Modeling and Analysis of Rumor Propagation Dynamics in Social Media

Exploring the nature of rumor propagation in social media is of great importance to mitigate the adverse effects of rumors and prevent rumor dissemination. To accurately model the complexity of rumor propagation dynamics in real social media and effectively ameliorate the effects of rumors on society, a new rumor propagation model named V-SEIR is proposed in this paper. The proposed V-SEIR model considers the node heterogeneity, individual behaviors, and network topologies, which can affect the node influence, prediction accuracy, and model generality, respectively. The rumor-free equilibrium point and the basic reproduction number \(R_0\) are calculated to assess the model stability. The nodes are divided into important and ordinary parts based on their degree, to characterize node heterogeneity. Three non-linear metrics are defined and estimated: influence, bandwagon effect, and forgetting mechanism, to capture the individual behaviors. The empirical analysis reveals the dynamic behavior and complexity of rumor propagation. The V-SEIR model is evaluated in different network topologies and the impact of each metric parameter is verified. By comparing the V-SEIR model with the classic propagation model and validating it with actual rumor datasets from Weibo, the results demonstrate that the V-SEIR model has higher prediction accuracy, especially in the early and late stages of rumor propagation. This research can enhance our understanding of rumor propagation, making it possible to forecast and manage online public opinion in social media.