Detecting Arabic Fake News on Social Media using Sarcasm and Hate Speech in Comments

Abstract

The rapidly increasing popularity of social networking sites and the widespread acceptance of anonymous users have encouraged an environment where unidentified accounts can act maliciously and propagate fake news. The motivation behind that could either be to begin hype or to gain individuals' attention and negatively impact society. Several studies attempt to establish models to detect fake news based on news content, source, or propagation path. However, fewer studies have investigated more profound signs, such as people's responses to the information posted on social media. We hypothesize that the existence of sarcasm or hate language in the comments and responses to a news post may be used as an indicator of the authenticity of the post itself. Therefore, this paper proposes a new technique incorporating hate language and sarcasm detected in users' comments as significant features for identifying fake news. We used three Arabic datasets to conduct this study and experimented with various state-of-the-art models. As a result, we conclude that considering these features in news responses can help detect fake news since we found that the existence of sarcasm or hate speech in comments of false tweets is approximately double that in true ones.