Analysis and detection of spam accounts in social networks
Chen Liu, Genying Wang · 2016
In recent years, social networks like Sina Weibo and Twitter have had rapid development. Meanwhile, social network platforms face threats imposed by spam accounts that propagate advertisements, phishing sites, fraud, etc. Such spam activities negatively affect normal users' experience and adverse to subsequent processing of users data. In this work, we present a new method using extreme learning machine (ELM), a supervised machine, for detecting spam accounts through their behavioral characteristics. Our analysis first collects messages crawling from Sina Weibo. Then, we select three categories of features extracted from message contents, social interactions and user profile properties applied to the ELM-based spam accounts detection algorithm. Finally, we verify the detectability of spam accounts through experiment and evaluation. Our proposed solution could achieve better results in system function and faster compared with other existing supervised machine leaning methods.