Research On Anonymity Techniques For Privacy-preserving Data Publishing In Social Networks

A social network describes a society of individuals, groups, and associations betweenthem. In the context of the extensive applications of Web2.0and social network servicesoftware, many users have their own e-mails, microblogs, online transaction records,community spaces, etc. these individual properties including relations among the users form avariety of social networks, Many of them have evolved as a complex network in which datahas the characteristics of high dimension, non-linear, scale-free, and small world. In order tomeet the needs of scientific research and data sharing, a great quantity of social network datahave been collected and released. However, the data publishing processes, which aim toinformation sharing, data mining, and knowledge discovery in database (KDD) and so on, areoften accompanied with the disclosure risk of sensitive privacy information. Meanwhile, theresearch of privacy preserving in the data publishing (PPDP) is presented for the reasonmentioned above, whose major target is to improve the released data security by losing someinformation in raw data appropriately under the premise of the data utility guarantee, and thento provide a very good trade-off between achieved privacy preserving and data utility.Nowadays, the research on preserving privacy in social networks is mainly focusing onrelational databases, and the research on graph datasets has just begun. this paper makes deepand detailed study on the anonymity techniques in social networks for privacy preserving datapublishing in the case of ensuring the strong information utility.Firstly, according to the problem of the privacy disclosures is brought about throughstructural attacks by malicious users. In this paper, we consider the privacy preservingpublication against structural attacks in social networks, which are expressed by simpleundirected graphs. Firstly, we present a k~+-isomorphism anonymity graph model based on thetheoretical principle of k-isomorphism, and then a k~+-isomorphism anonymity problem facedthe simple undirected graphs is formally defined. In additional, we propose a global structuralpartition-based k~+-isomorphism algorithm (GSPB k~+-isomorphism) to solve the problemdefined above. The experimental results show that under the equal conditions, our method notonly produces similar information loss to that of the traditional k-isomorphism method, butalso needs nearly time cost, which achieves more effective privacy preservation. Secondly, according to the problem of the privacy disclosures which is brought aboutthrough neighborhood attacks by malicious users. In this paper, we consider the privacypreserving publication against neighborhood attacks in social networks, which are expressedby simple undirected graphs. Firstly, we present a (d, k)-anonymity graph model based on thetheoretical principle of k-anonymity, combining the d-neighborhood attack in social networks,and then a (d, k)-anonymity problem faced the simple undirected graphs is formally defined. Inadditional, we propose a high degree vertex first (d, k)-anonymity algorithm (HDVF (d,k)-anonymity) to solve the problem defined above. The experimental results show that underthe equal conditions, our method not only produces less information loss than that of theexisted method, but also needs nearly time cost, which effectively resists d-neighborhoodattacks and preserves privacy information of the publishing networks.Thirdly, according to the problem of the privacy disclosures that is brought about throughstructural attacks by malicious users. In this paper, we consider the privacy preservingpublication against sensitive edges identification attacks in social networks, which areexpressed by bipartite graphs. Firstly, we present a positive one-way (c1, c2)-security algorithm,a negative one-way (c1, c2)-security algorithm, and a two-way (c1, c2)-security algorithm. Theproposed algorithms are based on the k-security group theory and the sensitive edgesidentification attacks in social networks. Furthermore, a bipartite graph anonymous problem isformally defined to against sensitive edges identification attacks. Meanwhile, aclustering-based bipartite graph anonymous method, Clustering-based bipartite (c1,c2)-security algorithm (CBB(c1, c2)-security), is proposed. The experimental results show thatunder the equal conditions, our method not only produces less information loss than that of theexisting method, but also effectively resists sensitive edges identification attacks and realizesecurity release of bipartite graphs.Finally, according to the privacy disclosures problem which is brought about throughcomposite attacks (structural attacks and attributes attacks) by malicious users, in this paper,The social networks are considered which are expressed by simple undirected graphs and withrich attributes. A (k, l)-anonymity graph model based on the theoretical principle ofk-isomorphism and the l-diversity is presented, and then a (k, l)-anonymity problem faced thesimple undirected graphs is formally defined. In addition, we propose a generalizationalgorithm based on k-anonymity and l-diversity to solve the problem defined above. In the meanwhile, the experimental results show that under the equal conditions, our method not onlyproduces less information loss than that of the departed method, but also needs nearly timecost, which effectively resists composite attacks and preserves privacy information of thepublished networks.