A Novel Rumor Detection Method Based On Labeled Cascade Propagation Tree

Nowadays Sina Weibo has become a fashionable social media platform in China. Meanwhile, the public and anonymous environment provides rumors a perfect hotbed to breed and spread. The negative influence on society from rumors cannot be ignored. Traditional rumor detection methods based on features always focus on static or flat features coming from content, users, propagation and etc. but it often ignores the effect of information's propagation structure. Aiming at this problem, first we introduce information's propagation cascade model into LPT (Labeled Propagation Tree) and propose an improved model-Labeled Cascade Propagation Tree (CA-LPT) which allows us to consider the effect of information's propagation structure. Second, we investigate users' influence assessment by a dynamic method. Finally, we predict whether a microblog post is a rumor by proposing 10 new features and combine with hybrid kernel SVM based on random walk graph kernel. Experiment results on real-world data from Sina Weibo demonstrate the efficacy of CA-LPT related new features. We conclude that CA-LPT can help building a more exact propagation tree and provide clues for a better classification accuracy.