Search

With the rise of social networks, various types of information have emerged in the vision field in a complex manner, making it crucial to analyze the propagation patters of online public opinion to effectively guide information dissemination. To elucidate the dynamics of information dissemination, this paper proposes a directed network information based on self-organized network information dissemination scenario. This model takes into account the influence of networks formed by users spontaneously and distinguishes the dissemination population based on the in- and out-degree of user nodes in the network. To assess the model’s performance, it is evaluated using real retweets from Chinese Sina Weibo, considering the impact of user interactions on information dissemination. Comparing real data with model-fitted data, the proposed model-based evaluation and numerical analysis demonstrate that the forwarding and transfer probabilities align with actual information dissemination. Furthermore, the evaluation sensitivity analyses describe the key factors influencing information dissemination, aiding decision-making in formulating strategies to guide public opinion. To quantify the importance of these factors, assessment metrics are introduced, such as the propagation regeneration number.

In order to avoid forming an information cocoon, the information propagation of COVID-19 is usually created through the action of "proactive search", an important behavior other than "reactive follow". This behavior has been largely ignored in modeling information dynamics. Here, we propose to fill in this gap by proposing a proactive-reactive susceptible-discussing-immune (PR-SFI) model to describe the patterns of co-propagation on social networks. This model is based on the forwarding quantity and takes into account both proactive search and reactive follow behaviors. The PR-SFI model is parameterized by data fitting using real data of COVID-19 related topics in the Chinese Sina-Microblog, and the model is calibrated and validated using the prediction accuracy of the accumulated forwarding users. Our sensitivity analysis and numerical experiments provide insights about optimal strategies for public health emergency information dissemination.