Rumour Detection Based on Graph Convolutional Neural Net

Rumor detection is an important research topic in social networks, and lots of rumor models are proposed recent years. For the task, structural information a conversation can be used to extract effective features. However, many existing focus on local features while global between source tweet its replies not effectively used. To make full use content information, we propose Source-Replies relation Graph (SR-graph) for each conversation, which every node denotes tweet, feature weighted word vectors, edges denote interaction tweets. Based SR-graphs, Ensemble Convolutional Neural Net with Nodes Proportion Allocation Mechanism (EGCN) task. In experiments, first verify that extracted effective, then show effects different word-embedding dimensions multiple test indices. Moreover, our EGCN model comparable or even better than current state-of-art machine learning models.