Rumor Suppression in a Three-Layer Network: A Reinforcement Learning Algorithm

Abstract

Rumor propagation poses a significant threat to social stability and public order, and controlling its spread can effectively reduce unnecessary panic and misunderstanding. Rumor control is primarily achieved by simulating rumors spread on social networks and disseminating the truth or restricting propagation pathways. However, current studies usually only apply the optimal control theory, which leads to difficulties in coping with complex and stochastic network propagation environments. To address these issues, this paper constructs a three-layer network rumor control model (SICR-3M3W) that considers the dual refutation mechanism and formulates an optimal control problem for this model. Based on the reinforcement learning framework, we design a Proximal Policy Optimization (PPO) algorithm to solve this problem intelligently. Finally, experiments based on a real-world data case are conducted, and the results demonstrate that our three-layer model can effectively simulate the rumor propagation process. Moreover, the designed PPO controller can achieve optimal control outcomes.