Public Opinion Dissemination and Prediction based on Exponential Random Graph Models

Description

This study investigates the mechanisms and prediction of public opinion dissemination in social networks using Exponential Random Graph Models (ERGMs). By analyzing social network data from TikTok and Weibo, the research explores the differences in public opinion dissemination characteristics across these platforms. Through Exploratory Data Analysis (EDA) and ERGMs modeling, the study identifies key factors influencing network relationships, finding that network sparsity, sentiment homophily, geographic location, and user interaction intensity significantly affect information spread. On TikTok, content quality and sentiment matching are crucial for information dissemination, with the platform’s algorithm-driven recommendation system accelerating content spread. In contrast, Weibo’s dissemination is more reliant on social relationships and the existing network structure. The study also finds that the effect of triadic closure on both platforms is minimal, suggesting that the influence of "friends of friends" on public opinion dissemination is limited. These findings provide a theoretical foundation for understanding and predicting public opinion pathways in social networks and offer practical insights for opinion management and public crisis control. By comparing the dissemination mechanisms of the two platforms, the study highlights the significant impact of platform characteristics, particularly during periods of information overload. The algorithm-driven content spread on TikTok and the social network structure on Weibo create distinct dissemination patterns. However, the study acknowledges its limitations in terms of sample size and time span, laying the groundwork for further
cross-platform comparisons and dynamic analyses in future research.

Period	1 Jan 2024 → 24 Dec 2024
Degree of Recognition	International