Content based Fake News Detection Using Knowledge Graphs

This paper addresses the problem of fake news detection. Although there are many works already in this space, most of them are not using the content itself for the decision making. In this paper, we propose some novel approaches to detecting fake news by making use of knowledge graphs. There are a few technical challenges. Firstly, state of the art triple extraction tools are still far from perfect. Secondly, it is challenging to validate the correctness of the extracted triples. Thirdly, open knowledge graphs, such as DBPedia, are not comprehensive enough to cover all the relations needed for fake news detection. To address these challenges, we propose three approaches, which are evaluated with Kaggle's "Getting Real about Fake News" dataset. Our studies indicate some insights, despite the above mentioned challenges, on using knowledge graph for fake news detection.