Detecting Social Spammers in Sina Weibo Using Extreme Deep Factorization Machine.

Online social networks (OSNs) are a form of social media that allow users to obtain news and information as well as connect with others to share content. However, the emergence of social spammers disrupts the normal order of OSNs significantly. As one of the most popular Chinese OSNs in the world, Sina Weibo is seriously affected by social spammers. With the continuous evolution of social spammers in Sina Weibo, they are gradually indistinguishable from benign users. In this paper, we propose a novel approach for social spammer detection in Sina Weibo using extreme deep factorization machine (xDeepFM). Specifically, we extract thirty features from four categories, namely profile-based, interaction-based, content-based, and temporal-based features to distinguish between social spammers and benign users. Furthermore, we build a detection model based on xDeepFM to implement the effective detection of social spammers. The proposed approach is empirically validated on the real-world data collected from Sina Weibo. The experimental results show that this approach can detect social spammers in Sina Weibo more effectively than most of the existing approaches.