A privacy-preserving interactive messaging scheme based on users credibility over online social networks

Abstract

Due to the tremendous growth in user population over online social networks (OSNs), interactive messaging traffic has experienced exponential increase, which consequently elevates the privacy leakage. Malicious users may utilize camouflage, phishing and other unconventional techniques to lure ordinary users to reveal their sensitive attributes in their interactive messages, when they least expect that the other sides are malicious. Therefore, it is very important to identify malicious users quickly and successfully when socializing over OSNs. In this paper, we propose a privacy-preserving interactive messaging scheme by relying on user credibility, which characterizes users' reputation through users' social behaviors, and use the item response theory in psychometrics to assess the risk levels of users' interactive messages. We have conducted extensive studies by sending interactive messages with sensitive attributes and have observed that the success rate of identifying malicious users based on user credibility is significantly higher and our scheme can effectively lower the risk of privacy leakage in our interactive messaging system.