User Dual Intents Graph Modeling for Information Diffusion Prediction

The goal of the information diffusion prediction task is to provide a set of information and predict the target users during the information diffusion process. In this study, we start from user intention and divide the process of information dissemination into two parts: interest intention driven information dissemination and social intention driven information dissemination. Based on this, we propose a user dual intents graph modeling for information diffusion prediction model. For interest intentions, hypergraphs are used to process historical interaction information. For social intention criteria, user social relationships and global information dissemination relationships are considered. In the processing of global information dissemination relationships, we add temporal features to capture time preferences in information dissemination. Finally, the intention is fused to achieve the prediction task. We conducted experiments on four datasets and demonstrated good results.