Protecting The Identify Privacy In Social Networks

In the real world, publishers, e.g., Facebook, would publish social networks to a third party. By technical means, such as data mining, value principles hidden could be explored by researchers. Meanwhile, privacy involved in social networks will be under threats. Therefore, protecting identity privacy and guaranteeing the utility of social networks when publishing social network data become an important concern.The main research emphasis of this paper is to design a privacy protection model witch satisfied with security and utility. This paper mainly includes the following works:First of all, thorough analyzing the importance of degree in social network analysis field, original social network is modeled as a undirected graph. Besides, the components and process of system model are stated.Next, this paper identifies a novel type of privacy attack, termed neighborhood relation attack. It assumed that an attacker could acquire the knowledge of the degree sequence of target’s neighborhoods, in addition to the target’s degree and relationship between them. Through the experiment on real social network data set, it proves that the neighborhood relation attack this paper assumed is feasible.Then, this paper proposes a SSSA privacy protection model to generate an anonymized social network against neighborhood relation attack. This model could make the unique node similar in original social network, and any node has the same neighborhood relation graph with at least k-1 other nodes. Under the neighborhood relation attack, the probability of each node identified should be below 1/k to make sure the node is highly secured. Meanwhile, that seems to guarantee a better utility of the published social network.Finally, This paper takes Cond_Mat、Hep_Hp、Enron、Facebook real social network data sets as the object of study, and performs an experiment on security and availability. The empirical study indicates that the SSSA privacy protection model could not only protect identity privacy, but also make published social network be used to answer some graph structure characters.