Graph Based Digital Spammer Identification by User Click Behavior

Abstract

As social media network popularity increases day by day, content writer are spamming on these platform for potential benefits of some organizations. Some of blogs are fabricate which invite users to external phishing sites or malware downloads huge security issue online and undermined the user experience. This work has proposed an un-supervised technique for identifying the real users from the social network spammers. Here graph based clustering algorithm was proposed to develop a binary cluster on the basis of serial or sequential action perform by the user. As per series of action social network graph is reduce into spanning tree where highly distance nodes are identified as abnormal behavior. So group of highly distance nodes are consider as the social spammers while other are real users. Experiment was perform on real social sequential dataset of twitter. Results were compared on various evaluation parameters and it was obtained that proposed approach has improved all such parameter values as compared to previous approach adopt by researcher.