Emotion-enhanced Cross-modal Contrastive Learning for Fake News Detection

This paper proposes a fake news detection model based on emotion-enhanced cross-modal contrastive learning (ECC-FND), which aims to fully utilize the information from images, text, and user comments in social media news to improve the accuracy and efficiency of fake news detection. ECC-FND combines cross-modal contrastive learning and sentiment analysis methods to effectively explore the semantic associations between multimodal information and dynamically weights features from different modalities using an attention mechanism. Experimental results demonstrate that ECC-FND significantly outperforms existing state-of-the-art methods on the Weibo dataset, which validates its effectiveness and generalization ability in fake news detection tasks. This study provides new insights for multimodal fake news detection and offers valuable references for future research.