Two Phase Based Spammer Detection in Weibo

Recent years, with the growth of social network users, the social network have become a more and more important platform for sharing information. The growth of social network users inevitably leads to the appearance of spammers, who usually spread spam information, post reduplicate messages to mess up legitimate users´ reading timeline or affect their operation feeling on social networks. As one of the largest online social networks in China, although Weibo system has already adopted certain kinds of spammer detection approaches, there are still many spammers existing in Weibo. Hence, in this paper, we propose a two phase based spammer detection approach. In this approach, firstly we take the existing works about the user feature as the first phase. In addition, we introduce content mining as the second phase in spammer detection. Technically, LDA is introduced to analyze the topics of a user in Weibo. More specifically, if his/her topics are usually spammed, we can claim that the user is a spammer. Experimental results show that our proposed approach is better than the existing approaches on spammer detection.