Detecting Spam on Sina Weibo

Online social network becomes greatly prevalent and evolves a communication channel for billions of users. Unfortunately, due to the ease of reaching these users, it has been penetrated by spammers who post inappropriate content. After revealing the transmission mechanism of the spam, an automatic detecting framework is designed to identify spam information. The profiles which have multiple discriminative features are extracted for the Machine Learning techniques. In the experiment phase we collected 562K messages posted by 28,679 users on Sina Weibo, then analyzed the different behaviors between malicious accounts and normal ones. W e evaluate our approach on a real large-scale dataset. The results demonstrate the effectiveness of the detecting system.