Rumor Detection with News Environment Enhanced Propagation Structure

Yang Wu, Yanqiang Zhang, Zhen Xu, Qian Tan, Yan Zhang, Pengwei Zhan

Abstract: The pervasive proliferation of rumors on social media has led to significant societal harm. As a result, numerous studies have concentrated on detecting rumors by examining their propagation patterns or their surrounding news context. However, current methods overlook the interplay between rumor propagation and the news environment, which can provide significant insights. For instance, the user responses in the news environment may help find relevant news for detecting the underlying rumor. To address this gap, we propose a novel rumor detection model, Enhanced Propagation Graph Convolutional Network (EPGCN), which captures rumor signals by considering both propagation and the news environment context. We develop the original propagation graph by incorporating relevant news nodes and then utilize Graph Convolutional Network (GCN) to learn the global structural features of the enhanced propagation graph. Experiments conducted on two public datasets demonstrate that our model outperforms existing methods.