关键词: 网络暴力, 多阶段博弈, 不完全信息, 信念更新

Abstract:

Cyber violence incidents are frequent, and current governance methods often lag behind. This paper introduces a multi-stage dynamic game model framework to explore optimal strategies for governing cyber violence between governments and platform operators. The authors employ Bayesian updating rules to address conditional probability issues in scenarios with incomplete information and perform equilibrium analyses for both single-stage and multi-stage games. A recursive algorithm (Q-Strategy) is developed to determine equilibrium strategies at each stage. The model's validity is confirmed through case studies and numerical simulations. The results indicate that strong regulatory measures by the government significantly reduce the dissemination of false information by platform operators and effectively curb cyber violence. Increasing the initial belief value shortens the propagation period and decreases overall dissemination intensity. Furthermore, high-cost strategies employed by platform operators effectively reduce the motivation to spread false information. Using the "woman defamation incident during package pickup" in Hangzhou as a case study, the authors propose governance measures for the incubation, dissemination, and waning phases.

Key words: cyber violence, multi-stage games, incomplete information, belief updates