Fake News Detection Based on Subjective Opinions

Abstract

Fake news fluctuates social media, leading to harmful consequences. Several types of information could be utilized to detect fake news, such as news content features and news propagation features. In this study, we focus on the user spreading news behaviors on social media platforms and aim to detect fake news more effectively with more accurate data reliability assessment. We introduce Subjective Opinions into reliability evaluation and proposed two new methods. Experiments on two popular real-world datasets, BuzzFeed and PolitiFact, validates that our proposed Subjective Opinions based method can detect fake news more accurately than all existing methods, and another proposed probability based method achieves state-of-art performance.