Weight Penalty Mechanism for Noncooperative Behavior in Large-Scale Group Decision Making With Unbalanced Linguistic Term Sets

Abstract

This paper proposes a novel framework to manage subgroups' non-cooperative behavior by weight penalty in largescale group decision making (LSGDM). To do that, a trustconsensus index (TCI) is defined by combining trust score and consensus degree among experts expressed by unbalanced linguistic term sets. A Louvain algorithm clustering process based on undirected graph composed of TCI is introduced to detect the subgroups in large network. Hence, a weight penalty feedback model is established to manage the subgroups detected as discordant and non-cooperative. The proposed method novelty resides in that the minimum adjustment cost can be obtained with respect to the penalty parameter. A detail analysis regarding the computation of the optimal penalty parameter to prevent excessive penalization is reported. Finally, a detailed numerical and comparative analyses are provided to verify the validity of the proposed method.