DeSAN: De-anonymization against Background Knowledge in Social Networks

Abstract

Social network de-anonymization is a challenging research problem. Gigantic volumes of social network data get collected by third-party applications to mine knowledge for devising government policies, business decisions, health records, and many more. Social network data is vulnerable to privacy leakage due to the presence of sensitive information. Furthermore, attackers knowledge and their manipulation capabilities have also expanded in multi-folds. As a result, modelling the attacker’s knowledge helps design a practical privacy model that could overcome attackers capabilities. Semantic knowledge has the potential to disclose privacy where the information is imprecise and inaccurate. This paper proposes a deanonymization technique, DeSAN, against imprecise and inaccurate attacker knowledge. The proposed technique assumes the attacker’s knowledge, comprehensive and realistic. We have implemented the proposed DeSAN technique on a real social dataset, which exhibits encouraging result in terms of deanonymization accuracy.