SBRidAPI – DETECT DECIMATE AND PREVENT SOCIALBOT IN OSN USING DEEP LEARNING TECHNIQUES

Abstract

WEB 2.0-BASED Online Social Networks (OSNs) are one of the high impacting human innovations of the 21st century that facilitate their users to express views and thoughts on current affairs and personal life, connect with friends and celebrities, and get updated with the breaking news. OSNs facilitate their users in terms of connectivity, information sharing, knowledge acquisition, and entertainment, but these are not without any repercussions. The real-time message broadcasting and anonymity have exposed. The malicious and anti-social elements generally perform such activities using fake profiles in the form of bots, human-assisted cyborgs, Sybil, and compromised accounts. Recently, OSN platforms have witnessed emerging threats, having serious repercussions that are much more sophisticated in comparison to the classical cyber threats like spamming, DDoS attack, and identity theft. Among OSN-specific threats, automated profiles (aka social bots) are one of the major enablers of advanced illicit activities like political astroturfing. Social bots are very deceptive; they mimic human behavior to gain trust in an OSN and then exploit it for illicit activities. As a result, researchers are analyzing different malicious aspects of social bots. This project presents SBRidAPI, to profile users for detecting social bots on OSNs. To the best of our knowledge, this is the first deep learning-based approach that jointly models a comprehensive set of profile, temporal, activity, and content information for user behavior representation, which is fed to a two-layer stacked BiLSTM, whereas content information is fed to a deep CNN.