Cross‐Networks and Cross‐Hierarchy Dissemination of Hot Issues Based on <scp>BERTopic</scp> Model Topic Clustering and Social Network Analysis Methods

ABSTRACT In the social media era, understanding how hot issues diffuse across user communities and influence hierarchies is critical for mapping public‐opinion dynamics. This study combines BERTopic clustering, social network analysis, and hierarchical user stratification to examine diffusion mechanisms in two emblematic Weibo hot issues. First, a temporal sentiment analysis reveals a sustained predominance of negative sentiment across both issues. Second, BERTopic clustering identifies 6 dominant viewpoint groups, showing that diffusion occurs within relatively homogeneous discursive clusters. Third, network‐structural analysis further demonstrates that both issues exhibit a multi‐center, loosely coupled subnetworks configuration. Yet, the nature of cross‐community connectivity diverges substantially: participation coefficient metrics indicate that the policy debate features a distributed bridging structure jointly occupied by mainstream media and high‐influence opinion leaders, whereas the commercial issue is characterized by a near‐monopoly of cross‐network flow by opinion leaders and an absence of media presence. Finally, cross‐hierarchy diffusion analysis shows a positive association between the influence level of mainstream media/opinion leaders and viewpoint diversity, but a weak negative association between cross‐tier retweet frequency and the breadth of perspectives transmitted, which suggests that upward and downward information flows remain structurally possible yet functionally constrained. Summary: these findings reveal that issue diffusion on Chinese social media is shaped not only by content and sentiment but also by the stratified and uneven structure of cross‐network connectivity. The study provides empirical evidence that mainstream media and opinion leaders function as key agenda setters, whose heterogeneous bridging capacities critically determine whether hot issues circulate widely or remain siloed within homogeneous clusters.