Zombie Fans Detection in Sina Microblog a Machine Learning Approach

Abstract

Online Social network have become very popular in recent years. Among these sites, Sina Weibo is the top microblogging network in China. The popularity and open structure of Sina Weibo have attracted a large number of automated programs, known as Zombie Fans, which appear to be a threat to Weibo. Using the API methods provided by Sina Weibo, Around 22,914 Weibo account, 182,923 tweets, and around 190 M follower/friend relationship in total are collected from public available data on Weibo. To facilitate the Zombie Fans detection, 7 single features, 6 correlation features are extracted. A machine learning approach is proposed to distinguish the Zombie Fans from normal ones, we evaluated our detection scheme using various classifiers on these features. Our Zombie Fans detector can achieve 96.0% accuracy and 95.1% precision using the Support Vector Machines classifier.