Temporally evolving graph neural network for fake news detection

The proliferation of fake news on social media has the probability to bring an unfavorable impact on public opinion and social development. Many efforts have been paid to develop effective detection and intervention algorithms in recent years. Most of the existing propagation-based fake news detection methods focus on static networks and assume the whole information propagation network structure is accessible before performing learning algorithms. However, in real-world information diffusion networks, new nodes and edges constantly emerge. Therefore, in this paper, we introduce a novel temporal propagation-based fake news detection framework, which could fuse structure, content semantics, and temporal information. In particular, our model can model temporal evolution patterns of real-world news as the graph evolving under the setting of continuous-time dynamic diffusion networks. We conduct extensive experiments on large-scale real-world datasets and the experimental results demonstrate that our proposed model outperforms state-of-the-art fake news detection methods.