Spammer,as a disseminator of spam,has become the focus of Weibo's anti-spam research.Existing research on spammer detection is confined to traditional binary classification problem,which is simply to determine the user for spammer and non-spammer.However,there are many types of spammers in the Weibo platform,if all kinds of spammers are considered as the same category,there will be the problem that spammers'characteristics can affect each other,so that the overall detection performance decreases.To solve this problem,the behavior of many kinds of spammers is analyzed in this thesis.First of all,according to spammers'behavior purposes and behavior patterns,spammers are classified into four categories.Secondly,the data sets are obtained by the crawler program,and a set of samples for analyzing the characteristics are constructed and labeled,then the statistical characteristics of users are calculated.Finally,the characteristics of the four types of spammers are analyzed quantitatively,and the characteristics of each type of users are summarized.The experimental results show that there are highly distinguishable features between various types of spammers and non-spammer,which can effectively distinguish various types of spammers and non-spammer and improve the detection accuracy.
维普
万方数据
