Modeling Information Diffusion With Sequential Interactive Hypergraphs

In online social networks, numerous users generate and distribute tremendous content simultaneously, how information spreads in society has attracted critical attention. Recent studies introduce deep learning to help mine the complex diffusion pattern forecast trend. However, there are two limitations of state-of-art methods. On one hand, existing models only focus on internal influence within flow but ignore external from dissemination other contents. dynamics user interest barely considered, while changes preference for contents also have a considerable impact future diffusion. To address these issues, we hypergraph structure sequential framework model interactions networks. Then, propose dual-channel neural networks tackle prediction problem, denoted by HyperINF. Specifically, channel, build interactive hypergraphs learn dynamic representation, construct graph capture cross-diffusion relation. At last, consider relation make prediction. Experimental results three datasets suggest effectiveness practicability proposed framework.