The rapid development of social networks has significantly accelerated information dissemination, but it has also intensified the risk of large-scale rumor propagation in areas of public concern, such as healthcare. To address this issue, this paper proposes an intelligent rumor governance system that integrates text recognition, diffusion simulation, and reinforcement learning control. The system constructs a dataset containing factual information and potential rumors, and employs TF-IDF features and a Random Forest model to achieve high-accuracy rumor detection. For information diffusion modeling, the system simulates rumor spread on an ErdsRnyi network and designs a Q-learning-based controller to intervene in the process. Experimental results show that the Random Forest-based rumor detection model achieves an accuracy of over 85%, demonstrating strong classification performance. The reinforcement learning control strategy effectively curbs information diffusion, reducing the number of infected nodes by 40%60%, and achieves the best overall performance among multiple strategies, particularly in balancing containment effectiveness, implementation cost, and user experience, highlighting its strong potential for real-world applications.
Yufei Jin (Wed,) studied this question.