097 * 0.1 + 0.124 * 0.4 + 0.217 * 0.5] = 0.486. Based on the partition rule mentioned in risk evaluation part, this supply chain risk level is medium and the action required is risk mitigation required to reduce the risk level to low is optional, monitoring is required. Or according to the actual needs, the level should to be divided into more grades and more suggestions will be displayed. Total risk score

Risk level

0.486

medium

Risk event Probability Risk event Probability

Supplier 0.11 Natural 0.03

Action required risk mitigation required to reduce the risk level to low is optional, monitoring is required Customer Transportation Manufacture 0.16 0.10 0.21 Political Commodity Market 0.09 0.13 0.17

The related information of the household appliance industry supply is fully displayed table above. It shows the final risk level and the suggestions to the enterprise management, and the probabilities of each risk event from the supply chain also provide the crucial information about which risk event most likely to occur and which is small probability event.

360

Y. Shi et al.

4 Conclusions and Future Work This study proposed a framework for supply chain risk identification, data fusion, risk analysis and risk evaluation. The framework combines mathematics and decision making methods for an effective assessment of supply chain risks. The risk identification part was developed to identify the scores of the risks considering risk elements, risk probability and the hazard index towards each element from different experts. Given the individual and aggregated risk scores, decision makers can either see the final risk score of the supply chain and the risk level or the probability of each risk event and focus on the significant risks that can affect their business operations. The greatest problem in supply chain risk assessment is how to define a risk and how to relative accurately define the prior probability of each risk event. In this paper, the type of risk and structure is clearly defined. The prior probability of each risk event are collected by the different experts and fused by the Dempster Shafer theory. Because of the limitation of the Dempster Shafer theory, this paper believe that the risk elements are independent of each other from a risk event. And then use a AND and OR rules to merge each risk element from a risk event. This approach is feasible logically, but the accuracy of the algorithm needs further study to prove. Acknowledgments. This work was one of the “Cyberspace Security” key projects in People’s Republic of China.

References 1. Bloch, I.: Information combination operators for data fusion: a comparative review with classification. IEEE Trans. SMC Part A 26(1), 52–67 (1996) 2. Aqlan, F., Ali, E.M.: Integrating lean principles and fuzzy Bow-Tie analysis for risk assessment in chemical industry. J. Loss Prev. Process. Ind. 29, 39–48 (2014) 3. Yao, J.T., Raghavan, V.V., Wu, Z.: Web information fusion: a review of the state of the art. Inf. Fusion 9(4), 446–449 (2008) 4. Bogataj, D., Bogataj, M.: Measuring the supply chain risk and vulnerability in frequency space. Int. J. Prod. Econ. 1–2(108), 291–301 (2007) 5. Buchmeister, B., Kremljak, Z., Polajnar, A., Pandza, K.: Fuzzy decision support system using risk analysis. Adv. Prod. Eng. Manag. 1(1), 30–39 (2006) 6. Cao, Y., Chen, X.: An agent-based simulation model of enterprises financial distress for the enterprise of different life cycle stage. Simul. Model. Pract. Theory 20(1), 70–88 (2012) 7. Carvalho, H., Barroso, A.P., Machado, V.H., Azevedo, S., Cruzz-Machado, V.: Supply chain redesign for resilience using simulation. Comput. Ind. Eng. 62, 329–341 (2012) 8. Dinarvand, R.: A new pharmaceutical environment in Iran: marketing impacts. Iran J. Pharm. Res. 2, 1–2 (2010) 9. Farshchi, A., Jaberidoost, M., Abdollahiasl, A., et al.: Efficacies of regulatory policies to control massive use of diphenoxylate. Int. J. Pharmacol. 8(5), 459–462 (2012) 10. Naraharisetti, P., Karimi, I.: Supply chain redesign and new process introduction in multipurpose plants. Chem. Eng. Sci. 65(8), 2596–2607 (2010)

A Dempster Shafer Theory and Fuzzy-Based Integrated Framework

361

11. Jaberidoost, M., Abdollahiasl, A., Farshchi, A., et al.: Risk management in Iranian pharmaceutical companies to ensure accessibility and quality of medicines. Value Health 15 (7), A616–A617 (2012) 12. Jüttner, U., Christopher, M., Baker, S.: Demand chain management-integrating marketing and supply chain management. Ind. Mark. Manag. 36(3), 377–392 (2007) 13. Breen, L.: A preliminary examination of risk in the pharmaceutical supply chain (PSC) in the national health service (NHS), UK. J. Serv. Sci. Manag. 1(2), 6 (2008) 14. Goh, M., Lim, J.Y., Meng, F.: A stochastic model for risk management in global supply chain networks. Eur. J. Oper. Res. 182(1), 164–173 (2007) 15. Hult, G.T., Craighead, C.W., Ketchen Jr., D.J.: Risk uncertainty and supply chain decisions: a real options perspective. Decis. Sci. 41(3), 435–458 (2010) 16. Jacinto, C., Silva, C.: A semi-quantitative assessment of occupational risks using Bow-Tie representation. Saf. Sci. 48(8), 973–979 (2010) 17. Kumar, S., Tiwari, M.: Supply chain system design integrated with risk pooling. Comput. Ind. Eng. 64(2), 580–588 (2013) 18. Mele, F.D., Guillen, G., Espuna, A., Puigjaner, L.: An agent-based approach for supply chain retrofitting under uncertainty. Comput. Chem. Eng. 31(5), 722–735 (2007) 19. Neiger, D., Rotaru, K., Churilov, L.: Supply chain risk identification with value focused process engineering. J. Oper. Manag. 27(2), 154–168 (2009) 20. Norman, A., Jansson, U.: Ericson’s proactive supply chain risk management approach after a serious sub-supplier accident. Int. J. Phys. Distrib. Logist. Manag. 34(5), 434–456 (2004) 21. Pavlou, S., Manthou, V.: Indentifying and evaluating unexpected events as sources of supply chain risk. Int. J. Serv. Oper. Manag. 4(5), 604–617 (2008)

Knowledge Re-presentation and Reasoning

A Minimal Temporal Logic with Multiple Fuzzy Truth-Values Xinyu Li1 , Xudong Luo1,2(B) , and Jinsheng Chen1 1

Institute of Logic and Cognition, Sun Yat-sen University, Guangzhou, China [email protected] 2 School of Computer Science and Information Engineering, Guangxi Normal University, Guilin, China

Abstract. Temporal logic is a very important branch of non-classical logic, systematically studying formal reasoning over time, which actually is a kind of modal logic with the truth-value set of {0, 1}. However, in real life, propositions that concern with tense are not always absolutely true or false. To this end, this paper fuzzifies the minimal temporal logic system. Specifically, we fuzzify propositions’ truth values to six fuzzy linguistic truth values, and thus we build a new multi-valued temporal logic system. We also prove the completeness and soundness of our logic system. In addition, we illustrate our system by a real life example. Keywords: Knowledge representation and reasoning · Temporal logic · Multi-valued logic · Fuzzy logic · Linguistic variable · Completeness and soundness

1

Introduction

Temporal logic is a kind of modal logic, studying tense propositions and formal reasoning about tense propositions. The research on modal logic is originated by Aristotle in his book entitled “Physics”, where he gave a rough form of firstorder two-valued temporal logic. In modern time, it is logician Arthur Prior who started the study of temporal logic, and established the first formal axiom system of temporal logic [20]. Since 1960s, the results of temporal logic gradually increased considerably. For example, Pnueli [18] established a linear temporal first order logic with two temporal operator: next and until. In 1984, Emerson [9] established Branching Time Logic, which covers Linear Temporal Logic as its special case. In 1986, Moszkowski [16] established Interval Temporal Logic, which is also linear and first-order, but can only deal with finite sequence of statements. In 1990, Emerson [8] introduced Proposition Linear Temporal Logic and Branching Time Logic, and proved that these two kinds of temporal systems are modal complete. In 1997, Marx and Venema [14] introduced dimensional temporal logic, called Arrow Logic, discussing the relation theory of interval temporal logic and modal logic. In 2007, Akama [1] had noticed the uncertainty c Springer International Publishing AG 2017  L. Uden et al. (Eds.): KMO 2017, CCIS 731, pp. 365–377, 2017. DOI: 10.1007/978-3-319-62698-7 30

366

X. Li et al.

of future, and thus established a new temporal logic system which is threevalued, and proved its completeness and soundness. All in all, temporal logic has developed into a much more mature logic branch, and has an important impact in computer science and philosophy. The propositions in these existing various forms of temporal logic are still either absolutely true or absolutely false. Nonetheless, in real life it is not always the case, but multiple truth-values are required. For example, in weather forecast, for a prediction of future weather, it cannot be absolutely correct. So a kind of temporal logic with multiple fuzzy truth-values is required to make temporal logic more practical, worthy while and reliable. To this end, in this paper we try to fuzzify the minimal temporal logic system [15] to establish a new temporal logic system. More specifically, we will give its syntax, semantics, and axiom system, and prove that this system is complete and sound. There are some work on fuzzy temporal logic, but few of them deal with multiple fuzzy truth-values of temporal proportions. For example, [10,17] focus on fuzzy notations of time, [19] employs fuzzy logic to deal with time constraints of incomplete information, and [12] mainly discusses fuzzy temporal events and fuzzy temporal states defined on a linear time model. On the other hand, some studies (e.g., [4,11]) deal with modal logic with multiple fuzzy truth-values, but their modal logic is not a kind of temporal logic. The rest of the paper is structured as follows. Section 2 recaps some basic notations and methods of fuzzy set theory. Section 3 defines our fuzzy logic system. Section 4 proves the completeness and soundness of our fuzzy temporal logic. Section 5 gives a real life example to illustrate the usability of our logic system. Finally, Section 6 concludes the paper and points out future work.

2

Preliminaries

This section recaps some basic concepts and notations of fuzzy set theory. The following is Zadeh’s concept of fuzzy sets [21]: Definition 1 (Fuzzy set). A fuzzy set is a pair (U, μ), where U is a set (called the universe of the fuzzy set) and μ is a mapping from U to [0, 1] (called the membership function of the fuzzy set). And ∀x ∈ U , the value of μ(x) is called the membership degree of x in U . Baldwin [2] defined the following fuzzy linguistic truth-value set, which is used in some fuzzy logic systems (e.g., [11,13]): Definition 2 (Fuzzy Linguistic Truth). A linguistic truth-value set is defined as follows: LT T S = {very-true, true, fairly-true, fairly-false, false, very-false}

(1)

For convenience, we denote LT T St = {very-true, true, fairly-true},

(2)

Fuzzy Minimal Temporal Logic

367

Table 1. The membership functions of linguistic truth (where x ∈ [0, 1]) Linguistic truth Membership function Very-true

μvery-true (x) = μtrue (x)2

True

μtrue (x) = x

Fairly-true

μf airly-true (x) = μtrue (x)1/2

Fairly-false

μf airly-f alse (x) = μf alse (x)1/2

False

μf alse (x) = 1 − x

Very-false

μvery-f alse (x) = μf alse (x)2

Fig. 1. The curves of linguistic truth

LT T Sf = {very-false, false, fairly-false}.

(3)

The semantics of the terms in this term set are defined as shown in Table 1 and drawn graphically in Fig. 1. In real life, it is difficult for people to deal with continuous numbers, thus we can assume the membership functions of the linguistic truths is on the set of discrete points 0, 0.1, 0.2, . . . , 1.0. Thus, for example, we can represent τφ = true and τψ = f alse as follows: 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 + + + + + + + + + + , (4) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 τψ = + + + + + + + + + + . (5) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

τφ =

In fuzzy logic, t-norms () and t-connorms () are used to deal with conjunction and disjunction, respectively. The following is their definition [6,7]: Definition 3 (t-norm and t-connorm). A binary operator ◦ on [0, 1] is a triangular norm (t-norm), denoted as , if it satisfies: (1) commutativity: a ◦ b = b ◦ a,

368

X. Li et al.

(2) associativity: (a ◦ b) ◦ c = a ◦ (b ◦ c), (3) monotonicity: a ≤ b ⇒ a ◦ c ≤ b ◦ c, (4) unit element: a ◦ 1 = a. A binary operator ◦ on [0, 1] is a triangular connorm (t-connorm), denoted as , if it satisfies: (1), (2), (3) and (4 ) unit element: a ◦ 0 = a. A typical pair of typical t-norm  and t-conorm  is as follows: (x, y) = min{x, y},

(6)

(x, y) = max{x, y}.

(7)

The following definition gives the properties that a complement operator should obey: Definition 4 (Complement operator). A mapping C : [0, 1] → [0, 1] is a complement operator if it satisfies the following conditions: (1) C(0) = 1, C(1) = 0; (2) C(C(x)) = x; and (3) if a ≤ b then C(a) ≥ C(b). A common complement operator is: C(x) = 1 − x.

(8)

The following method can be used to extend a function on the crisp sets to the one on fuzzy sets. Definition 5 (Extension Principle). Suppose f is a function with n arguments x1 , . . . , xn . Let Ai be the fuzzy set of xi . Then we can extend function f to taking fuzzy arguments A1 , . . . , An , which result is a fuzzy set defined by: μB (y) = sup{μA1 (x1 ) ∧ . . . ∧ μAn (xn ) | f (x1 , . . . xn ) = y}, where sup denotes the supremum operation on a set. For convenience, the operation of the extension principle is denoted as ⊗. If the results of an operator on fuzzy linguistic true-values is not closed on the set of fuzzy linguistic true-values, we need the following linguistic approximation technology: Definition 6 (Linguistic Approximation). Given τ ∈ LT T S, τ1 ∈ LT T S being the closest to τ should satisfy: ∀τ2 ∈ LT T S, ED(τ, τ1 ) ≤ ED(τ, τ2 ), where ED is the Euclidean Distance, which is defined as follows: for two fuzzy linguistic terms τ1 and τ2 on U = {0.1, 0.2, · · · , 1}:  ED(τ1 , τ2 ) = (9) {(μτ1 (x) − μτ2 (x))2 | x ∈ U }. For convenience, the operation of linguistic approximation is denoted as .

Fuzzy Minimal Temporal Logic

3

369

Logic System

This section will present the syntax and semantics of our fuzzy temporal logic system (denoted as F Kt ). 3.1

Syntax

First we present the rules that define a logic formula in our logic system. Definition 7 (Language). Let P V be a countable set of proposition letters. A proposition formula in F Kt is defined by φ ::= p | ¬φ | φ ∧ ψ | P φ | F φ, where p ∈ P V , P is the weak past operator, meaning “happened once in past”; and F is the weak future operator, meaning “will happen once in future”. And strong future operator, denoted as G, which means “will happen all the time in future”, is defined as: Gφ =df ¬F ¬φ;

(10)

and strong past operator, denoted as H, which means “happened all the time in past”, is defined as: Hφ =df ¬P ¬φ.

(11)

We also need the following abbreviations: φ ∨ ϕ =df ¬(¬φ ∧ ¬ϕ),

(12)

φ → ϕ =df ¬φ ∨ ϕ.

(13)

The axioms (which are used as the start points of inference in our logic system) and proof rules (which are used for inference) are as follows: Definition 8 (Axioms and proof rules). Suppose that ϕ1 and ϕ2 are formulas in F Kt . The axiomatization of F Kt contains: (A1) (A2) (A3) (A4) (A5)

all propositional tautologies, G(p → q) → (Gp → Gq), H(p → q) → (Hp → Hq), p → GP p, p → HF p,

with the following rules of proof: (1) Modus ponens: if φ and φ → ϕ, then ϕ. (2) Uniform substitution: if ϕ is obtained from φ by uniformly replacing proposition letters in φ by arbitrary formulas, then φ implies ϕ. (3) Future Generalisation: if φ, then Gφ.

370

X. Li et al.

(4) Past Generalisation: if φ, then Hφ. Once we have a set of axioms and a set of inference rules, under certain conditions we can prove a logic formula. Formally, we have: Definition 9 (Proof ). Let Γ is a set of formulas in F Kt . In F Kt a formula φ is provable from Γ , denoted as Γ  φ, if there is a sequence of formulas < φ1 , · · · , φn > in F Kt such that φn = φ and for each φi (1 ≤ i < n), either φi is one of the axioms or φi ∈ Γ or φi can be obtained from the earlier items by applying the rules of proof. If Γ is empty, we abbreviate it as  φ. Intuitively, if from a set of propositions we can prove a proposition not only true but also false, then the set of proposition is inconsistent. Formally, we have: Definition 10 (Consistent). Suppose that Γ is a set of formulas in F Kt . Γ is inconsistent if there exists a formula φ such that Γ  φ and Γ  ¬φ. Γ is consistent if it is not inconsistent. 3.2

Semantics

Basically, the syntax of our logic is the same as that of crisp temporal logic. However, the semantics of our fuzzy one is quite different from that of crisp one. Informally, in each world (time point), we designate a fuzzy linguistic truth value for each proposition in temporal logic F Kt . Formally, we have: Definition 11 (Model). A fuzzy temporal model M is a triple (T, R, V ), where: (1) T is a non-empty finite set of all time possible worlds; (2) R is a binary relation on T ; and (3) V is a mapping from Var × T → LT T S, called a fuzzy truth evaluation, where Var = {p1 , . . . , pn } is a set of countable propositional variables, and LT T S is the linguistic truth-value set defined as formula (1). The rules for calculate the truth-value of a logic formula are as follows: For any formula φ, let Definition 12 (Truth evaluation of F Kt ). V (φ, M, t) be the fuzzy truth value of φ in world t. Then (1) V (¬φ, M, t) = (⊗(V (φ, M, t), C)); (2) V (φ ∧ ψ, M, t) = (⊗(V (φ, M, t), V (ψ, M, t), )); (3) V (φ ∧ (φ → ψ), M, t), denoted as τψ , is defined as follows: τψ = (τψ ),

(14)

where μτψ (y) = sup{μV (φ,M,t) (x)  μV (φ→ψ,M,t) (x, y)}; 





(4) V (F φ, M, t) = max{V (φ, M, t ) | t ∈ T, R(t, t ))}; and    (5) V (P φ, M, t) = max{V (φ, M, t ) | t ∈ T, R(t , t)}.

Fuzzy Minimal Temporal Logic

371

Table 2. The conjunctive operator of fuzzy temporal logic τφ

τψ Very-true

True

Fairly-true Fairly-false False

Very-false

Very-true

Very-true

True

Fairly-true Fairly-false False

Very-false

True

True

True

Fairly-true Fairly-false False

Very-false

Fairly-true Fairly-true Fairly-true Fairly-true Fairly-false False

very-false

Fairly-false Fairly-false Fairly-false Fairly-false Fairly-false False

Very-false

False

False

False

False

False

False

Very-false

Very-false

Very-false

Very-false

Very-false

Very-false

Very-false Very-false

Table 3. The complement operator of fuzzy temporal logic τφ

Very-true True Fairly-true Fairly-false False Very-false

τ¬φ Very-false False Fairly-false Fairly-true True Very-true

By Definition 12, from V (φ, M, t) and V (ψ, M, t), we can find the truthvalue of φ ∧ ψ as showed in Table 2, and that of ¬φ as showed in Table 3. Now we give an example of how we calculate the truth evaluation of τφ ∧ τψ . Suppose that τφ = true, which is defined as formula (10), and τψ = f alse, which is defined as formula (11). Then by the extension principle (see Definition 5) and the evaluation of conjunction operator (see Definition 12), we have: μφ∧ψ (0.1) = max{ min(μφ (0.1), μB (0.1)), min(μφ (0.1), μB (0.2)), min(μφ (0.1), μψ (0.3)), min(μφ (0.1), μψ (0.4)), min(μφ (0.1), μψ (0.5)), min(μφ (0.1), μψ (0.6)), min(μφ (0.1), μψ (0.7)), min(μφ (0.1), μψ (0.8)), min(μφ (0.1), μψ (0.9)) min(μφ (0.1), μψ (1.0)), min(μφ (0.2), μψ (0.1)), min(μφ (0.3), μψ (0.1)) min(μφ (0.4), μψ (0.1)), min(μφ (0.5), μψ (0.1)), min(μφ (0.6), μψ (0.1)), min(μφ (0.7), μψ (0.1)), min(μφ (0.8), μψ (0.1)), min(μφ (0.9), μψ (0.1)), min(μφ (1.0), μψ (0.1))} = max{ min{0.1, 0.9}, min{0.1, 0.8}, min{0.1, 0.7}, min{0.1, 0.6}, min{0.1, 0.5}, min{0.1, 0.4}, min{0.1, 0.3}, min{0.1, 0.2}, min{0.1, 0.1}, min{0.1, 0}, min{0.2, 0.9}, min{0.3, 0.9}, min{0.4, 0.9}, min{0.5, 0.9}, min{0.6, 0.9}, min{0.7, 0.9}, min{0.8, 0.9}, min{0.9, 0.9}, min{1.0, 0.9}} = 0.9.

372

X. Li et al.

The calculation of μφ∧ψ at other points are similar to the above. Then we have τφ∧ψ =

0 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 + + + + + + + + + + . 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Finally, by Linguisitic Approximation (see Definition 6), τφ∧ψ is nearest to linguistic truth-value false. That is, the truth evaluation of φ ∧ ψ is false. In Definition 12, we did not give the formulas for calculating V (Gφ, M, t), V (Hφ, M, t) and V (φ → ψ, M, t), but we can use formulas (10), (11) and (13) to transfer them into the ones that can be calculated according to the formulas given in Definition 12. That is: 









V (Gφ, M, t) = min{V (φ, M, t ) | t ∈ T, R(t, t )}, 

V (Hφ, M, t) = min{V (φ, M, t ) | t ∈ T, R(t , t)}.

(15) (16)

Definition 13 (Satisfaction). Given a temporal model M = (T, R, V ), any proposition p in M is satisfiable in world t, denoted as M, t  p, which can be defined recursively as follows: (1) M, t  ψ iff V (ψ, M, t) ∈ LT T St or (V (φ → ψ, M, t) ∈ LT T St and V (φ, M, t) ∈ LT T St ). (2) M, t  ¬φ iff V (φ, M, t) ∈ LT T Sf . (3) M, t  φ ∧ ψ iff M, t  φ and M, t  ψ. (4) M, t  F φ iff there exists at least one u ∈ T and R(t, u) such that M, u  φ. (5) M, u  P φ iff there exists at least one t ∈ T and R(t, u) such that M, t  φ. By the above definition and formulas (10)–(13), we have: (1) (2) (3) (4)

M, t  Gφ iff for all u ∈ T and R(t, u), M, u  φ. M, u  Hφ iff for all t ∈ T and R(t, u), M, t  φ. M, t  φ ∨ ψ iff M, t  φ or M, t  ψ. M, t  φ → ψ iff M, t  ¬φ or M, t  ψ.

Definition 14 (Semantic consequence). Suppose that Γ is a set of formulas in F Kt and φ is a formula in F Kt . φ is the semantic consequence of Γ (denoted by Γ  φ) if for all valuations, models and possible worlds such that every formula in Γ is satisfiable, φ is satisfiable. If Γ is empty, then we abbreviate it as  φ and say that φ is valid.

4

Soundness and Completeness

This section will prove the soundness and completeness of our logic system.

Fuzzy Minimal Temporal Logic

4.1

373

Soundness

The soundness of a logic system means that if a formula in the logic can be proved from its axioms of the logic by its inference rules, then the formula is a semantic consequence. Formally, we have: Theorem 1 (Soundness). Fuzzy minimal temporal logic system is sound with respect to the class of all frames, meaning if Σ  φ then Σ  φ. Proof. First, we will prove that the rules of proof preserve validity: (i) Suppose modus ponens does not preserve validity, then we have  p → q,  p but  q. Hence, there is a model M = (T, R, V ) such that M, t  p → q, M, t  p and M, t  q. Since M, t  p → q, M, t  p or M, t  q. But we have M, t  p and M, t  q, contradicting the result above. Thus, modus ponens preserves validity. (ii) It is easy to verify that future generalisation preserves validity. Suppose  p, then by Definition 14, for any model M = (T, R, V ) and any t ∈ T , M, t  p. Then  Gp. If not, there is a model and a world in this model such that p is not satisfiable, which contradicts to our hypothesis. So, future generalisation preserves validity. The proof of past generalisation is similar. Now we will prove that every axiom is valid: (i) Suppose φ is propositional tautology, then in every possible world t, the truth value of φ is in LT T St . (ii) To prove  G(p → q) → (Gp → Gq), we need prove that for every possible world t in any model M, V (G(p → q) → (Gp → Gq), M, t) ∈ LT T St , which means that when V (G(p → q), M, t) ∈ LT T St , definitely V ((Gp → Gq), M, t) ∈ LT T St . So, when V (G(p → q), M, t) ∈ LT T St , in possible world t satisfying R(t, t ) we have V ((p → q), M, t ) ∈ LT T St . That is, when V (p, M, t ) ∈ LT T St , there must be V (q, M, t ) ∈ LT T St . When V (Gp, M, t) ∈ LT T St , there must be V (Gq, M, t) ∈ LT T St , and thus V ((Gp → Gq), M, t) ∈ LT T St . So, V (G(p → q) → (Gp → Gq), M, t) ∈ LT T St is proved. (iii)  H(p → q) → (Hp → Hq) can be proved similarly to (ii). (iv) We prove  φ → GP φ. For any φ, on the one hand, suppose V (φ, M, t) ∈ LT T St . In possible world t which is accessed by possible world t, by Definition 12, V (P φ, M, t ) = max{V (φ, M, t ) | R(t , t )}, and t access to t . So since V (φ, M, t) ∈ LT T St , V (P φ, M, t ) must be in LT T St . When we look back to see possible world t, by Definition 12 we know V (GP φ, M, t) = min{V (P φ, M, t ) | R(t, t )}, which means V (GP φ, M, t) ∈ LT T St . On the other hand, if V (φ, M, t) ∈ LT T Sf then V (¬φ, M, t) ∈ LT T St . So,  φ → GP φ. (v) We prove  φ → HF φ. For any φ, on the one hand, suppose V (φ, M, t) ∈ LT T St . In possible world t , which is accessed by possible world t, by Definition 12, V (F φ, M, t ) = max{V (φ, M, t ) | R(t , t )}

374

X. Li et al.

and t access to t . So since V (φ, M, t) ∈ LT T St , V (F φ, M, t ) must be in LT T St . When we look back to see possible world t, by Definition 12, V (HF φ, M, t) = min{V (F φ, M, t ) | R(t , t)}, which means V (HF φ, M, t) ∈ LT T St . On the other hand, if V (φ, M, t) ∈   LT T Sf , V (¬φ, M, t) ∈ LT T St . So,  φ → HF φ. 4.2

Completeness

Now we are going to discuss the completeness of our logic, meaning that if a formula is correct semantically, then it is correct syntactically. Before giving the completeness theorem of our logic, we need more concepts. Definition 15 (Maximal consistent set). A set of formulas Γ is maximal consistent if and only if (1) Γ is consistent; and (2) Γ is maximal: there does not Γ  such that Γ ⊂ Γ  and Γ  is consistent. Definition 16 (Fuzzy consistency). Let Γ be a set of formulas, we say Γ is consistent, if for any formula φ ∈ Γ , it is impossible that V (φ, M, t) ∈ LT T St and V (¬φ, M, t) ∈ LT T St . Definition 17 (Canonical model). The canonical model Mc = (T c , Rc , V c ) is defined as follows: (1) T c is the set of all possible worlds; (2) Rc is a transitive, irreflexive, and asymmetrical binary relation on T c such that (t, t ) ∈ Rc if for all formulas ψ, t  F ψ implies t  ψ, and t  P ψ implies t  ψ; and (3) V c : Var × T → LT T S is a fuzzy truth evaluation such as V c (p, M, t) ∈ LT T St if and only if t  p, where p ∈ Var = {p1 , . . . , pn } (i.e., a set of countable propositional variables). In order to prove completeness, we also need some lemmas as follows: Lemma 1. Suppose that Γ is a maximally consistent set, then: (1) for all formulas φ: φ ∈ Γ or ¬φ ∈ Γ ; (2) for all formulas φ, ϕ: φ ∨ ϕ ∈ Γ iff φ ∈ Γ or ϕ ∈ Γ . The proof of the above lemma can be found in [3]. Lemma 2 (Lindenbaum’s Lemma [5]). If Γ is consistent, then there exists a Γl is maximal and consistent, where Γ ⊆ Γl . Lemma 3 (Truth Lemma). Suppose that t ∈ T c . For any φ, t  φ iff (Mc , t)  φ.

Fuzzy Minimal Temporal Logic

375

Table 4. Truth value of φ Day City c1

c2

c3

d1

Fairly-true True

d2

True

Very-true False

d3

Very-true

True

c4

Fairly-false True Very-true

Fairly-false Very-false

Table 5. Truth value of ψ Day City c1

c2

d1

True

Fairly-true Fairly-false False

d2

Fairly-true False

False

d3

Very-true

Undecided Very-false

True

c3

c4 Fairly-false

Proof. We prove by induction on the length of φ. The base case follows from the definition of V c . The boolean cases follow from Lemma 1. It remains to deal with the modalities. Suppose that φ = F ϕ. Mc , t  F ϕ iff ∃t ∈ T c (tRc t ∧ Mc  ϕ) iff (by induction hypothesis) ∃t(tRc t ∧ t  ϕ) iff t  F ϕ. The proof for the case φ = P ϕ is similar to the proof above.   Now we can give our completeness theorem. Theorem 2 (Completeness). Fuzzy minimal temporal logic system is complete, i.e., Γ  φ implies Γ  φ. Proof. To prove it, we need prove the converse-negative proposition: if Γ  φ, then Γ  φ. If Γ  φ, we can know Γ ∪ {¬φ} is a consistent set, by Lindenbaum’s Lemma (i.e., Lemma 2), there is a maximal consistent set Γl . We will use canonical model to find this maximal consistent set Γl to make Γl ⊇ Γ ∪{¬φ}. By Truth Lemma (i.e., Lemma 3) and fuzzy consistent, we know V c (φ, M c , t) ∈ LT T St iff φ ∈ Γl , then ¬φ ∈ Γl , ¬φ ∈ Γ . Thus Γ  φ is proved.  

5

Example

Let us consider an example in real life. Suppose Mary plans to have a holiday of three days d1 , d2 , and d3 . During the holiday, she plans to visit two different cities within two days, and there are four optional cities c1 , c2 , c3 , and c4 . She likes a city which is not rainy and above 20 Celsius degree. Now suppose we know the truth-values of that these four cities in three days are “not raining” (denoted as φ) and “above 20 Celsius degree” (denoted as ψ) as shown in Tables 4 and 5, and the accessible relation between these four cities are: R(c1 , c2 ), R(c1 , c4 ), R(c2 , c3 ), R(c2 , c4 ), R(c3 , c4 ).

376

X. Li et al.

We give an example to show how to calculate the fuzzy linguistic truth-value of a modal proposition. By Definition 12 and Table 2, we have: V (Gφd1 , M, c1 ) = min{V (φd1 , M, c1 ), V (φd2 , M, c2 )} = min{f airly-true, very-true} =f airly-true, V (Gφd2 , M, c1 ) = min{V (φd2 , M, c1 ), V (φd3 , M, c2 ) = min{true, true} =true, V (Gψd1 , M, c1 ) = min{V (ψd1 , M, c1 ), V (ψd2 , M, c2 )} = min{true, f alse} =f alse, V (Gψd2 , M, c1 ) = min{V (ψd2 , M, c1 ), V (ψd3 , M, c2 ) = min{f airly-true, true} =f airly-true. Then again by Definition 12 and Table 2, we have: V (Gφd1 ∧ Gψd1 ), M, A) = min{V (Gφd1 , M, c1 ), V (Gψd1 , M, c1 )} = min{f airly-true, f alse} V (Gφd2

=f alse, ∧ Gψd2 ), M, A) = min{V (Gφd2 , M, c1 ), V (Gψd2 , M, c1 )} = min{true, f airly-true} =f airly-true.

Accordingly, Mary should spend the holiday in city c1 on day 2.

6

Conclusions

On the one hand, temporal logic has proved to be an important kind of modal logic since 1960s. On the other hand, the development of fuzzy logic lets us see that the original two-valued logic system can have much more widespread applications in real-life if they are extended by fuzzy logic theory. As a result, in this paper we fuzzified the minimal temporal logic system by extending its truth-value set of {0, 1} to the one of six linguistic truth-values. In particular, we define the semantics of this logic system, and we also prove that this logic system is soundness and completeness. Moreover, to show the applicability of our new system, we discuss a real-life example with our system. However, fuzzy temporal logic system can be more applicable not only in real life but also in computer science. So in further work, it is worth exploring how to apply fuzzy temporal logic into different areas, specially the area of Artificial Intelligence.

Fuzzy Minimal Temporal Logic

377

Acknowledgments. This research was partially supported by the Natural Science Foundation of Guangdong Province, China (No. 2016A030313231) and the National Fund of Social Science (No. 14ZDB015).

References 1. Akama, S., Nagata, Y., Yamada, C.C.: A three-valued temporal logic for future contingents. Log. Anal. 50(198), 99–111 (2007) 2. Baldwin, J.F.: A new approach to approximate reasoning using a fuzzy logic. Fuzzy Sets Syst. 2, 309–325 (1979) 3. Blackburn, P., De Maarten, R., Venema, Y.: Modal logic. Bull. Symb. Logic 8(2), 286–289 (2002) 4. Chen, J., Luo, X.: A multi-linguistic-valued modal logic. In: Kang, B.H., Bai, Q. (eds.) AI 2016. LNCS (LNAI), vol. 9992, pp. 317–323. Springer, Cham (2016). doi:10.1007/978-3-319-50127-7 26 5. Crossley, J.N., Ash, C.J., Brickhill, C.J., Stillwell, J.C., Williams, N.H.: What is Mathematical Logic?. Oxford University Press, London (1972) 6. Dubois, D., Prade, H.: A class of fuzzy measures based on triangular norms: A general framework for the combination of uncertain information. Int. J. Gen. Syst. 8(1), 43–61 (1982) 7. Dubois, D., Prade, H.: Criteria aggregation and ranking of alternatives in the framework of fuzzy set theory. Stud. Manage. Sci. 20, 209–240 (1984) 8. Emerson, A.: Temporal and model logic. In: van Leeuwen, J. (ed.) Handbook of Theoretical Computer Science, Volume B: Formal Models and Sematics, pp. 995– 1072. MIT Press, Cambridge (1990) 9. Emerson, E.A., Sistla, A.P.: Deciding full branching time logic. Inf. Control 61(84), 175–201 (1984) 10. Frigeri, A., Pasquale, L., Spoletini, P.: Fuzzy time in linear temporal logic. ACM Trans. Comput. Logic 15(4), 1–22 (2014) 11. Jing, X., Luo, X., Zhang, Y.: A fuzzy dynamic belief logic system. Int. J. Intell. Syst. 29(7), 687–711 (2014) 12. Moonand, S.I., Lee, K.H., Lee, D.: Fuzzy branching temporal logic. IEEE Trans. Syst. Man Cybern. Part B Cybern. 34(2), 1045–1055 (2004) 13. Luo, X., Zhang, C., Jennings, N.R.: A hybrid model for sharing information between fuzzy, uncertain and default reasoning models in multi-agent systems. Int. J. Uncertain. Fuzziness Knowl. Based Syst. 10(4), 401–450 (2002) 14. Marx, M., Venema, Y.: Multidimensional modal logic. J. Appl. Logic 4, 1–9 (1997) 15. Mchedlishvili, L.I.: Minimal temporal logic with a modalized temporal operator. In: Studies in Logic & Semantics, pp. 71–89 (1981) 16. Moszkowski, B.: Executing Temporal Logic Programs. Cambridge University Press, Cambridge (1986) 17. Mukherjee, S., Dasgupta, P.: A fuzzy real-time temporal logic. Int. J. Approx. Reason. 54(9), 1452–1470 (2013) 18. Pnueli, A.: The Temporal Logic of Programs. Weizmann Science Press of Israel, Jerusalem (1977) 19. Poli, V.S.R.: Fuzzy temporal predicate logic for incomplete information. In: IEEE International Conference on Fuzzy Theory and Its Applications, pp. 86–90 (2015) 20. Prior, A.N.: Time and modality: Being the John Locke lectures for 1955–6 delivered in the University of Oxford. Clarendon Press (1957) 21. Zadeh, L.A.: Fuzzy sets. Inf. Control 8, 338–353 (1965)

Modified Similarity Algorithm for Collaborative Filtering Kaili Shen ✉ , Yun Liu, and Zhenjiang Zhang (

)

Laboratory for Communications and Information System, Beijing Jiaotong University, Beijing, China {14120206,liuyun,zhangzhenjiang}@bjtu.edu.cn

Abstract. Collaborative filtering (CF) is one of the most applied techniques in recommendation systems and has been widely used in various conditions. The accuracy of the CF method requires further improvement despite the method’s advancement. Numerous issues exist in traditional CF recommendation, such as data scarcities, cold start and scalability problems. Since the data’s sparsity, the nearest neighbors formed around the target user would cause the loss of the infor‐ mation. When the new recommended system started, the information about eval‐ uating is poor, the result of recommend is also poor. About the scalability prob‐ lems, under the background of big data, the complexity and accuracy of calcula‐ tion is facing a great challenge. Global information has not been fully used in traditional CF methods. The cosine similarity algorithm (CSA) uses only the local information of the ratings, which may result in an inaccurate similarity and even affect the target user’s predicted rating. To solve this problem, a modified simi‐ larity algorithm is proposed to provide high accuracy recommendations, and an adjustment factor is added to the traditional CSA. Finally, a series of experiments are performed to validate the effectiveness of the proposed method. Results show that the recommendation precision is better than those of traditional CF algo‐ rithms. Keywords: Collaborative filtering · Similarity measures · MovieLens

1

Introduction

Retrieving relevant information has become increasingly challenging with the devel‐ opment of network and technologies. Information overload causes people to spend much time in searching for valuable information online. Recommendation systems analyze user data, learn user preferences, and recommend items of interest to users. They solve information overload (Polatidis, et al. 2015). Recommendation systems have been used in many companies, such as Amazon and TaoBao, to increase sales and profits. They can be divided into four categories: knowledge-based recommendation, collaborative filtering (CF), content-based recommendation, and hybrid recommendation. CF technologies can be categorized into user- and item-based CF algorithms (Bobadilla, et al. 2012). User-based CF algorithm is the first automatic CF technology. Item-based CF was first introduced by Sarwar et al., and has been used by Amazon.com. On the basis of these studies, we propose a modified similarity measure aimed to enhance the accuracy of recommendations. The said method is based on the traditional © Springer International Publishing AG 2017 L. Uden et al. (Eds.): KMO 2017, CCIS 731, pp. 378–385, 2017. DOI: 10.1007/978-3-319-62698-7_31

Modified Similarity Algorithm for Collaborative Filtering

379

cosine similarity algorithm (CSA). An adjustment factor is added to the traditional CSA. Experiments are performed to prove that the modified method can improve the accuracy of user similarity and obtain better recommendation results. The subsequent parts of the paper are as follows. Section 2 contains a brief intro‐ duction of the CF. Section 3 describes the proposed method. Section 4 presents the experiments and explains the evaluations. Section 5 states the conclusion and future work.

2

Related Work

The current section discusses the implementation of the CF recommendation algorithm and introduces the different similarity measures in detail. 2.1 Collaborative Filtering The CF recommendation algorithm in the e-commerce recommendation system can be divided into the rating matrix representation, the generation of nearest neighbors, and the recommendation. The rating matrix comprises the set of users U = {u1 , u2 , … , um }, the set of items I = {i1 , i2 , … , in }, and rating Ru,i, which represents the rating of user u for item i. The rows define users and the columns define items. The generation of recommendation has three steps, namely, similarity computation, generation of the nearest neighbor, and calculation of the prediction rating. Similarity computation involves calculating the similarity between users or items according to the rating matrix. The similarity algorithm is discussed in detail in Sect. 2.2. Similarities are arranged from largest to smallest. The top-k is selected as the nearest neighbor set N = {N1, , N2 , … , Nk }, where k indicates the number of nearest neighbors. The ratings and similarities of all neighbors are combined to predict the target user’s rating. The target user is u, the target item is i, and the predicted rating is ∑ Pu,i = Ru +

l∈Nu

( ) sim(u, l) × Rl,i − Rl ∑ l∈Nu |sim(u, l)|

(1)

Where Ru is the average of the user u’s ratings, Nu is the nearest neighbor of user u, and sim(u, l) is the similarity value of users u and l . After calculating the target user’s predicted ratings for the target items, the predicted ratings are ranked in descending order. The first n items are picked out to generate the top-n recommendation list, which is recommended to the target user. 2.2 Similarity Measures Several traditional similarity metrics are subsequently introduced in this section.

380

K. Shen et al.

Cosine Similarity Algorithm (CSA): User rating data can be viewed as a vector in the n dimensional item space. Cosine similarity uses the cosine between two user-item vectors to measure the similarity between two users. ( ) ( ) sim ui , uj = cos u⃖⃖⃗i , u⃖⃖⃗j = √∑

∑n k=1,k∈Ii,j

n k=1,k∈Ii,j

R2u ,k

Rui ,k Ruj ,k √∑ n k=1,k∈Ii,j

i

R2u ,k

(2)

j

Rui ,k, is the rating of user ui for item k. Ruj ,k is the rating of user uj for item k. Ii,j is the item set that ui , uj both rated, where Ii,j = II ∩ IJ. II is the item set user ui rated and IJ is the items set user uj. k is the number of items that ui , uj both rated.

Adjusted CSA (ACSA): Different users have different rating habits. Thus, deviation exists with the basic cosine similarity. The ACSA uses average rating to offset the user rating habit difference.

∑n

) ( sim ui , uj = √ ∑n

(Rui ,k − Rui )(Ruj ,k − Ruj ) √∑ n (Rui ,k − Rui )2 (Ruj ,k − Ruj )2 k=1,k∈I

k=1,k∈Ii,j

k=1,k∈Ii,j

(3)

i,j

Where Rui ,Ruj are respectively the average rating scores of users ui and uj of items that ui , uj both rated. Pearson Correlation Coefficient (PCC): PCC is also called correlation-based simi‐ larity. PCC

sim

( ) ui , uj = √ ∑n

∑n

(Rui ,k − Rui )(Ruj ,k − Ruj ) √∑ n (Rui ,k − Rui )2 (Ruj ,k − Ruj )2 k=1,k∈I

k=1,k∈Ii,j

k=1,k∈Ii,j

(4)

i,j

Where Rui , Ruj represent the average ratings of users ui and uj. The values range from –1 to 1. The difference between the several traditional similarity metrics mentioned above is not huge. CSA is the basic method to evaluate the difference between the uses, but the defect of this algorithm is that it doesn’t consider the different standard for evaluation in different users. To overcome this defect, ACSA and PCC take the decentralized into account, but there is also some difference between the two methods. ACSA considers more about the influence that each user’s average score, who grades in the item i. And PCC considers more about the average score on the item i.

3

Proposed Method

A detailed description of the proposed method is introduced in this section.

Modified Similarity Algorithm for Collaborative Filtering

381

3.1 Adjustment Factor The number of co-rated items reflects connection between users. The high number of co-rated items may show high similarity. The traditional similarity metrics do not consider the number of co-rated items. To solve this problem, the concept of adjustment factor, which considers the number of co-rated items to compensate for the shortcoming of the traditional similarity calculation method, is introduced in this paper. The adjust‐ ment factor is calculated as follows: ( ) ω ui , uj =

(

)𝜑 1 ( ) Jaccard ui , uj

(5)

Jaccard uses all the ratings information provided by a pair of users. The value ranges from –1 to 1. However, it uses only the numerical values of the ratings and suffers from few co-rated items problem. | | ( ) ||Iui ∩ Iuj || Jaccard ui , uj = | | |Iui ∪ Iuj | | |

(6)

| | | | Where|Iui ∩ Iuj |is the number of co-rated items of users i and j. |Iui ∪ Iuj | is the number | | | | of rated items of users i and j. ) ) ( ( Jaccard ui , uj is calculated by Eq. (6) and Jaccard ui , uj ∈ [0, 1]. Thus, 1 ( ) ∈ [1, +∞). φ is a weight index of the adjustment factor and Jaccard ui , uj φ ∈ [0, +∞), and we use the SA algorithm to obtain a relatively accurate value in the ) ( 1 next experiments. When Iui ∩ Iuj is larger, Jaccard ui , uj is larger, then ( ) Jaccard ui , uj ( ) becomes smaller and ω ui , uj is smaller. Thus, a negative correlation between Iui ∩ Iuj ( ) and ω ui , uj exists. 3.2 Improved Similarity Algorithm

( ) ( )ω(u ,u ) simproposed ui , uj = sim ui , uj CSAi j

(7)

( ) The proposed user similarity algorithm simproposed ui , uj is based on the CSA ) ( ( ) sim ui , uj CSA. It adds an adjustment factor ω ui , uj to the result. The CSA is a basic and ) ( commonly used technology to measure similarity in CF. Because sim ui , uj CSA ∈ [0, 1], ( ) ( ) according the exponential curve, when ω ui , uj is smaller, simproposed ui , uj is larger. Therefore, simproposed becomes higher as Iui ∩ Iuj becomes larger. As a result, a positive correlation between the co-rated items and the similarity exists to obtain more accurate similarities between users.

382

4

K. Shen et al.

Experimental Evaluation

Experiments were conducted to verify the result of the proposed algorithm. The experi‐ ments were based on the MovieLens dataset and four error metrics. These experiments were used to verify the precision of the proposed similarity method compared to the traditional CF algorithms. 4.1 Dataset The experiments were conducted using the MovieLens 100 K dataset (http://Movie‐ Lens.umn.edu) collected by the University of Minnesota. The MovieLens dataset, a personalized movie recommendation website, was constructed in 1997 by the Grouplens research team. MovieLens is an open system and it provides researchers with three data sets of different orders of magnitude. The MovieLens 100 K dataset contains 1682 movies, 943 users, and 100,000 ratings. The degree of drainage is 1–100000/ (1682 * 943) = 0.936953. Each user rated at least 20 times. The dataset contains three small datasets, which are the user’s attribute table [user id], the film attribute table [item id], and the user’s rating data table [rating]. Each item belongs to several types of movies, such as action, comedy, and crime. A total of 19 types exist. Integers 1–5 were used to rate the movies, with 5 as the highest score. The dataset used in this paper was divided into two for the training set (80%) and the test set (20%). 4.2 Measures The parameters, namely, MAE, precision, and recall, are used to measure the accuracy of the recommendations. MAE is a measure of the deviation of recommendations from their true user-specified values. The MAE value can be obtained by calculating the deviation between the actual and predicted scores between users. The lower the values of MAE, the higher the accu‐ racy of the recommended algorithm. MAE can be calculated by Eq. (8): MAE =

∑N | p − qu.i || i=1 | u.i N

(8)

where N is the number of test ratings, pu.i is the predicted rating of user u for item i, and qu.i is the real rating of user u for item i. If the predicted rating is from 3 to 5, the item will be recommended to users. On the basis of this assumption, four different conditions exist: true positive (TP), false negative (FN), false positive (FP), and true negative (TN) (Chen, et al. 2015). TP recommends to users both the predicted and actual ratings. FN recommend the actual rating, but not the predicted rating. FP is the opposite of FN. If the predicted and actual ratings are not recommended to users, TN occurs. The higher the precision and recall values, the higher the accuracy of the recommended algorithm.

Modified Similarity Algorithm for Collaborative Filtering

383

Precision and recall are defined (9) and (10): precision = recall =

TP TP + FP

TP TP + FN

(9)

(10)

4.3 Experimental Analysis To display the results of the proposed method better, several CF algorithms are compared. In the experiment, the adjustment factor weight was set to φ = 0.75. The nearest neighbor set was K = {10, 20, 40, 60, 80, 100}. The experimental results are shown in Figs. 1, 2 and 3. Figure 1 presents he MAE results for the MovieLens dataset, Fig. 2 presents he precision results, and Fig. 3 presents the recall results.

Fig. 1. MAE results

Figure 1 shows that the horizontal axis represents the number of nearest neighbors and the vertical axis represents the MAE values. The figure and data show the gradual decline of the overall trend. The proposed algorithm committed less errors compared to the other algorithms. The MAE results consistently declined with the increasing number of nearest neighbors. Figures 2 and 3 show that the performance of the proposed algorithm was the best as indicated by its high recommendation accuracy. The experimental results clearly indicated that the proposed method can provide a more accurate recommendation than the others. The adjustment factor made the calculation of the user similarity more accu‐ rate. When the selected neighbor set was small, the results obtained were poor, and sparse data problem existed. However, the proposed CF recommendation method produced better results even with the sparse data problem.

384

K. Shen et al.

Fig. 2. Precision results

Fig. 3. Recall results

5

Conclusions and Future Work

In this paper, a new CF algorithm was proposed based on the traditional similarity measure. An adjustment factor was introduced to improve the similarity value. Local user ratings and global information were both considered. The proposed scheme was compared with other CF algorithms, such as CSA and PCC, and experiments validated the good performance of the proposed scheme. The proposed method can provide more suitable recommendations than the other CF algorithms. The proposed algorithm demonstrated a better performance, but its accuracy still requires enhancement. Future research may be focused on the following issues to effectively integrate and utilize the dataset in many aspects:

Modified Similarity Algorithm for Collaborative Filtering

385

• User activity can be considered in the future work, provided that the user activity includes login time, time interval, and operations habits in different periods. • A better solution to the problem of cold start should be devised. • User group recommendation is also a possible research direction in the future. Users of the recommendation system sometimes want to receive recommendations as a group instead of as individuals.

References 1. Polatidis, N., Georgiadis, C.K.: A multi-level collaborative filtering method that improves recommendations. Expert Syst. Appl. 48, 100–110 (2015) 2. Cui, Y., Song, S., He, L., et al.: A collaborative filtering algorithm based on user activity level. In: 2012 Fifth International Conference on Business Intelligence and Financial Engineering, pp. 80–83. IEEE (2012) 3. Chen, D.E.: The collaborative filtering recommendation algorithm based on BP neural networks. In: International Symposium on Intelligent Ubiquitous Computing and Education, pp. 234–236. IEEE (2009) 4. Jia, C.X., Liu, R.R.: Improve the algorithmic performance of collaborative filtering by using the interevent time distribution of human behaviors. Physica A: Stat. Mech. Appl. 436, 236– 245 (2015) 5. Bobadilla, J., Ortega, F., Hernando, A., et al.: Generalization of recommender systems: Collaborative filtering extended to groups of users and restricted to groups of items. Expert Syst. Appl. 39(1), 172–186 (2012) 6. Chen, M.H., Teng, C.H., Chang, P.C.: Applying artificial immune systems to collaborative filtering for movie recommendation. Adv. Eng. Inform. 29(4), 830–839 (2015) 7. Mohammed, B., Mouhoub, M., Alanazi, E., et al.: Data mining techniques and preference learning in recommender systems. Comput. Inf. Sci. 6(4), 88 (2013) 8. Moreno, M.N., Segrera, S., López, V.F., et al.: Web mining based framework for solving usual problems in recommender systems. A case study for movies’ recommendation. Neurocomputing 176, 72–80 (2015) 9. Breese, J.S., Heckerman, D., Kadie, C.: Empirical analysis of predictive algorithms for collaborative filtering. In: Fourteenth Conference on Uncertainty in Artificial Intelligence, pp. 43–52 (1998) 10. Yang, R.L.: Convergence of the simulated annealing algorithm for continuous global optimization. J. Optim. Theory Appl. 104(3), 691–716 (2000)

Healthcare-Related Data Integration Framework and Knowledge Reasoning Process Hong Qing Yu1 ✉ , Xia Zhao2, Zhikun Deng1, and Feng Dong1 (

)

1

2

Department of Computer Science, University of Bedfordshire, Luton, UK [email protected] School of Computing and Digital Technology, Birmingham City University, Birmingham, UK

Abstract. In this paper, we illustrate sensor data based healthcare information integration framework with semantic knowledge reasoning power. Nowadays, more and more people start to use mobile applications that can collects data from a variety of health and wellbeing sensors and presents significant correlations across sensors systems. However, it is difficult to correlate and integrate data from these varieties provided users with overall wellbeing picture and hidden insights about systematic health trends. The paper presents a data semantic integration solution using semantic web technologies. The process includes knowledge lifting and reasoning process that could feedback many hidden health factors and personal lifestyle analysis using semantic rule language (SWRL). Keywords: Healthcare · Knowledge discovery · Semantic web · Data mining

1

Introduction

There has been indeed growing interest in the ‘initiative of self-monitoring’, evidenced by the sharp market expansion in life-logging devices and apps. These sensors are capable of constantly monitoring personal health behaviours and activities (e.g. walking, calories, heart rate, and diet), leading to unprecedented opportunities in self-care. Corre‐ spondingly, significant research effort has started to harvest and integrate the sensors for long-term health data collections – examples include MyHealthAvatar [1] and MyLifeHub [2]. Such a long-term data collection is extremely valuable to individualised disease prediction and prevention, and to promoting healthy lifestyles. Although many data can be collected from different devices or applications, it is currently still quite difficult for people to discover correlations about themselves. This paper illustrates an innovative framework that manages and integrates multiple health-related data resources from wearable sensors, mobile and web applications for discovering people’s health knowledge, which will exert influence on the future direc‐ tion of people’s self-care empowerment, disease prevention and importantly promote better lifestyles. With fast development of health-oriented smart sensor devices, mobile applications and social network technologies, various different kinds of information related to human health can be more efficiently collected. Two important challenges the paper would like to address are: © Springer International Publishing AG 2017 L. Uden et al. (Eds.): KMO 2017, CCIS 731, pp. 386–396, 2017. DOI: 10.1007/978-3-319-62698-7_32

Healthcare-Related Data Integration Framework

387

1. Efficiently storing, querying and integrating the big dataset that collected from heterogeneous data resources. 2. Providing scale and easily understandable knowledge discovery mechanism to observe significant factors from a big and day-to-day data repository.

2

Research Background and Related Work

2.1 MyHealthAvatar Project MyHealthAvatar framework [1] is a proof of concept EU funded 3 million euros project for providing a digital representation of patient health status. It is designed as a lifetime companion for individual citizens that facilitates the collection of, and access to, longterm health-status information that includes citizens’ social and sensor data together with major data resources from traditional healthcare organisations. Currently, the inte‐ grated are stored in a cloud and NoSQL based data repository with Web APIs to query and access. This paper will focus on the semantic reasoning framework that will extract health event knowledge from the IoT raw data into semantic representation and to be queried to perform further semantic based reasoning tasks. 2.2 Related Work Traditionally, the data integration task is mainly identified as a data warehousing process problem [3]. There are two typical approaches of wrappers and mediators. The goal of a wrapper is to access a source, extract the relevant data, and present such data in a specified format [4]. The role of a mediator is to collect, clean, and combine data produced by different wrappers (or mediators), so as to meet a specific information need of the integration system [5]. However, these techniques are struggling to deal with integration requirements on flexible data structures and are less feasible for datasets that are frequently updated, which is mostly the case since Web Services/Web APIs are mostly applied as data providers nowadays. The paper [6] presents the Mobile Health Mashups system, a mobile service that collects data from a variety of health and well-being sensors and presents significant correlations across sensors in a mobile widget as well as on a mobile web application. The work focuses on analyzing and detailing the technical solution, such as: integration of sensors, how to create correlations between two data sets, and the presentation of the statistical data as feeds and graphs. The work applies an iterative design process that involved a two-month field trial, the outcome of this trial, and implications for design of mobile data mashup systems. However, the system only mashes up two data resources – from Fitbit and Withings. The mashup simply federates the two data sets together without any filtering and semantic mapping. In recent years, the work in data integration research concerns the semantic integra‐ tion problem. For example, in the healthcare research domain, [7] proposes to build an interoperable regional healthcare system among several levels of medical treatment organizations. The approaches include the ontology based approach is introduced as the methodology and technological solution for information integration, the integration

388

H.Q. Yu et al.

framework of data sharing among different organisations are proposed and the virtual database to realize data integration of hospital information systems is established. This research work supports the interoperable regional healthcare system with functions of modularization and expansibility, and the stability of the system is enhanced by the hierarchy structure. However, the complexity and the size of the data have not been significant reduced in the semantic layer to utilize further knowledge discovery and reasoning because of the graphical data repository features. It suggests that the semantic data integration should have pre-processing steps and only significant data should be lifted into the RDF repositories instead transforming all the original data into semantic layer.

3

Healthcare-Related Data Integration from Multiple Data Resources in Private Cloud Environment

Collecting and meaningfully integrating heterogeneous data resources is a longstanding problem in data management and engineering research areas. In our research, we collect desired data from multiple data resources including mobile applications (Moves), wear‐ able sensors or digital measuring devices (Fitbit [8, 9] and Withings1) and MHA plat‐ form. Each different data resource provides different and useful data information, as Table 1 shows. The data collection process applies Web API technologies following the OAuth security protocol2. Whenever the user logs in to the MHA platform, the data from other devices can be synchronized into the system. Table 1. JSON semantic mapping 1 JSON structure syntax Mapped property defined in the ontology Ranking value mha: rank Duration, mha: time Destination mha: located in Distance mha: distance Step mha: step Activity_group mha: hasEvent

Range value (0,1] Seconds Annotation text or unknown Metres Count Activity type

For data integration, the column-based NoSQL database (Cassandra) [10] has been developed to mash up the heterogeneous data as whole. The most important advanced features supported by NoSQL database are: • Nature of the data to be stored: To store and access the lifelong information of each citizen in MyHealthAvatar, there will be a large volume of data in unstructured, semistructured and structured formats. For example, data collected from wearable devices and social networks may come with different formats (e.g. XML and JSON) and 1 2

http://www.withings.com/. http://oauth.net/2/.

Healthcare-Related Data Integration Framework

389

involves multiple data types (e.g. objects, lists and customized data types), which are difficult to incorporate into the rows/columns table style in SQL. Furthermore, data from different resources with unknown structure are likely to be added in the future. It is not feasible to define the data type and structured beforehand as required by SQL databases. • Data analytics: It is important to perform data analytics and possibly near real-time analytics in MyHealthAvatar in order to support risk detection. • System scalability and performance: Although the MyHealthAvatar is a proof-ofconcept project, we foresee the future of such systems will manage large amounts of user data with high demands on updating and reading. It is vital the system can scale and adapt to distributed computation paradigms (e.g. cloud computing) when the volume of the data grows. • Agility of the application development: MyHealthAvatar aims to collect and access lifelong health data for citizens. The uncertainty of the data types from different source centres requires that the development is done in iterations and is flexible and dynamic when new data resources are added. Figure 1 shows the detail designed NoSQL database structure that bases on the column and key query data storing mechanism.

Fig. 1. NoSQL data model design examples

Each column family stores a group of rows that contains a set of individual columns in a specific data structuring requirement. For example, one row in the ‘activities column’ groups all the data columns that store activity type, step counting and duration data elements. The other row in the family can store the places information that the user has travelled to or plans to visit. The ‘profile’ column family completely focuses on managing basic user profile information such as name and contact. The main structure of the Cassandra deployment on the cloud is depicted in Fig. 2.

390

H.Q. Yu et al.

Fig. 2. Cassandra cluster cross clouds implementation

As seen in Fig. 2, the Cassandra nodes are distributed in both public and private clouds, while within each respective cloud environment itself, nodes are deliberately distributed on different server racks. The configuration allows three replications of every piece of information coming into the system; in Fig. 2, R1, R2, R3 are on separate clouds (physical environment, network connection, etc.) and separate server racks (separate physical server and separate hard drive). This deployment copes with a potential disaster situation: if power is cut in one city, MyHealthAvatar’s data repository should still work as we have distributed the cluster in different geographical locations. In case of hardware failure of a server rack within one data centre, it should not affect the operation of the data repository since we have replications on different server racks.

4

Semantic Based Knowledge Representation

We have illustrated the semantic ontology in paper [11] on 2015 KMO conference. Based on initial evaluation, we modified the ontology into new version as description below: TMO terminologies [12] with some existing semantic concepts from well-known domain ontologies and our defined personal activity and treatment terms. • Event is defined as the same as TMO.Processual_entity is a super concept to clas‐ sify an interesting event that is related to the health of an individual user. The event is the super class of (discover a) Symptom, (taking a) Treatment, (diagnosed a) TMO.Disease_progression and (having a) Activity. Each event associates with a particular time point on the user’s time.TemporalEntity. In addition, Event is the

Healthcare-Related Data Integration Framework

•

•

•

•

•

•

• •

3 4

391

central point of the whole ontology, which can be detected from the data mining layer. Person is the concept to describe a MHA user using the FOAF ontology. The FOAF ontology includes all possible aspects of a general profile of a person, such as health history, gender and height. In our proposed framework, the semantic layer nominal‐ izes the user’s name and address information that is stored in the lower-level NoSQL database with more secured data management infrastructure; we do not discuss the security topic in this paper. (significant) Activity is a subclass of the Event concept to identify the activity that is more significant to the user, rather than including all daily activities. In general, all the significant events should be related to understand the user’s health situation or lifestyle. The activity type can be grouped by exercise type such as ‘Running’, ‘Driving’ or ‘Shopping’, but also can be categorized by the places and social activity type. Each significant activity should also record the duration, place, and possibly with distance, calory consumption and steps. Symptom imported from TMO is a subclass of the Event concept to present unusual health-related conditions that are detected and concluded from the user’s data. The same as all other events, the symptoms have a time stamp and place. Currently, the subclasses of Symptom include low/high blood pressure, unusual heart rate, poor sleeping and significant weight/fat changes. Other non-sensor symptoms can also be added but have to rely on the user’s manually inputs. Annotation defined in the AO (Annotation Ontology) [13] is used as a semantic vocabulary link to an event. The annotations should use controlled vocabularies or semantic identifiers to define the meaning. The annotation can be automatically added through a linked data annotation engine, e.g. DBpedia Spotlight3 or can be added by users via annotation tools from the MHA platform. Treatment is a subclass of the Event concept for recording the treatments that have been taken by the user from a medical health organisations or by the user themself. Treatment refers to any medical health-related actions with respect to the user such as taking drugs, operations, and physical and mental therapies. In addition, the treat‐ ment requires identifying the exact time point on the user’s timeline. Disease_progression reused from TMO is a subclass of the Event concept and presents the medical situations that were diagnosed in the past according the user’s timeline or are a potential risk for the user. The Health Condition, Treatment and Symptom concepts structure a triangle relation that could be very valuable knowledge for an individual user or a group of users. Risk defined by ICO is used as a concept to evaluate the possibility or progress levels to a particular health condition. Lifestyle is imported using the Intention concept in the MWLA ontology4 that defines 25 lifestyle instances. Since MWLA is still a live EC project CARRE [14], the numbers of the lifestyle definitions can be enriched in the future.

https://github.com/dbpedia-spotlight/dbpedia-spotlight/wiki. https://bioportal.bioontology.org/ontologies/MWLA.

392

H.Q. Yu et al.

• Knowledge is an association term to the Event class that means that each item of knowledge is related to an instance of a discovered event such as an activity, a symptom, a treatment or a disease progress.

5

Semantic Knowledge Lifting and Reasoning

5.1 Semantic Lifting The semantic lifting process is a generalization of RDF triples based on the proposed ontology, which includes two steps of semantic mapping: domain mapping and property mapping with range assignment. Step 1: Domain mapping In the first step the domain matching algorithm is applied to identify the domain element; this is the simplest algorithm in these three steps. According to our JSON structure composing the summary data analysis, only activity type or symptom name elements are suitable candidates for the domain that can be lifted as subject elements of the instance RDF triples. If the element is under the activities JSON structure, then a URI will be generated and specified as an Activity class defined in the OWL ontology. For example, if the activity is classified as running, then the URI will be http://myhealth‐ avatar.org/activity/running and the definition triple will be

A similar process is generated for the symptom event. For example, the high blood pressure symptom event for a particular user can be defined as

Healthcare-Related Data Integration Framework

393

Step 2: Property and range mapping According to the JSON input structure, the property and range mapping are performed together based on the pre-defined mappings in Tables 1 and 2. Table 2. JSON semantic mapping 2 JSON structure syntax Mapped property defined in the ontology Value mha: hasValue Time mha: time Location mha: located in

Range value Text value with unit Time spot/date information Annotation text or unknown

5.2 Semantic Reasoning Examples Example 1: Travel/long commute lifestyle for the past month. Definition: Long transport (more than 2 h) activity events have been lifted in to the RDF at least four times in the last month. The reasoning process will be: (1) Construct (SPARQL query) the last month activity event RDF memory-model based on the ontology retrieved from the triple storage. (2) Specify the SWRL rule or SPARQL query for reasoning. In this example, Code 1 is used as the SPARQL query.

394

H.Q. Yu et al.

PREFIX mha: SELECT (COUNT(?numberOfLongTravelling) AS ?howMany ?p) WHERE { ?p mha:hasEvent ?e . ?e rdf:type mha:transport . ?e mha:time ?d . FILTER (?d >= 7200) } HAVING ( ?howMany > 4 ) CODE 1 Code 1 specifies a query that will return transport times (COUNT ?howMany) a month if the transport activity events have been detected more than 4 times (use ASK query to filter ?howMany > 4) in a targeted month. (3) If the query returns a value, then it means the user satisfies the defined reasoning query. Then we can construct the semantic links between the person to the Travel term defined in the MWLA. The steps of (2) and (3) can be integrated by combining Construct update query to the model as one step (Code 2).

PREFIX mha: mwla: < http://purl.bioontology.org/ontology/MWLA> CONSTRUCT (?p mha:lifestyle mwla:Travel) SELECT (COUNT(?numberOflongTravelling) ?howMany ?p) WHERE { ?p mha:hasEvent ?e . ?e rdf:type mha:transport . ?e mha:time ?d . FILTER (?d >= 7200) } HAVING ( ?howMany > 4 ) CODE 2

AS

The similar processes can be applied to detect other lifestyle, such as lack of activity or lack of exercise and so on. Example 2: SWRL-based health-condition risk alarm reasoning. Definition: If a person lacks activity and has age > 60, then there is a risk of high blood pressure. The rule can be defined as Code 3 in the Jena rule engine.

[rule: (?p mha:lifestyle mwla:noActivity), (?p foaf:age ?i), greaterThan(?i, 60) -> (?p tom:has_risk ?x), (?x tom:is_about ?d), (?d rdf:type tom:High_Blood_Pressure)] CODE 3

Healthcare-Related Data Integration Framework

395

Reasoning rules should be defined by healthcare professionals, for example, the blood pressures have systolic blood pressure that measures how hard the heart’s left ventricle contracts to circulate blood through the body. Diastolic blood pressure meas‐ ures the pressure in the blood vessels when the heart’s chambers are relaxed and filling with blood. UK National Health Service (NHS) guidelines5 indicates that normal adult blood pressure should be between 90/60 and 140/90, where the top (first) number is the systolic pressure and the diastolic is the bottom (second) number. In addition, readings higher than 140/90 can be defined as high blood pressure and lower than 90/60 as low blood pressure. The other mining methods are defined in our reasoning rules based on similar UK NHS guidelines. Heart rate range should generally be in [60, 100], otherwise it is too slow or too fast. The sleep hours should generally be between six and nine hours

6

Conclusions and Future Work

This paper presents a NoSQL and Semantic Web based framework to address health related data integration from heterogeneous data resources and semantic knowledge lifting and reasoning. We firstly integrated the data from four different kinds of datasets that contain plenty valuable personalised healthcare information such as steps, sleeps, heat bits, weights and so on into a private cloud NoSQL repository. Secondly, the data are lifted into semantic layer by applying the semantic mapping rules dynamically as backend services. As results, our system can provide the efficient knowledge discovery process by using semantic rule language that is express enough for describing desired query questions. Our future work will focus on collaborating with healthcare professionals to provide more semantic reasoning rules and knowledge discovery requirements that could help on doctor diagnose or tracking treatment process. In addition, we would like to build more mobile applications on top of our framework to gain more users to do more accurate evaluations.

References 1. Kondylakis, H., Spanakis, M., Sfakianakis, S., Sakkalis, V., Tsiknakis, M., Marias, K., Xia, Z., Yu, H.Q., Dong, F.: Digital patient: personalized and translational data management through the MyHealthAvatar EU project. In: International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), Milan, Italy (2015) 2. MyLifeHub. http://gow.epsrc.ac.uk/NGBOViewGrant.aspx?GrantRef=EP/L023830/1 3. Calvanese, D., De Giacomo, G., Lenzerini, M., Nardi, D., Rosati, R.: Data integration in data warehousing. Int. J. Coop. Inf. Syst. 10(3), 236–271 (2001) 4. Wiederhold, G.: Mediators in the architecture of future information systems. IEEE Comput. 25(3), 38–49 (1992) 5. Ullman, J.D.: Information integration using logical views. In: Afrati, F., Kolaitis, P. (eds.) ICDT 1997. LNCS, vol. 1186, pp. 19–40. Springer, Heidelberg (1997). doi: 10.1007/3-540-62222-5_34 5

http://www.nhs.uk/NHSEngland/thenhs/about/Pages/overview.aspx.

396

H.Q. Yu et al.

6. Tollmar, K., Bentley, F., Viedma, C.: Mobile health mashups: making sense of multiple streams of wellbeing and contextual data for presentation on a mobile device. In: 2012 6th International Conference on Pervasive Computing Technologies for Healthcare (PervasiveHealth), San Diego, CA, pp. 65–72 (2012) 7. Yang, H., Li, W.: An ontology-based approach for data integration in regionally interoperable healthcare systems. In: 11th International Conference on Informatics and Semiotics in Organisations (ICISO 2009), 11–12 Apr 2009, Beijing, China, pp. 93–96 (2009) 8. Fitbit Ltd, Fitbit Healthy Futures Report (2013). http://www.trajectorypartnership.com/wpcontent/uploads/2014/02/Fitbit-Healthy-Futures-Report-September-2013.pdf 9. Mackinlay, M.Z.: Phases of accuracy diagnosis: (in) visibility of system status in the Fitbit. Intersect Stanf. J. Sci. Technol. Soc. 6(2), 1–9 (2013). Date accessed 27 Apr 2016 10. DATASTAX Corporation, Introduction to Apache Cassandra – White Paper, July 2013. http:// www.datastax.com/wp-content/uploads/2012/08/WP-IntrotoCassandra.pdf 11. Yu, H.Q., Zhao, X., Deng, Z., Dong, F.: Ontology driven personal health knowledge discovery. In: Uden, L., Heričko, M., Ting, I.-H. (eds.) KMO 2015. LNBIP, vol. 224, pp. 649– 663. Springer, Cham (2015). doi:10.1007/978-3-319-21009-4_48 12. Luciano, J.S., Andersson, B., Batchelor, C., et al.: The translational medicine ontology and knowledge base: driving personalized medicine by bridging the gap between bench and bedside. J. Biomed. Semant. 2(Suppl 2), S1 (2011). doi:10.1186/2041-1480-2-S2-S1 13. Ciccarese, P., Ocana, M., Castro, L.J.G., Das, S., Clark, T.: An open annotation ontology for science on web 3.0. J. Biomed. Semant. 2(Suppl 2), 4 (2011). http://doi.org/ 10.1186/2041-1480-2-S2-S4 14. https://www.carre-project.eu/

Data Mining and Intelligent Science

Algorithms for Attribute Selection and Knowledge Discovery Jorge Enrique Rodríguez R.(&), Víctor Hugo Medina García(&), and Lina María Medina Estrada(&) District University “Francisco José de Caldas”, Bogotá, Colombia [email protected], [email protected], [email protected]

Abstract. The features relevant selection is a task performed prior to the data mining and can be seen as one of the most important problems to solve in the data preprocessing stage an in the machine learning. With the feature selection is mainly intended to improve predictive or descriptive performance of models and implement faster and less expensive algorithms. In this paper an analysis about feature selection methods is made emphasizing on decision trees, entropy measure for ranking features, and estimation of distribution algorithms. Finally, we show the result analysis of execute the three algorithms. Keywords: Features selection
Complexity
Decision trees
Entropy Estimation of distribution algorithms
Machine learning
Data mining




1 Introduction Usually, database contains millions of tuples and thousands of attributes, presenting dependencies among attributes [1]. Database may contain irrelevant features to the data mining or feature redundancy. For example, if it has to classify customers into popular-music-CD buyers and non-buyers, characteristics such as the telephone number and the number of children they have may be irrelevant, while age and academic level may be relevant characteristics to classify customers. The essential purpose of data preprocessing is to manipulate and transform each dataset, making the information contained within them more accessible and coherent [2, 3]. Feature relevance selection is applied to reduce the number of features in applications where data has hundreds or thousands of features. Feature selection algorithms should be distinguished from feature extraction. Feature extraction creates new features from functions of the original features, whereas feature selection returns a subset of the features. Feature selection algorithms are often used in domains where there are many features and comparatively few samples; this algorithms can be used to identify and remove unneeded, irrelevant and redundant attributes from data that do not contribute to the accuracy of a predictive model or may in fact decrease the accuracy of the model.

© Springer International Publishing AG 2017 L. Uden et al. (Eds.): KMO 2017, CCIS 731, pp. 399–409, 2017. DOI: 10.1007/978-3-319-62698-7_33

400

J.E. Rodríguez R. et al.

The existing selection algorithms focus, mainly, on finding relevant features [4]. This is a process, in which the most relevant characteristics are selected, improving the knowledge discovery database. This review paper is structured as follows: feature selection theory, feature relevance selection algorithms, analysis results, and conclusions.

2 Feature Selection Feature selection is the most important process within data preprocessing stage, since it is the stage in which the most significant attributes are left within the dataset. In some cases, the selection is applied through trial and error until a more-efficient model or pattern has been gotten, but this is not the most accurate, since time and complexity increase, regarding the number of variables in the dataset. With the selection it is sought to leave only the attributes with which it can be reached make a prediction or description as accurate as possible. Feature selection, applied as a data preprocessing stage to data mining, proves to be valuable in that it eliminates the irrelevant features that make algorithms ineffective. Sometimes the percentage of instances correctly classified is higher if a previous feature selection is applied, since the data to be mined will be noise-free [5]. Feature selection task is divided into four stages [6]: the first one determines the possible set of attributes to perform the representation of the problem, then the subset of attributes generated in step one is evaluated. Subsequently, it is examined whether the selected subset satisfies the search criteria. The final stage verifies the quality of the subset of attributes that was determined. These processes can be classified differently depending on the stage in which we focus, in order to make this distinction in three categories [7]: Filters, Wrappers [8, 9] and hybrids [10]. In Filters methods the selection procedure is performed independently of the evaluation function. In these can be distinguished four different evaluation measures: distance, information, dependence and consistency. Respective examples of each one of these measures can be found in: [11–13]. Wrappers methods combine search in the attribute space with the machine learning evaluating the set of attributes and choosing the most appropriate. The disadvantage they present is that they are more expensive than the Filters [7] although they tend to get better results. Hybrid models present the advantages that Filters and Wrappers models provide. Since the feature selection is applicable to dissimilar real situations, it is difficult to reach a consensus as to which is the best possible choice; this makes possible multiple algorithms of this type [14].

3 Feature Selection Algorithms Feature selection algorithms are intended to find subsets of feature relevant that reduce the loss of information and reduce noise, eliminating feature irrelevant and selecting the relevant ones. The feature relevant provide higher quality information for further

Algorithms for Attribute Selection and Knowledge Discovery

401

analysis of data mining. This algorithms used may differ in the way of working, but all of them are focused on perfecting the minable view, which is the general objective of data preprocessing. It is therefore important to bear in mind that not only the fact of feature selection will allow to carry out the data mining in a suitable form, but that many factors, to be taken into account, for example, which data must be analyzed and, consequently, the domain in which it is being worked, as well as the validity of the data that they have. 3.1

Decision Trees

A decision trees are a representation in which each set of possible conclusions is implicitly established by a list of known class samples [15]. This are a form of representing knowledge easily and understandably. It is stated that these are more understandable than other algorithms, such as support vector machines or neural networks artificial; since they are more intelligible to represent knowledge in a symbolic way. Decision tree is a tree structure that classifies an instance of entrance in one of its possible classes; they are used to extract knowledge in large volumes of data, generating rules, which are used to support decisions. Decision tree has a simple form that efficiently classifies new data [16]. Figure 1 illustrates a tree [17].

Fig. 1. Decision tree

These trees are considered as an important tool for data mining; compared to other algorithms, decision trees are faster and more accurate [18]. Learning in a decision trees is a method to approximate an objective function of discrete values, in which the learning function is represented by a tree. It can also be represented as a set IF – THEN rules to improve the readability by the human being. These learning methods are the most popular inductive inference algorithms and they have thriving application in various machine learning tasks [19–21]. There are three fundamental steps to feature selection with decision trees: first of all generate the tree; then, from this generate the rules and later select the attributes by

402

J.E. Rodríguez R. et al.

observing the ones most used in the rules. The construction is largely related to designing it efficiently but short, that is, to generate the smallest possible tree. The theory of information provides a mathematical model (Eq. 1) to measure the total disorder in a database; however this does not guarantee that a smaller tree is built, that is, with low height and branch. disAver ¼

  Xb nb Xc nbc nbc  log  c 1 nt 1 nb nb

ð1Þ

Where: nb is the number of instances of attribute b nt is the total number of instances nbc is the number of instances of attribute b belonging to class c According to the above, the disorder of all attributes is calculated to select the root of the tree, which would be the most relevant. The calculation of the disorder is repeated until a tree with the most relevant attributes is built. Some advantages of decision trees are: • • • • • •

Simple to understand and to interpret. Trees can be visualized. Requires little data preparation. The complexity is logarithmic in the number of data points used to train the tree. Able to handle both numerical and categorical data. Able to handle multi-output problems. Possible to validate a model using statistical tests. That makes it possible to account for the reliability of the model. • Performs well even if its assumptions are somewhat violated by the true model from which the data were generated.

3.2

Entropy Measure for Ranking Features

Ranking methods may filter features leading to reduced dimensionality of the feature space. This is especially effective for classification methods that do not have any inherent feature selections build in, such as the nearest neighbor methods or some neural networks. Ranking of features determines the importance of any individual feature, neglecting their possible interactions. Ranking methods are based on statistics, information theory, or on some functions of classifiers outputs. This is a simple algorithm for feature selection based on the measure of entropy applied to problems considered within unsupervised machine learning. However, for a high number of attributes, its complexity increases significantly. The basic idea is that all instances are given as a vector of attribute values without a class acting as output from each instance. All the process is based on eliminating n attributes, without losing the basic characteristics of the dataset. This algorithm is based on the measure of similarity S that is inversely proportional to the distance D between two instances of n dimensions [22–24].

Algorithms for Attribute Selection and Knowledge Discovery

403

The distance D is small with instances close to 0 and large with instances close to 1. When the features are numerical, the measure of similarity S of two instances can be as shown in Eq. 2. Si j ¼ e D i j

ð2Þ

Where Dij is the distance between the instances Xij and Yij, and a a parameter expressed in mathematic terms (Eq. 3). a¼

lnð0:5Þ D

ð3Þ

D is the average distance between the samples in the dataset. In practice it is close to 0.5. The Euclidean distance are calculated as follows (Eq. 4). sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi Xn  Xik  Xjk 2 Di j ¼ k¼1 max  min k k

ð4Þ

Where n is number of attributes, maxk and mink are the maximum and minimum value used for the normalization of k-attributes. When the attributes are categorical, the hamming distance is used (Eq. 5). sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi   Xn Xik ¼ Xjk  Si j ¼ k¼ n

ð5Þ

  Where Xik ¼ Xjk  is 1, otherwise it is 0. In order to deal with numerical and categorical data, the numeric attributes must be discretized in order to transform them into categorical ones. From information theory, it is known that entropy is a global measure, and that it is small for ordered data configurations and high for disordered configurations. The algorithm in question compares the entropy for a set of data before and after deleting attributes. For a data set of N instances, the measure of the entropy is (Eq. 6). E¼

XN1 XN i¼1

j¼i þ 1



      Si j log Si j þ 1  Si j log 1  Si j

ð6Þ

This algorithm gradually removes the least significant attributes, maintaining the original order of the data; the process of removing attributes can end in any iteration of the algorithm. Another good alternative for entropy estimation are the techniques which bypass the density estimation, estimating entropy directly from the data. These are usually based on the distances between the data samples. Distances are not more complex in high dimensional feature space. Then, the complexity of these approaches does not depend on the number of dimensions, but on the number of samples.

404

3.3

J.E. Rodríguez R. et al.

Estimation of Distribution Algorithms (EDAs)

In EDAs the population of individuals is generated without using neither crossover nor mutation operators. Instead, the new individuals are sampled starting from a probability distribution estimated from the database containing only selected individuals from the previous generation. At the same time, while in other heuristics from evolutionary computation the interrelations between the different variables representing the individuals are kept in mind implicitly, in EDAs the interrelations are expressed explicitly through the joint probability distribution associated with the individuals selected at each iteration. In fact, the task of estimating the joint probability distribution associated with the database of the selected individuals from the previous generation constitutes the hardest work to perform. In particular, the latter requires the adaptation of methods to learn models from data that have been developed by researchers in the domain of probabilistic graphical models. EDA algorithm is a stochastic search technique based on population, which uses a probability distribution model to explore candidates (instances) in a search space. This distribution is estimated iteratively with the candidates until a standstill criterion is reached. EDAs, replace the mechanisms of variation (mix and mutation) traditionally used by evolutionary algorithm for the generation of individuals obtained by simulating a probability distribution [25, 26]. EDAs have been recognized as a strong algorithm to optimize. They have shown a better performance in comparison with evolutionary algorithm, in problems where these have not presented satisfactory results. This is mainly due to the explicit nature of the relations or dependencies between the most important variables associated with some particular problems that are estimated through probability distributions [27, 28]. Table 1 shows the EDA algorithm; in the first place an initial population of individuals is generated. These individuals are evaluated according to an objective or aptitude function. Table 1. EDA Algorithm pseudo-code [29]

Algorithms for Attribute Selection and Knowledge Discovery

405

This evaluates how appropriate each individual is as a solution to the problem. Based on this evaluation, a subset of the best individuals is selected. Thus, from this subset it is learnt a probability distribution to be used to sample another population [27]. The biggest problem with EDAs is how to estimate the probability distribution. Obviously the computation of all the parameters necessary to specify the distribution of joint probability is not practical. This problem results in the approximation of the joint probability distribution by means of different factorizations with a certain degree of complexity. EDAs can be distinguished and classified by the type of variables they deal with; categorical or numerical. Although the real objective is to classify them according to the type of link Used for its distribution. Depending on the type of connection used in their model distribution, EDAs can be classified as univariate, bivariate and multivariate. EDA [30, 31] approximately optimizes a cost function by building a probabilistic model of a pool of promising sub-optimal solutions over a given search space. For very-high dimensional search spaces, storing and updating a large population of candidates may imply a computational burden in both time and memory. The compact approach circumvents storage limitations by incrementally updating the probability model using just two candidates at any step of the algorithm, instead of the entire population. This feature makes the compact EDA framework practical for large-scale optimization, a soon-to-be commonplace setting in scientific domains such as bioinformatics, particle physics, chemical crystallography, or social network analysis, to name a few.

4 Results Finally we show the results of applying the three algorithms in dataset different.

4.1

Decision Trees

Figure 2 shows the behavior of three dataset according to the percentage of selection and percentage of selected attributes (for the test the UDClear version 1.0 software was used). You can see that they have totally different behaviors and the reasons are: Soybean dataset: shows behavior in which a low percentage of feature is selected, the highest percentage of selection and the maximum of selected attributes does not exceed 70%. The dataset does not cover the total of the attributes, that is, so dataset has 35 attributes and 683 instances, therefore, the 683 instances are not sufficient to cover all possible combinations that may occur. This problem has been called the curse of dimensionality. Chess dataset: the behavior of these data is totally opposite to the Soybean dataset, since it covers the largest amount of attribute space. Means that most data is relevant and only begins to be deleted with a high features selection percentage. This dataset

406

J.E. Rodríguez R. et al.

Fig. 2. Feature selection tests

theoretically provides all the necessary data to create strong and useful models that can extract patterns and relevant knowledge. Census dataset: dataset shows a balanced behavior due to the fact that they are data extracted from real databases and with a high number of instances. This would be the set of data chosen to perform selection and subsequently data mining, since in practice, forming a data set that covers the entire space of dimensions is quite high complexity, when generating a model from them, taking into account the increasing volume of databases.

4.2

Entropy Measure for Ranking Features

For test this algorithm we used a dataset of four features (X1, X2, X3, and X4) with one thousand instances. Features contain categorical data; therefore we are used the hamming distance to compute the similarity (Eq. 5) between the instances and then the entropy is computed (Eq. 6). The result generates that the feature X3 is the least relevant; since in computing the difference between the total entropy and the entropy without the feature three, it is the one closest to 0. Therefore X3 must be removed from the dataset.

4.3

Estimation Distribution Algorithms (EDAs)

We used one dataset applied to solve the classical OneMax problem with 100 variables, using a population size of 30, 1000 maximum number of evaluations, and 30 candidates per iteration. For the execution of the EDAs algorithm, used Orange Suit with the widget Goldenberry. The result show in the Table 2.

Algorithms for Attribute Selection and Knowledge Discovery

407

Table 2. EDA applied result Best: {1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1} Cost: 89 Evals: 464 Argmin: 22 Argmax: 459 Min val: 44 Max val: 89 Mean: 68.4590517241 Stdev: 10.5445710397

5 Conclusions The features selection is a fundamental task in data preprocessing. Which directly affects data mining process, in terms of the effectiveness of its Algorithms and complexity is concerned; because of this, the researchers in this field of knowledge seek to create new algorithms to reduce noise in data in order to make it the main purpose of data mining, which is the support for decision-making. Decision trees are used as an algorithm to feature selection are a good option. However, it must be taken into account that: the data set must be categorical, only applies for predictive problems, which limits the field of applications, and if the dataset is incomplete, the selection is not considered as good. Feature selection based entropy measure for ranking features is applicable only to descriptive tasks, limiting the field of application just like decision trees. Its main weakness lies in the high complexity since it makes a comparison combining all the instances. As for the EDAs, these enjoy a good reputation within the feature selection algorithms; however, it has some weaknesses, such as: redundancy in generating the dependency trees, in free code applications has only been done with languages interpreters, and there is limited evidence of having used multivariate data.

References 1. Larose, D.: Data Mining: Methods and Models, pp. 1–3. Wiley-Interscience, USA (2006) 2. Pyle, D.: Data Preparation for Data Mining, pp. 15–19. Morgan Kaufmann Publisher, USA (1999) 3. Bradley, P., Mangasarian, O.: Feature selection via concave minimization and support vector machine. J. Mach. Learn. ICML, 82–90 (1998). USA 4. Lei, Y., Huan, L.: Efficient feature selection via analysis of relevance and redundancy. J. Mach. Learn. Res. 5, 1205–1224 (2004). USA 5. Guyon, I., Elissee, A.: An introduction to variable and feature selection. J. Mach. Learn. Res. 3, 1157–1182 (2003). USA

408

J.E. Rodríguez R. et al.

6. Dash, M., Liu, H.: Feature selection for classification. J. Intell. Data Anal. 1(3), 131–156 (1996). USA 7. Liu, H., Lei, Y.: Toward integrating feature selection algorithms for classification and clustering. IEEE Trans. Knowl. Data Eng. 17(4), 491–502 (2005). USA 8. Kohavi, R., John, G.: Wrappers for feature subset selection. Artif. Intell. 97(12), 273–324 (1997). USA 9. Jennifer, G.: Feature selection for unsupervised learning. J. Mach. Learn. Res. 5, 845–889 (2004). USA 10. Das, S.: Filters, wrappers and a boosting-based hybrid for feature selection. In: Proceedings of the 18th International Conference on Machine Learning, USA, pp. 74–81 (2001) 11. Cardie, (2001): Using decision trees to improve case-based learning. In: Utgo, P. (ed.) Proceedings of the 10th International Conference on Machine Learning, USA, pp. 25–32 (1993) 12. Mucciardi, A., Gose, E.: A comparison of seven techniques for choosing subsets of pattern recognition. IEEE Trans. Comput. 20, 1023–1031 (1971). USA 13. Ruiz, R., Riquelme, J., Aguilar-Ruiz, J.: Projection-based measure for efficient feature selection. J. Intell. Fuzzy Syst. 12, 175–183 (2003). USA 14. Pérez, I., Sánchez, R.: Adaptación del método de reducción no lineal LLE para la selección de atributos en WEKA. In: III Conferencia Internacional en Ciencias Computacionales e Informáticas, Cuba, pp. 1–7 (2016) 15. Winston, P.: Inteligencia Artificial, pp. 455–460. Addison Wesley, USA (1994) 16. Chourasia, S.: Survey paper on improved methods of ID3 decision tree classification. Int. J. Sci. Res. Pub. 3, 1–4 (2013). USA 17. Rodríguez, J.: Fundamentos de minería de datos. Fondo de publicaciones de la Universidad Distrital Francisco José de Caldas, Colombia, pp. 63–64 (2010) 18. Changala, R., Gummadi, A., Yedukondalu, G., Raju, U.N.P.G.: Classification by decision tree induction algorithm to learn decision trees from the class-labeled training tuples. Int. J. Adv. Res. Comput. Sci. Softw. Eng. 2(4), 427–434 (2012). USA 19. Michell, T.: Machine Learning, pp. 50–56. McGraw Hill, USA (1997) 20. Han, J., Kamber, M.: Data Mining: Concepts and Techniques, vol. 3, pp. 331–336. McGraw Hill, USA (2012) 21. Russell, S., Norvig, P.: Artificial Intelligence: A Modern Approach, pp. 531–540. Prentice Hall, USA (2012) 22. Kantardzic, M.: Data Mining: Concepts, Models, Methods and Algorithms, pp. 46–48. IEEE Press Wiley-Interscience, USA (2003) 23. Liu, H., Motoda, H.: Feature Selection for Knowledge Discovery and Data Mining, vol. 2, pp. 30–35. Kluwer Academic Publisher, USA (2000) 24. Liu, H., Motoda, H.: Feature Extraction, Construction and Selection. A Data Mining Perspective, pp. 20–28. Kluwer Academic Publisher, USA (2000) 25. Larrañaga, P., Lozano, J.: Estimation of Distribution Algorithms. A New Tool for Evolutionary Computation, pp. 1–2. Kluwer Academic Publishers, USA (2002) 26. Pelikan, M., Sastry, K.: Initial-population bias in the univariate estimation of distribution algorithm. In: Proceedings of the 11th Annual Conference on Genetic and Evolutionary Computation, GECCO 2009, USA, vol. 11, pp. 429–436 (2002) 27. Pérez, R., Hernández, A.: Un algoritmo de estimación de distribuciones para el problema de secuencia-miento en configuración jobshop, vol. 1, pp. 1–4. Communication Del CIMAT, Mexico (2015)

Algorithms for Attribute Selection and Knowledge Discovery

409

28. Mühlenbein, H., Paaß, G.: From recombination of genes to the estimation of distributions I. Binary parameters. In: Voigt, H.-M., Ebeling, W., Rechenberg, I., Schwefel, H.-P. (eds.) PPSN 1996. LNCS, vol. 1141, pp. 178–187. Springer, Heidelberg (1996). doi:10.1007/3540-61723-X_982 29. Rodríguez, N.: Feature relevance estimation by evolving probabilistic dependency networks and weighted kernel machine. A thesis submitted to the District University Francisco José de Caldas in fulfilment of the requirements for the degree of Master of Science in Information and Communications, Colombia, pp. 3–4 (2013) 30. Bengoetxea, E., Larrañaga, P., Bloch, I., Perchant, A.: Estimation of distribution algorithms: a new evolutionary computation approach for graph matching problems. In: Figueiredo, M., Zerubia, J., Jain, A.K. (eds.) EMMCVPR 2001. LNCS, vol. 2134, pp. 454–469. Springer, Heidelberg (2001). doi:10.1007/3-540-44745-8_30 31. Pelikan, M., Sastry, K., Cantú-Paz, E. (eds.): Scalable Optimization via Probabilistic Modeling: From Algorithms to Application. SCI, vol. 33. Springer, Heidelberg (2006). doi:10.1007/978-3-540-34954-9

P-IRON for Privacy Preservation in Data Mining G. Arumugam1 and V. Jane Varamani Sulekha2 ✉ (

1

)

Department of Computer Science, Madurai Kamaraj University, Madurai, Tamilnadu, India [email protected] 2 Fatima College, Madurai, Tamilnadu, India [email protected]

Abstract. Data mining includes extracting useful and interesting patterns from large dataset, to create and enhance decision support systems. Due to this, data mining has become an important component in various fields of day-to-day life including medicine, business, education, science and so on. Numerous data mining techniques have been developed. These techniques make the privacy preservation an important issue. When applying privacy preservation techniques, importance is given to the utility and information loss. In this paper we propose Preference Imposed Individual Ranking based microaggregation with Optimal Noise addition technique (P-IRON) for anonymizing the individual records. Through the experimental results, our proposed technique is validated to prevent the disclosure of sensitive data without degradation of data utilization. Our work highlights some discussions about future work and promising directions in the perspective of privacy preservation in data mining. Keywords: PPDM · Microaggregation · Privacy · Perturbation · Differential privacy · Individual ranking · Optimal Noise

1

Introduction

Privacy preservation is an essential need for all the data mining applications where there exists a large dataset which needs to be analyzed without the analyst or third party data miner obtaining the data directly. To eliminate this problem, researchers have developed many techniques to hide or anonymize the data before analysis. Privacy Preserving Data Mining (PPDM) made of two parts. The simplest technique is anonymization that is deidentification of the data, whereby sensitive raw data (identifiers, quasi identifiers, sensitive attributes) such as name, age, address, phone number, income, disease, SSN (social security number), SIN (social insurance number) is transformed, modified, or eliminated from the data records. In this method, non-sensitive attributes are transmitted without any change. In some cases, anonymized data can be reconstructed. Next, sensi‐ tive information mined from a database by using data mining algorithms should also be preserved because that too may compromise data privacy. Privacy Preserving Data Mining (PPDM) was first proposed by [1, 2]. To overcome the privacy issues, numerous solutions have been proposed by researchers. One could add noise to the original data so that the transformed data does not reveal the sensitive information. Some techniques are designed for vertical or horizontal partitioned data. © Springer International Publishing AG 2017 L. Uden et al. (Eds.): KMO 2017, CCIS 731, pp. 410–423, 2017. DOI: 10.1007/978-3-319-62698-7_34

P-IRON for Privacy Preservation in Data Mining

411

Other techniques include Perturbation, Blocking, Anonymization, Aggregation, Swap‐ ping, Sampling, Sanitization, Differential privacy, Condensation, Cryptography, Evolu‐ tionary algorithms like Genetic Algorithm, Artificial bee colony algorithm and Ant colony optimization algorithm. There are other techniques that emphasize on protecting the confidentiality of logic rules and patterns discovered from data. These solutions fall into two main categories. The first category is Secure Multiparty Computation (SMC) and the second one is Data Modification. SMC method provides robust level of privacy. But, these algorithms are extremely expensive in practice, and impractical for real use. The second category of the data modification approach trades privacy with improved performance in the sense that malicious data miners may infer certain properties of the original data from the masked data. This paper reviews privacy preservation techniques, the challenges in privacy preserving data mining and proposes a novel P-IRON privacy preservation. The remainder of this paper is organized as follows. Section 2 analyses Perturbation and microaggregation based PPDM methods. Section 3 introduces our proposed P-IRON. Section 4 presents experimental results and Sect. 5 describes considerations about future extensions and promising directions in the perspective of privacy preserving data mining.

2

Perturbation and Microaggregation

2.1 Perturbation Data perturbation technique is one of the most widely used techniques for privacy preserving data mining. Identifying sensitive attribute and modifying that attribute is known as perturbation technique in PPDM. It is especially suitable for applications where the data owners need to publish the sensitive data. In data perturbation, data owner randomly change the data to preserve the sensitive information before publishing the data. Perturbation techniques are often evaluated with three metrics, data utility, privacy loss and information loss. An efficient data perturbation algorithm should aim at mini‐ mizing both privacy loss and information loss. But, these three metrics are not balanced in many of the existing perturbation techniques. The loss of information typically refers to the amount of sensitive information preserved, after the perturbation. Different data mining tasks, such as classification, clustering and association mining, typically utilize different set of properties of a dataset. So, the information that is considered critical to clustering may differ from those critical to classification. We say that the exact infor‐ mation that needs to be preserved after perturbation should be “Preference based”. The following are the few techniques adopted in perturbation. Replacing the original values by a sample from the distribution where the data belongs, adding a small amount of noise to the original values, adding noise to the results of a query, sampling from a result query and swapping values between records [3].

412

G. Arumugam and V. Jane Varamani Sulekha

2.2 Representation Consider a dataset consisting of both Sensitive variables (S) as well as non-sensitive variables (N). The Sensitive and non- Sensitive variables are either numerical or cate‐ gorical and S and N together account for all numerical and categorical variables. Let g(·) and G(·) represent the probability density and cumulative density functions, respec‐ tively. While the data may consist of a set of key identification variables and they are often non-numerical in nature and/or may not permit numerical manipulation, they will not be considered further. The main aim of perturbation is to generate a set of trans‐ formed (masked) values M, so that the following requirements are satisfied. Accuracy or Data Utility: The statistical characteristics of M are the same as that of S (i.e., g(M) = g(S)), and the relationship between M and N is the same as that between S and N (i.e., g(M, N) = g(S, N)). Privacy Loss or Disclosure Risk: The confidentiality of S is maintained and the released microdata (M, N) does not increase privacy loss, (i.e., g(S | N, M) = g(S | N)). 2.3 Perturbation Techniques There are two main categories in data perturbation, one based on probability distribution and another one fixed data perturbation (Value Distortion technique). In the probability distribution, sensitive value is replaced with some distribution sample. In Fixed data perturbation noise is added to the sensitive attribute before it is released to the data miner. Fixed data perturbation methods are used for numerical, categorical data. In this paper, we mainly focus on the value distortion techniques. 2.3.1 Additive Data Perturbation (ADP) In ADP [4] approach original value is replaced with ui + v, where ui , the original data and v is a random value drawn from a certain distribution. Commonly used distributions are the uniform distribution over an interval [−α, α] and Gaussian distribution with mean μ = 0 and standard deviation α. Merits/Demerits: ADP is good if the data owner wants to make minor changes to the original data. In many cases, the original data can be accurately reconstructed from the perturbed data. 2.3.2 Condensation Based Perturbation The condensation approach [5] partitions the original data into k record groups using iterative approach. In the first step, a record is selected as the center of the group then k − 1 nearest neighbors are added to this group. In the second step, other k − 1 elements are formed as another group. The first groups of k records are removed from the original dataset D and the other records are formed into the next group G. Next k records in G are deleted from the database D, and the process is repeated iteratively, until the database D is empty. Merits/Demerits: Condensation based perturbation technique is designed to preserve the inter attribute correlations of the data. However, the condensation approach is weak in protecting the private data.

P-IRON for Privacy Preservation in Data Mining

413

2.3.3 Rotation Perturbation A rotation perturbation [6] is denoted by G(X) = RX, where RdXd is a randomly generated dXd orthogonal matrix. Here orthogonal matrix is represented as rotation matrix. The definition infers that RT R = R RT = I It also indicates that by altering the order of the rows or columns of rotation matrix, the resulting matrix is still a rotation matrix. If information about the original dataset is not known, then rotation perturbation is a good technique to preserve privacy. Merits/Demerits: Rotation perturbation guarantees privacy and preserves accuracy. Inferring original dataset from the perturbed dataset is possible in rotation perturbation. 2.3.4 Multiplicative Data Perturbation (MDP) Each data element is randomized independently by multiplying with a random number [7]. Two techniques of multiplicative noise are available in the literature. The first tech‐ nique is based on generating random numbers that have been derived from a Gaussian distribution with mean one and small variance. The second technique involves the following steps, taking logarithmic transformation, computing the covariance, and generating random noise using a Gaussian distribution with mean zero and variance equaling a constant times the covariance. Merits/Demerits: MDP assures higher security than ADP and still maintains the data utility very well. Multiplicative data perturbation overcomes the scaling problem. Two possible drawbacks of the MDP are Known input-output (I/O) attack and Known sample attack. 2.3.5 Geometric Data Perturbation Geometric data perturbation [8] includes multiplicative transformation (R), translation matrix (Ψ), and noise matrix Δ. G(X) = RX + Ψ + Δ. The additional components Ψ and Δ are used to address the weakness of rotation perturbation, while still preserving the data quality for classification modeling. Merits/Demerits: Geometric perturbation provides privacy guarantee with small amount of loss in accuracy. 2.3.6

Translation Data Perturbation (TDP), Scaling Data Perturbation (SDP), Hybrid Data Perturbation (HDP) TDP, SDP and HDP [9] are proposed in this paper. Considering D as a Data Matrix and V as a vector space, in TDP, each sensitive attribute xi ϵ V is perturbed using an additive noise perturbation. For the attributes age, salary the sample noise vector N = ({Add, −3}, {Add, 5000}). In SDP, the observations of sensitive attributes in each xi ϵ V are perturbed using a multiplicative noise perturbation. The HDP combines the strength of Translation, Scaling, and Rotation Data Perturbation. Merits/Demerits: New concepts and methods are addressed. This technique preserves the main features of the clusters mined from the original database. It guarantees privacy and preserves accuracy. 2.3.7 Perturbation with Multiple Sensitive Attributes A Privacy Protection Model for Patient Data with Multiple Sensitive Attributes [10] is proposed. This method extends K-anonymity and L-diversity to data with multiple

414

G. Arumugam and V. Jane Varamani Sulekha

sensitive attributes. In the first step, data are divided into a partition such that each partition contains at least K records and satisfies L-diversity. The algorithm follows the K-d tree approach to generate the partition. The benefits of K-d tree is that records with similar values will be put in the same group, thus there will be less data distortion. In the second step data are anonymized. 2.3.8 Attack Techniques in Perturbation However, perturbation is vulnerable to many attacks [11] such as ICA (Independent Component Analysis) Based Attacks, Distance-Inference Attacks, Naive Estimation approach, Rotation center based attack, Attribute Linkage, Known sample and Known Inputs/Outputs attacks. Attribute Linkage: Correlated attributes can be used to identify the original value. Known Sample: The attacker has previous knowledge about the original data or a collection of independent perturbed samples. Known Inputs/Outputs: The attacker knows a small set of sensitive data and the mapping between these known original data and their perturbed counterparts. 2.4 Microaggregation Microaggregation is a perturbative data preserving method. In Microaggregation the individual values are replaced by values computed on small aggregates prior to releasing. In other words, instead of releasing the actual values of the individual records, the system releases the mean of the group (or median, mode, weighted average) to which the obser‐ vation belongs. Microaggregation technique has two phases, partitioning and aggrega‐ tion. In partitioning, the original micro dataset is partitioned into several disjointed clusters/groups so that all records in the same group are very much related to each other and, simultaneously, dissimilar to the records in other groups and in this process cohe‐ sion and coupling is introduced among the data. Additionally, each group is forced to contain at least k records. Different methods exist in microaggregation. In Univariate microaggregation, microaggregation is applied to every individual variable. In contrast, multivariate microaggregation applied to all variables (or subset) in the cluster. Micro‐ aggregation methods can be classified into two types, namely fixed size and data oriented microaggregation. For fixed size microaggregation, the partition is done by dividing a dataset into clusters that have fixed size k, except one cluster which has a size between k and 2k − 1, it depends on the total number of records n and the anonymity parameter k. For the data oriented microaggregation, the partition is based on the data with cluster sizes between k and 2k − 1. Fixed Size microaggregation For Fixed Size microaggregation [12], the grouping is done by dividing a dataset into clusters that have size k, but one cluster may have a size between k and 2k − 1. It depends on the value k and total numbers of records n. Data Oriented univariate microaggregation based on Wards hierarchical algorithm, Genetic algorithm and Fixed size Multivariate microaggregation based on Wards algorithm are discussed. Merits/Demerits: Fixed Size methods reduce space complexity, and thus are more efficient than Data Oriented methods.

P-IRON for Privacy Preservation in Data Mining

415

Data Oriented microaggregation Data oriented methods [13] may achieve lower information loss than Fixed Size methods. The basic idea is to use fixed size heuristics or other algorithms such as nearest point next (NPN) to construct a path traversing all points in a multivariate dataset. Then the multivariate adaptation of Hansen–Mukherjee’s algorithm (MHM) is used on that path. The result is a data oriented k-partition. Merits/Demerits: Data oriented micro‐ aggregation with fixed size k are more efficient than the data oriented variable group size. But variable size microaggregation minimizes the information loss. Maximum Distance based Microaggregation The Maximum Distance (MD) Method [12] is proposed with univariate and multivariate microaggregation method. Using the Euclidean distance two distant records r, s are identified. After the partition, micro aggregated data are computed by replacing each record by the centroid of the group to which it belongs. Merits/Demerits: Effective portioning is possible by using MD based microaggregation but its computational complexity is higher. Maximum Distance to Average Vector Method (MDAV) Maximum Distance to Average Vector Method (MDAV) [14], is a Multivariate Fixed size microaggregation method. It is based on forming groups based on the distance between centroid and distinct data. In MDAV, a square matrix of distances between all records is calculated. After calculating the matrix of distances, MDAV iterates and builds two groups. Merits/Demerits: MDAV is better than MD in terms of computa‐ tional complexity while maintaining the performance in terms of resulting SSE. The disadvantage of MDAV is it’s not flexible. Performance degradation will occur if the data points are scattered in the clusters. Variable-MDAV Variable Size MDAV or V-MDAV [15] in contrast with fixed size MDAV, produces k partitions with group sizes varying between k and 2k − 1. It produces variable size partition. This flexibility can be used to achieve similarity within the group and optimal partition of data. Merits/Demerits: V-MDAV overcomes the limitation of MDAV with the same computational cost. Density based microaggregation A Density Based Microaggregation Algorithm (DBA) [16] is proposed. The DBA has two phases. First Phase (DBA-1), partitions the dataset into groups in which each group contains at least k records. The second phase (DBA-2) is then applied to further tune the partition in order to achieve small information loss and maximum data utility. DBA-2 may decompose the formed groups or may merge its records to other groups. Merits/ Demerits: Minimizes information loss. This method works well with univariate numer‐ ical value. Multivariate Categorical and mixed data values should be researched further. Median based Microaggregation Microdata Protection Method through Microaggregation based on Median [17] is proposed. It divides the whole micro dataset into a number of exhaustive and mutually

416

G. Arumugam and V. Jane Varamani Sulekha

exclusive groups before publication. After grouping it publishes the median instead of individual records. It promises that the modification does not affect the result. Modified data and the original data are similar in this method. T-Closeness through Microaggregation T-Closeness through Microaggregation [18] primarily generates a cluster of size k based on the quasi-identifier attributes. Then the cluster is iteratively refined until t-closeness is satisfied. In the refinement, the algorithm checks whether t-closeness is satisfied and, if it is not, it selects the closest record not in the cluster based on the quasi-identifiers and swaps it with a record in the cluster selected. It takes the t-closeness requirement into account at the moment of cluster formation during microaggregation and this provides best results. Individual ranking based Microaggregation In order to reduce the amount of noise needed to satisfy differential privacy, Utility Preserving Differentially Private Data Releases via Individual Ranking Microaggrega‐ tion [19] is proposed. In individual ranking, each variable is treated independently. Data vectors are sorted by the first variable, then groups of k successive values of the first variable are formed and, inside each group, values are replaced by the group average. Merits/Demerits: Individual ranking owes its popularity to its simplicity and to the fact that it usually preserves more information than one-dimensional projection. Data Recipient centered Microaggregation A data recipient centered de-identification method to retain statistical attributes [20] is proposed. Based on the input from the recipient (the researcher) de-identification can be done because the researchers have a plan of how to use the data. Using Microaggre‐ gation synthetic data are generated. In our work, we are combining perturbation and microaggregation technique. All the existing PPDM techniques including Perturbation and Microaggregation are applied to the whole dataset. In our work, the proposed P-IRON (Preference imposed Individual Ranking based microaggregation with Optimal Noise addition) technique is applied to the dataset with some utility based preferences imposed on certain parameters in the dataset. Preference may be of any kind and different attributes may have different utility. Preference based Variable can be represented as PBV. The following are some examples where utility based preferences can be applied. Disease between age group 30 to 50, PBV = age. Buying pattern of the metropolitan population, PBV = pincode. Buying pattern of a particular age group, PBV = age. Buying behavior and consumption, PBV = salary. Raised cholesterol and obesity level in males over 40, PBV = age. Climatic disease in a particular area, PBV = pincode. Depression, Transportation acci‐ dents, Respiratory conditions and Drug use disorder among the age group 10 to 19, PBV = age. Stress, depression, metabolism and bone problem in females over 40, PBV = age. Data mining is the process of evaluating data from different perceptions and summarizing it into useful information. A typical data mining process depends on data owner to define what kind of pattern they are going to mine or interested in. According to the utility based pattern, selection of data can be done. Instead of releasing the whole dataset, the utility based on the preferences in the parameters of the dataset can be

P-IRON for Privacy Preservation in Data Mining

417

released to improve computing time and storage space. This method also reduces the risk of individual disclosure and data mining algorithm complexity.

3

P-IRON

As Han and Kamber [21] state, a data mining system has the capability to generate thousands or even millions of patterns. But a pattern is interesting if it is potentially useful. Though objective measures help identify interesting patterns, they are often insufficient. It should be combined with subjective measures that reflect a particular user’s interests and needs. For example, patterns describing the disease among patients of a hospital should be interesting to the hospital administration, but may be of little interest to other analysts studying the same database. It is very necessary for data mining systems to generate only interesting and useful patterns. This would be effective for users and data mining systems because neither would have to examine through the patterns generated to identify the really interesting ones. While considering the Elec‐ tronic Health Records (EHR), dataset might be useful for one purpose but useless for another. User provided constraints and interestingness measures should be added with data mining process to obtain completeness of mining. Generally, it is not the respon‐ sibility for a data owner to build models, but it is the responsibility for a data owner to keep privacy when the data are released. The data owner has to execute a privacy protection technique with different preference based parameters to attain a desired tradeoff between privacy and utility. Considering this in our mind we propose a novel privacy preserving technique P-IRON. P-IRON Combines preference based Microaggregation by Individual Ranking and optimal ɛ differential privacy based perturbation which ensures low information loss and guarantees privacy and utility. Existing microaggre‐ gation techniques replace the original values with computed aggregates like mean, median, mode and centroid. These aggregated values can be reconstructed and may violate privacy. Reconstruction won’t be possible in P-IRON. The data owner can also choose a preference based dataset [22] from a set of non-dominated dataset. P-IRON technique can be divided into two major parts Microaggregation and Optimal Noise Addition. In Microaggregation phase, K ward hierarchical clustering algorithm [23] is used to partition the dataset. Individual ranking is a popular microag‐ gregation method. In individual ranking, each variable is treated independently. In our work we are taking the variable age as PBV and individual ranking is done using age. By using K-ward algorithm, dataset is grouped into n partitions based on the PBV. Then groups of k successive values of the PBV are formed and, inside each group, values are replaced by the group mean. A similar technique is repeated for the rest of the variables if we want to use this method for multivariate microaggregation. Individual sorting usually preserves more personal information. After the microaggregation, optimal noise is added to each micro aggregated value and this perturbed dataset is released for mining. For numerical attributes noise is usually added using a random number. This random number is generally derived from a normal distribution with small standard deviation and zero mean. Noise is added in a controlled way so that it won’t affect the mining result. X denotes all the attributes of the original dataset. X′ denotes the perturbed

418

G. Arumugam and V. Jane Varamani Sulekha

dataset. When the original data is replaced with the cluster mean, the sensitivity of the dataset will be represented as Δx/k. where Δx is the distance between the most distant records in the cluster. The sensitivity of the whole dataset is n/k × Δx/k. To obtain differential privacy, Laplace noise (n/k × Δx/k)/ɛ is added to the numerical data. Laplace noise is not optimal. Optimal Noise can be expressed as follows. Let N1 and N2 be two random noise distributions. If N1 can be constructed from N2 by moving some of the probability mass towards zero, then N1 must always be preferred to N2. The reason is that the probability mass of N1 is more concentrated around zero, and thus the distortion introduced by N1 is smaller. A rational user always prefers less distortion and, therefore, prefers N1 to N2. A random noise [24] distribution N1 is optimal within a class C of random noise distributions if N1 is minimal within C; in other words, there is no other random N2 ∈ C such that N2 < N1.

4

Pseudocode of Our Proposed Work

Step 1. Form a cluster using individual ranking with the first k elements of the original dataset and another group with the last k elements of the original dataset Step 2. Use Wards method until all elements in the original dataset belong to a group containing k or more data elements. In this process of forming groups by Wards method, never join two groups which have both a size greater than or equal to k. Step 3. For each group in the final partition that contains 2k or more data elements, apply this algorithm recursively. Within each cluster, the entire attribute values are replaced by the cluster mean, so each micro aggregated cluster consists of k repeated mean values. Step 4. Add Optimal Noise (ON), (n/k *Δx/k)/ɛ to each attribute in the clusters. The first step ensures that in each recursive step the dataset is split into at least 2 groups. The second step ensures that the formed groups are never combined because of their size. Third step guarantees k anonymity, with 2k or more elements. The last step ensures privacy of individual record. We combine Individual ranking based microaggre‐ gation and optimal ɛ differential privacy. This combination gives better performance, low information loss and ensures privacy. The main difference between our proposed tech‐ nique with the previous microaggregation algorithm is that, the given method can get privacy preserved multi partitioned univariate (singe attribute based) numerical dataset. In each partition, the perturbation method applied is different (different noise addition for each partition), so it may restrict the reconstruction problem. The perturbed dataset obtained from original dataset will give the same mining result while applying classifica‐ tion or clustering algorithm. This method reduces the risk of individual disclosure. Chronic kidney disease (CKD) is now-a-days common among the middle age. For example, if a hospital wishes to know the CKD among the age group 40 to 50, here the preferred utility based pattern is CKD among the age group 40 to 50. In this work, age is individual ranking imposed PBV. Partition is done on age and the preference based perturbed dataset between age group 40 to 50 is released for mining. To ensure the individual’s privacy, the preference based dataset is micro aggregated and added with

P-IRON for Privacy Preservation in Data Mining

419

optimal Laplace Noise, before releasing it for mining. Considering the partition as n = 3 and the clusters are named as c1, c2, c3. The cluster c1 has values between 1 to 39, c2 has values between 40 to 50 and c3 has values between 51 to 90. Table 1 shows Sample dataset. Table 1. Sample patient data Age 39 68 41 20 33 80 75 44 49

BP 100 80 100 90 100 100 100 80 100

al 3 0 3 0 3 3 3 0 3

Rbcc 2.8 4.5 2.8 4.0 2.0 2.5 2.5 4.5 2.8

Alb 1 2 0 2 2 2 2 2 1

Class ckd notckd ckd notckd ckd ckd ckd notckd ckd

The proposed algorithm is applied to the sample patient dataset and the intermediate results of the clusters are shown in Table 2. Original dataset is partitioned into 3 groups. Each group cluster values are replaced with mean of that group and Optimal Noise is added to the mean value. In the final phase, preference based clusters are released for mining. Table 2. Clusters c1, c2, and c3 Age 20 33 39 Age 41 44 49 Age 68 75 80

BP 90 100 100 BP 100 80 100 BP 80 100 100

al 0 3 3 al 3 0 3 al 0 3 3

Rbcc 4.0 2.0 2.8 Rbcc 2.8 4.5 2.8 Rbcc 4.5 2.5 2.5

Alb 2 2 1 Alb 0 2 1 Alb 2 2 2

Class notckd ckd ckd Class ckd notckd ckd Class notckd ckd ckd

Table 3 shows the privacy preserved patient data. For the first partition the ON = 1.05, the second partition ON = 1.4 and for the third partition ON = 1. Here we are having 3 Partitons, our preference is age group between 40 to 50. So the second partion alone can be released to dataminers for analysis.

420

G. Arumugam and V. Jane Varamani Sulekha Table 3. Privacy preserved multi partitioned univariate numerical dataset Age 31 31 31 Age 46 46 46 Age 75 75 75

BP 90 100 100 BP 100 80 100 BP 80 100 100

al 0 3 3 al 3 0 3 al 0 3 3

Rbcc 4.0 2.0 2.8 Rbcc 2.8 4.5 2.8 Rbcc 4.5 2.5 2.5

Alb 2 2 1 Alb 0 2 1 Alb 2 2 2

Class notckd ckd ckd Class ckd notckd ckd Class notckd ckd ckd

Indeed, with individual ranking any intruder knows that the real value of an element in the ith group is between the average of the i − 1th group and the average of the i + 1th group. If these two averages are very close to each other, then a very narrow interval for the real value being searched has been determined. Individual ranking is less vulnerable to inference attack.

5

Experimental Results

CKD dataset obtained from Bethel hospital, Madurai, Tamilnadu, India is utilized. Original CKD dataset consists of 1200 records with 20 attributes. For building the Pref‐ erence based privacy preserved dataset the computing time is lesser. Then after applying P-IRON technique, the mining process also takes less time while using the preference based dataset. First we compared the time taken to mine the original dataset with the preference based dataset (2nd partition alone) using WEKA tool. We used ZeroR clas‐ sifier in WEKA tool to classify the CKD dataset. The synthetic dataset is generated from the original CKD dataset. The synthetic dataset consists of 11, 40, 243 records and storage space is 147 MB. After applying the P-IRON technique taking PBV as age, the preference based dataset consists of 5, 36, 346 records and storage space is 54 MB. Table 4, shows the time taken to mine the original dataset and preference based dataset. Next we compared the mining result of the original dataset with the privacy preserved full dataset using WEKA tool. Table 5, shows the classification results of the original and P-IRON technique imposed dataset. Our experiments reveal that our framework is effective, meets privacy requirements, and guarantees valid data mining results while protecting sensitive information. Our proposed method performed well and produced valid data mining results.

P-IRON for Privacy Preservation in Data Mining

421

Table 4. Time taken for the original and preference based dataset Original dataset Scheme: weka.classifiers.rules.ZeroR Instances: 1140243 Time taken : 0.42 seconds === Confusion Matrix === a b