Detecting rumors in social media using emotion based deep learning approach

Abstract

Social media has become a vital platform for information dissemination. However, this ease of sharing can also facilitate the spread of unverified and potentially damaging rumors, negatively a↵ecting society and individuals. Given the vast amount of content generated on social media, there is a critical need for methods to assess information veracity and ensure factual accuracy. Existing research has investigated various approaches for rumor detection, including feature engineering and deep learning techniques, leveraging propagation theory to identify rumors. Our research builds upon this foundation by emphasizing the role of emotions and sentiment analysis in tweets, employing deep learning methods to enhance rumor detection accuracy. Leveraging insights from prior studies, a Sentiment and EMotion driven TransformEr Classifier method (SEMTEC) is proposed. Unlike previous models, SEMTEC incorporates the extraction of emotional and sentiment tags alongside content-based information from the main tweet text. This comprehensive semantic analysis allows us to gauge user emotional states, leading to a remarkable improvement in accuracy in rumor detection. The proposed method is tested and compared with existing techniques on standard datasets and shown to be e↵ective. This performance significantly surpasses that of existing state-of-the-art models.