Trust and behavior analysis-based fusion method for heterogeneous multiple attribute group decision-making

• Published in: Computers & industrial engineering Vol. 152; p. 106992
• Main Authors: Yu, Su-min, Du, Zhi-jiao, Wang, Jian-qiang, Luo, Han-yang, Lin, Xu-dong
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.02.2021
• Subjects: heterogeneous multiple attribute group decision-making (HMAGDM); Individual alternatives; Individual attributes; Selection behavior; Trust and behavior analysis-based fusion method; Trust relationship; Trust and behavior analysis-based fusion method; Individual alternatives; Trust relationship; Individual attributes; heterogeneous multiple attribute group decision-making (HMAGDM); Selection behavior
• ISSN: 0360-8352; 1879-0550
• DOI: 10.1016/j.cie.2020.106992

•The selection behaviors for attributes are analyzed and classified into three cases.•Trust is used to aid in calculating the distance of heterogeneous information.•A weight determination method considering decision information clarity is proposed.•A trust and behavior analysis-based fusion method for HMAGDM is developed. In classical multiple attribute group decision-making (MAGDM), experts are usually required to evaluate a predefined set of alternatives according to a predefined set of attributes. However, in real-world MAGDM scenarios, experts may use individual sets of attributes to evaluate individual alternatives due to differences in knowledge and experience among experts and the nature of the decision-making problem. In this case, the individual sets of alternatives and attributes will be heterogeneous. A new trend in group decision-making methodology is related to social networks. Individuals decisions are influenced by those with whom they have close relationships or social connections. To address this type of heterogeneous MAGDM, this study proposes a trust and behavior analysis-based fusion method. First, we analyze the behaviors of experts in choosing alternatives and attributes, and classify them into three categories: empty, positive, and negative. Then, distance measures among heterogeneous evaluation information belonging to different categories of selection behaviors are defined and guided by trust values derived from the trust relationships among experts. A weight determination method for experts considering the clarity of evaluation information is developed. Based on these, a complement method that aims to populate positions that are not assigned values is designed. Finally, the case study and comparative analysis illustrate the feasibility and characteristics of the proposed fusion method.