Automatic Fake Followers Detection in Chinese Micro-blogging System

Yi Shen, Jianjun Yu, Kejun Dong, Kai Nan

Abstract: Micro-blogging, which has greatly influenced people’s life, is experiencing fantastic success in the worldwide. However, during its rapid development, it has encountered the problem of content pollution. Various pollution in the micro-blogging platforms has hurt the credibility of micro-blogging and caused significantly negative effect. In this paper, we mainly focus on detecting fake followers which may lead to a problematic situation on social media networks. By extracting major features of fake followers in Sina Weibo, we propose a binary classifier to distinguish fake followers from the legitimate users. The experiments show that all the proposed features are important and our method greatly outperforms to detect fake followers. We also present an elaborate analysis on the phenomenon of fake followers, infer the supported algorithms and principles behind them, and finally provide several suggestions for micro-blogging systems and ordinary users to deal with the fake followers.