Research On Privacy Preserving Social Network Based On The Use Of Data Publishing

In order to meet the needs of scientific development or commercial purposes, the social network data owners have to release a large number of social network data.However, since these data contain a large amount of personal information, it will cause personal privacy information exposure if these social network data is published without consideration and modification. Therefore, the social network data privacy technology has become one of the hotspots in the field of privacy protection. However,almost privacy protection technologies assume that the published data have the same publishing purpose, which is obviously unrealistic because of research diversity. For instance, some of the data is carried out for community center use, and some data is collected for link mining and so on. General privacy preserving algorithms did not considering about the application of published data afterwards, leading to the poor data availability in future scientific research or commercial use. Therefore, this thesis looks into the social network data privacy protection technology for specified data release purposes.This thesis firstly introduces the field background of both domestic and foreign researches of the social network data publishing privacy anonymity technology, and summarizes the related theories and concepts. Secondly, a research of privacy anonymity algorithm for social network data publishing, which is for community structure analysis purpose, has been done based on the social network data privacy anonymity technology. Based on the classical k-degree algorithm, a anonymity one has been proposed combined the evolutionary algorithm and community hierarchy model. This new algorithm uses the evolutionary method to get the k anonymous sequence, and regard social network graph as a topological one with hierarchical community structure. Under the guidance of the k degree anonymous sequence, the algorithm is making social network graph anonymous by changing the entropy of the hierarchical community. Then, cause the single k-degree anonymous algorithm is stillin high risk of leakage of privacy,(k, l)-anonymous algorithm has been proposed,which is based on evolutionary algorithm and community hierarchy model, and combined the l-diversity algorithm with the above k-degree anonymity algorithm..The algorithm generalizes the sensitive attribute values of the nodes in the anonymous graph, which improves the security of the privacy preserving algorithm. Finally, it validated the proposed algorithm, and compared the results of other algorithms using different experimental schemes. Experimental result shows(k,l)- anonymity algorithm,which is based on evolutionary algorithm and community hierarchy model, can make maintain the structural properties of a graph and the community structure better than common privacy protection algorithms, and the availability of anonymous data is higher.