Existing methods for the rumor detection(RD)considered either only temporal flow or propagation struc-ture in social media,and used numerous auxiliary information.Actually,both the temporal flow and propagation structure are helpful to RD and can complement each other.Moreover,the self-description of source user is more important than other auxiliary information and semantic information in source post.To address the above issues,this paper proposes a novel RD model named TPSS,which integrates the temporal flow based method and the prop-agation structure based method.Moreover,it only employs the self-description of source user as auxiliary feature and proposes a co-attention mechanism to enhance temporal flow and propagation structure based source post.The experimental results on Twitter15,Twitter16 and PHEME show that TPSS outperforms several state-of-the-art baselines.
维普
万方数据