POF-HG: Fusion of Public Opinion Field Effect and Heterogeneous Hypergraph for Information Diffusion Prediction

Information diffusion prediction is a core challenge in social network analysis, relevant to viral marketing, misinformation detection, and public health monitoring. Existing approaches face two key limitations: (1) traditional graph-based methods rely on pairwise user links and fail to capture the higher-order group interactions present in real social systems, and (2) current models treat information cascades independently, overlooking the fact that multiple topics compete for users’ limited attention. To overcome these issues, we propose POF-HG (Fusion of Public Opinion Field Effect and Heterogeneous Hypergraph), a framework that integrates the public opinion field effect with heterogeneous hypergraph learning. The opinion-field module quantifies competitive dynamics among coexisting topics by modeling how each topic attracts users based on its energy in the network. Meanwhile, the heterogeneous hypergraph captures multi-way relationships among users and messages to extract richer structural semantics. A co-attention fusion module combines social influence patterns with opinion-field features to generate interaction-aware user embeddings for next-user prediction. Experiments on four real-world datasets show that POF-HG significantly outperforms state-of-the-art baselines.