Network-based evidential three-way theoretic model for large-scale group decision analysis

Abstract

Social relationships are critical to the group decision-making (GDM) process, especially for large-scale scenarios. Conventional GDM models have several drawbacks when applied to large-scale GDM problems. In this paper, we propose an evidential three-way theoretic model for large-scale group decision analysis based on the introduction of ego networks. A similarity matrix of all individuals is obtained after ego network generation via social network feature extraction. Rough and smooth detection are then conducted in the framework of three-way decisions. Specifically, the degree of organizational influence is analyzed based on the generated basic probability assignments (BPAs), and the individuals are divided into several organizations. After an opinion collection process, preference evolution is implemented via a social influence network (SIN) technique and a fuzzy preference relation (FPR) model. Then, the global final scores of all the alternatives are obtained using an aggregation process. Finally, we conduct a simulation experiment to illustrate the entire procedure. Based on a comparison of related methods, we believe that the proposed method can reasonably solve real-world large-scale group decision-making (LSGDM) problems and has good practicability and effectiveness.