A Rumor Detection Method Based on Multimodal Information Fusion

With the development of social media, online rumors also spread rapidly through social media, causing serious negative impacts on politics, economy, and public safety. Therefore, how accurately detecting rumors on social media is a crucial issue. The existing rumor detection methods can be mainly divided into feature-based methods and propagation structure-based methods, but feature-based methods cannot capture the features of rumor propagation structure, and methods based on propagation structure cannot make good use of the text features of rumors. To this end, we consider combining the two methods and introduce an attention mechanism to learn the weights of individual features to fully fuse them. In addition, the aggregation structure of rumors is fully considered when extracting the propagation structure features. Experiments on real-world datasets demonstrate that our model achieves better results than the state-of-the-art methods, that is, the rumor detection method designed in this paper has better rumor recognition ability.