It is obvious to say that we should fight against fake news. Methods have been developed. Most of them are time consuming since they use large amounts of annotated data, and are mainly supervised.

To address this problem, some researchers proposed a novel unsupervised fake news detection framework based on structural contrastive learning by combining the propagation structure of news and contrastive learning to achieve unsupervised training.

Remember that Contrastive learning is a self-supervised learning approach in which Data provides supervision by comparing samples and inferring the underlying data structure.

For more information: https://cybersecurity.springeropen.com/articles/10.1186/s42400-024-00342-5

Authors: Yajie Guo, Shujuan Ji, Xianwen Fang, Dickson K. W. Chiu , Ning Cao & Hofung Leung