Abnormal user identification in online social networks based on user behavior

Recently, in online social networks zombie fans have been shown a more complex and humanizing form. The existing methods based on basic features such as the number of followee, follower scale, user name and information content are not very efficient, which may lead to many misrecognitions and missed detection. Behavior pattern is the most fundamental feature of online users and abnormal users certainly have particular actions which are different from the normal users. Therefore, in this paper, a zombie fans identification method has been proposed based on the behavior characteristics like retweet, comment and the corresponding regularity. Furthermore, with user behaviors, invalid user identification is also researched. The experimental results showed that the abnormal user recognition method proposed in this paper had high identification accuracy.