Real Time Profile Analysis and Fake Detection Model for Improved Profile Security in Online Social Networks

Abstract

The problem of social network security has been well studied. Numbers of approaches are identified towards securing the profile of social network users and identifying the fake profiles. Still the methods suffer to produce efficient results on detecting fake profiles. To improve the performance, an efficient profile analysis and fake detection model is presented in this article. The proposed Real Time Profile Analysis and Fake Detection Model (RTPAFDM) works on three phases: first the Post Level Trust Analysis (PLTA) is performed, which monitors the post updated on every day and share the posts internally with other users to claim the trustworthy of the post which computes the value of User Post Trust Weight (UPTW); second the method performs Profile Level Trust Analysis (PrLTA) which extracts the profile information like user name, personal details, contact details and friends list. According to the information extracted, the method searches on the entire social network to measure the Profile Match Similarity (PSM). Based on the PSM value, the method would identify the originality of the profile; finally, the method performs Profile Historic Trust Analysis (PHTA) which works on the earlier posts made and their truth values. It computes the value of Historical Trusted Post Weight (HTPW). Using the result of all these trust analysis values, the profile security and fake detection is performed. The method is evaluated with Twitter data set and achieved fake detection accuracy up to 97% and supports the improvement profile security efficiently.