Fake News Detection using Higher-order User to User Mutual-attention Progression in Propagation Paths

Social media has become a very prominent source of news consumption. It brings forth multifaceted, multimodal and real-time information on a silver platter for the users. Fake news or rumor mongering on social media is one of the most challenging issues pertaining to present web. Previously, researchers have tried to classify news propagation paths on social media (e.g. Twitter) to detect fake news. However, they do not utilize latent relationships among users efficiently to model the influence of the users with high prestige on the other users, which is a very significant factor in information propagation. In this paper, we propose a novel Higher-order User to User Mutual-attention Progression (HiMaP) method to capture the cues related to authority or influence of the users by modelling direct and indirect (multi-hop) influence relationships among each pair of users, present in the propagation sequence. The proposed higher order attention trick is a novel contribution which can also be very effective in case of transformer architectures[30]. Our model not only outperforms the state-of-the-art methods on two publicly available Twitter datasets but also explains the propagation patterns pertaining to fake news by visualizing higher order mutual-attentions.