点击次数：

论文类型：会议论文

发表时间：2015-01-01

收录刊物：CPCI-S

卷号：4

页面范围：9-12

关键字：Anti-Trustrank; Social Spam; Relationship Strength

摘要：How to find the features of spammers and link structure is a fundamental issue in social networks. Existing work solves this problem by classification and ranking. However, these methods ignore relationship strength on edges, i.e., the weight between a pair of users. Then the challenge is how to effectively estimate the relationship strength given a pair of users. To tackle this challenge, the paper proposes a Poisson regression-based latent variable model to estimate relationship strength by jointly modelling users' similarities and interactions' frequency. Furthermore, Anti-Trust Rank with relationship strength algorithm is proposed to fight link farming on Twitter. Experimental results show that the proposed scheme can demote spammers and penalize uses that link to spammers effectively.