RAMI PUZIS

In this paper, we propose an approach for estimating the authenticity of online discussions based on the similarity of online social media (OSM) accounts participating in the online discussion to known abusers and legitimate accounts. Our method uses similarity functions for the analysis and classification of OSM accounts. The proposed methods are demonstrated using Twitter data collected for this study and a previously published Arabic Honeypot dataset. The data collected during this study includes manually labeled accounts and a ground truth collection of abusers from crowdturfing platforms. Demonstration of the discussion topic’s authenticity, derived from account similarity functions, shows that the suggested approach is effective for discriminating between topics that were strongly promoted by abusers and topics that attracted authentic public interest.