Rumors can be propagated across online microblogs at a relatively low cost, but result in a series of major problems in our society. Traditional rumor detection approaches focus on exploring various propagation patterns or data interactions between a source microblog and its subsequent reactions. It is obvious that this causes missing interaction on rumor detection, especially in the absence of retweets or reactions. According to the communication theory of Allport and Postman (1947), Chorus (1953) and Rosnow (1988), the topic of a post can help determine its potential of being a rumor or not. Therefore, we develop a novel topic-driven rumor detection (TDRD) framework to determine whether a post is a rumor only according to its source microblog. Specifically, we first automatically perform topic classification on source microblogs, and then we successfully incorporate the predicted topic vector of the source microblogs into rumor detection. Our extensive experimental results demonstrate that our TDRD significantly outperforms state-of-the-art methods on both two English and two Chinese benchmark datasets.
- Latent dirichlet allocation (LDA)
- Rumor detection
- Source microblogs
- Topic vector