FND-MC: Fake News Detection Based on Crossmodal Alignment and Multimodal Fusion

Cybersecurity faces challenges posed by the spread of fake news in multimedia formats on social media platforms. Although there are various methods for verifying information on social media, the application of multimodal alignment and fusion techniques in fake news detection is still quite limited. Here, we construct a fake news detection method based on multimodal alignment and fusion (FND-MC). This method first employs advanced multimodal learning models to extract textual and visual features from news articles simultaneously. Secondly, it integrates the extracted key features through a text-visual feature alignment module and a multi-head attention mechanism, thereby incorporating the global information of the news. Finally, we conducted extensive experiments on two public datasets, Recovery and GossipCop, to verify the advantages of the proposed framework.. The experimental results indicate that FND-MC can detect multimodal fake news effectively.