Hierarchical Multi-head Attentive Network for Evidence-aware Fake News Detection

The widespread of fake news and misinformation in various domains ranging\nfrom politics, economics to public health has posed an urgent need to\nautomatically fact-check information. A recent trend in fake news detection is\nto utilize evidence from external sources. However, existing evidence-aware\nfake news detection methods focused on either only word-level attention or\nevidence-level attention, which may result in suboptimal performance. In this\npaper, we propose a Hierarchical Multi-head Attentive Network to fact-check\ntextual claims. Our model jointly combines multi-head word-level attention and\nmulti-head document-level attention, which aid explanation in both word-level\nand evidence-level. Experiments on two real-word datasets show that our model\noutperforms seven state-of-the-art baselines. Improvements over baselines are\nfrom 6\\% to 18\\%. Our source code and datasets are released at\n\\texttt{\\url{https://github.com/nguyenvo09/EACL2021}}.\n