Social Networks Spam Detection Using Graph-Based Features Analysis and Sequence of Interactions between Users

The growing interest in the different social network platforms leads to huge number of interactions between different users all around the globe. These unlimited interactions provide a suitable environment for spammers to spread as the complexity of the social networks increases. Automatic detection of such malicious users inside this crowd of complex interactions is one of the most difficult research problems. Different approaches have been adopted to contend against malicious activities. Among the different promising approaches is the one relying on using graph analysis techniques. In this paper, we suggest two representation models for social interaction's graph-based datasets. The representation models are mainly developed based on analyzing interactions and relations between users. The first model is developed based on graph-based analysis, while the other one is developed based on sequential processing of user interactions. Based on the conducted experiments, we conclude that the two representation models show high spam detection accuracy. However, graph-based analysis models produce higher accuracy levels compared to those produced by interaction sequences processing models.