Emotion Cum False Information Detection Based on Sentiment, Structural and Semantic Frameworks

The existence of false information in Online Social Networks (OSNs) can pose a serious threat to the ecology of social information dissemination, and this statement can arouse netizen's feelings of fear and sadness and affect their social, mental, and physical health. Many researchers focus on emerging problems related to detecting false information and investigating the emotional framing of misinformation in social media. However, detecting false information combined with examining public emotion dynamics over temporal social interaction needs to be fully explored. To tackle this problem, we propose a fusion model called Emotion-based False Information Detection (Emo-FID). The model develops using Sentiment/Emotion, Semantic, and Structural (3S) frameworks of information propagation in a social network context. The proposed model performed both false information detection tasks as well as systematically predicted public emotions by understanding the propagation pattern contained in a social network context. The ultimate goal of the proposed framework is to classify public emotions and detect rumour/non-rumour conversations by adopting the Bi-GCN, GRU and BERT methods. To understand the effectiveness of the proposed approach, we conduct extensive experiments on publicly available real-world Twitter (currently known as X) datasets covering different events. The experimental results indicate that our fusion model performs well and achieves state-of-the-art accuracy on existing datasets.