基于对比学习的社交媒体谣言检测

Social media is now one of the important channels for people to learn about information and news, but they also lead to the spread of rumors. Rumors have brought great harm to society, and how to accurately identify rumors is particularly important. This paper finds that the traditional Weibo rumor dataset is very old, and builds a new Weibo21 rumor dataset as a supplement; At the same time, a rumor detection model based on contrastive learning is proposed, which uses unsupervised contrastive learning to build a text encoder, and then uses graphattention network to perform feature extraction for classification. Finally, the experiments on the two datasets show that the model proposed in this paper is better than the previous models. ?????