A Privacy-Preserving Scheme for Interactive Messaging over Online Social Networks.

Ying Cai,Shunan Zhang,Hongke Xia,Yanfang Fan,Haochen Zhang

DOI: https://doi.org/10.1109/jiot.2020.2986341

IF: 10.6

2020-01-01

IEEE Internet of Things Journal

Abstract:Due to the tremendous growth in user population over online social networks (OSNs), interactive messaging traffic has experienced exponential growth, which consequently causes serious privacy concerns. Malicious users may utilize camouflage, phishing, and other unconventional social engineering techniques to lure ordinary users to reveal their sensitive attributes in their interactive messages when they least expect. In addition, during a communication session, attackers may intercept data packets with sensitive attributes to obtain private information. Therefore, it is of paramount importance for the system to successfully identify malicious users and prevent them from obtaining private information from messages containing sensitive attributes when socializing over OSNs. In this article, we propose a privacy-preserving scheme for interactive messaging by leveraging user credibility that characterizes users' reputation based on users' social behaviors and uses the item response theory in psychometrics to assess the risk levels of users' interactive messages. This scheme ensures the privacy preservation during message exchange through message obfuscation with the message replication and the replacement of sensitive attributes. Through the experiments on a Facebook-like Social Network Data Set, we have demonstrated that our proposed scheme has high availability and accuracy while achieving the high success rate to identify malicious users, which can thus greatly reduce the privacy leakage.