TaHiD: Tackling Data Hiding in Fake News Detection with News Propagation Networks

Anonymous

Keywords: Fake news detection, Graph neural networks, News propagation networks, Data hiding

Abstract: Fake news with detrimental societal effects has attracted extensive attention and research. Despite early success, the state-of-the-art methods fall short of considering the propagation of news. News propagates at different times through different mediums, including users, comments, and sources, which form the news propagation network. Moreover, the serious problem of data hiding arises, which means that fake news publishers disguise fake news as real to confuse users by deleting comments that refute the rumor or deleting the news itself when it has been spread widely. Existing methods do not consider the propagation of news and fail to identify what matters in the process, which leads to fake news hiding in the propagation network and escaping from detection. Inspired by the propagation of news, we propose a novel fake news detection framework named TaHiD, which models the propagation as a heterogeneous dynamic graph and contains the propagation attention module to measure the influence of different propagation. Experiments demonstrate that TaHiD extracts useful information from the news propagation network and outperforms state-of-the-art methods on several benchmark datasets for fake news detection. Additional studies also show that TaHiD is capable of identifying fake news in the case of data hiding.

Paper Type: long