Abstract Online social networks (OSNs) exhibit complex multi-relational and multi-group dynamics that traditional graph-based models struggle to capture. Existing SIR-based misinformation propagation models often overlook the influence of fake accounts and the disparities between single-modal and multi-modal information diffusion. To address these challenges, this paper introduces the B-TCSR evolutionary model, which is based on a hypergraph structure. This model innovatively integrates the co-evolutionary dynamics of different user roles, accounting for the interactions between fake and genuine users. Furthermore, the B-TCSR model examines the influence of single-modal and multi-modal content characteristics on the dissemination of misinformation. Theoretical analysis confirms the existence of a unique global positive solution in random models and establishes sufficient conditions for both misinformation extinction and persistence. Notably, it is observed that fake accounts significantly enhance the spread of misinformation. Moreover, analysis using Partial Rank Correlation Coefficients demonstrates that the hypergraph-based model is more sensitive to parameter variations than traditional graph structures. Through practical case studies, the reliability and applicability of the results are validated. This work pioneers a new paradigm for analyzing multi-group misinformation dynamics, offering actionable insights for designing robust misinformation mitigation strategies in OSNs.
Zhang et al. (Wed,) studied this question.