Abstract: In order to reduce divergences among experts and enhance the scientific rigor of decision results, the consensus reaching process(CRP) has become a crucial stage in group decision-making. During the CRP, conformity behavior may exert a significant influence on both the decision process and its outcomes. However, most existing studies assume conformity behavior to be static, which deviates from the complex and ever-changing reality. To address this limitation, this study proposes a social network large-scale group decision-making method that incorporates dynamic conformity behavior. Specifically, a K-medoids clustering algorithm is employed to partition the large group into several subgroups. A dynamic conformity degree is then defined by jointly considering individual and group factors, based on which a dynamic ideal adjustment point is introduced to model the dynamic conformity behavior. Furthermore, by dynamically adjusting the limited budget based on subgroup consensus levels, a maximum consensus model under dynamic conformity behavior is constructed. Numerical results demonstrate that the presented method not only substantially reduces decision-making costs but also effectively improves consensus levels. The presented method exhibits strong consensus efficiency and behavioral adaptability, thereby offering a novel theoretical perspective and a practical framework for modeling dynamic conformity behavior and supporting complex group decision-making under evolving environments.

Key words: dynamic conformity behavior, consensus reaching process, consensus model, social network large-scale group decision-making