Social Media Fake News Detection Integrating Multi-model Content Semantic Consistency
Zhang Guobiao,Li Jie
(School of Information Management, Wuhan University, Wuhan 430072, China)
(Institute for Information Retrieval and Knowledge Mining, Wuhan University, Wuhan 430072, China)
(School of Sociology, Soochow University, Suzhou 215000, China)
[Objective]In order to achieve early detection of fake news on social media and curb the widespread dissemination of false information. [Methods]Based on the news image and text content, by mapping the image to semantic tags, a method for calculating the semantic consistency between image and text content is designed, and a fake news detection model is constructed, and the fake news detection standard dataset FakeNewsNet is used to verity the performance of the model. [Results]The F1 value of the full-feature model that combines the semantic consistency features of news images and text is up to 0.775 on PolitiFact data and 0.879 on GossipCop data. [Limitations]Due to the limitations of the existing image semantic annotation models, the image content cannot be accurately described, and the calculated semantic consistency has bias. [Conclusions]Multi-modal feature fusion can effectively improve the performance of fake news detection, and the constructed news text and image content semantic consistency feature can enrich and expand the basis for fake news detection.