Privacy Preserving Social Network Against Dopv Attacks

Abstract

The published multi-social network graphs contain numerous private information. To protect these information, researchers try to simulate attack models and design protection schemes. In this paper, we propose a heuristic attack model based on Dopv (Degree of paired vertices) attack. The attacker by defrauding trust or browse homepage to acquires the victim’s degrees (number of friends) from two published social network graphs and combine them into Dopv. Based on Dopv attack, attacker locates target candidates then compare nodes similarity by the same attributes or labels to find out target. To avoid this attack and protect the individuals’ privacy, we propose a new solution called Pvk-degree anonymity (Paired vertices k-degree anonymous). In Pvk-degree anonymity, the probability of a real user being re-identified is no more than 1/k. We devise algorithms to achieve the Pvk-degree anonymity that preserves the original vertex set in the sense that we allow the edge modified but no deletion of vertices. The experimental results show that our approach can preserve the privacy and guarantee the utility of social network graphs effectively against Dopv attacks.