The study focuses primarily on information diffusion processes in higher-order networks. Revealing the patterns governing information diffusion in social networks serves as an important means of understanding the mechanisms underlying collective opinion formation and supporting cyberspace governance decisions. Traditional opinion dynamics models are typically based on static network topologies and pairwise interaction assumptions. However, such modeling approaches cannot simultaneously capture the dynamic evolution of users’ online and offline states during information diffusion and the synergistic effects generated by higher-order interactions, thereby introducing deviations into their descriptions of real-world diffusion processes.
To address these issues, the study first integrates a mechanism for switching between users’ online and offline states and constructs a temporal social hypernetwork tailored to real-world social scenarios. On this basis, the study proposes a continuous opinion dynamics model with a three-level coupled “individual-clique-group” mechanism. The proposed model incorporates individual forgetting and interest-retention mechanisms, time-varying opinion tolerance among clique members, and group consensus thresholds and consensus-pressure mechanisms, thereby characterizing opinion evolution jointly driven by individual cognition, interactions among clique members, and group-level effects. Second, the study employs effective mean-field approximation theory to analyze the global steady-state structure and bifurcation mechanisms. The study delineates the steady-state boundaries of the model and analyzes the dynamical mechanism underlying the transition from monostability to multistability. Finally, the study performs a parameter-sensitivity analysis on synthetic hypernetworks and selects three real-world social datasets for simulation-based validation, thereby evaluating the robustness of the model and its adaptability to real-world social evolution patterns.
The study delineates the boundary conditions for the local asymptotic stability of the model’s neutral monostable state through theoretical analysis and reveals the mechanism of asymmetric bifurcation and the emergence patterns of multistability under temporal higher-order interactions. In synthetic networks, simulation results show that individual opinion polarization in temporal social hypernetworks is positively correlated with the node activity ratio and opinion tolerance among clique members, but negatively correlated with the forgetting rate and consensus threshold. Moreover, compared with other opinion dynamics models, the proposed model demonstrates stronger preference responsiveness and a greater capacity to express opinion differentiation. The study selects three real-world social datasets and constructs corresponding temporal social hypernetworks based on their original social relationships. The experimental results show that the proposed model can effectively capture the multistability emerging during public-opinion evolution across multiple real-world scenarios. On this basis, empirical results suggest that hyperedge size may be associated with the emergence of multistability.
Based on the findings of the study, future work will further investigate the effects of hyperedge size on the proposed opinion dynamics model, other information diffusion paradigms, and diffusion behavior in complex groups. Future work will also analyze the mechanisms through which different hyperedge sizes and strengths of group interactions influence diffusion processes, with the aim of developing a more comprehensive and multidimensional theoretical framework for information diffusion and regulation. In addition, subsequent studies will build on the temporal social hypernetwork opinion dynamics model developed by the study to further explore the optimal design of information-control and diffusion-control strategies under resource-constrained scenarios.