Reserch On (α,k)-Anonymity Method Based On Social Networks

With the rapid development of network technology, a large number of the online socialnetworks were established and more and more people are participate in social networks toshare and exchange information in this interactive process, it will produce large amounts ofdata. These data contain part of what the user does not want others to know, for example,somebody suffered from one disease, and these can usually be considered to be the personalprivacy information.A variety of social network analysis and data mining were used in orderto expose the hidden social model. A number of social network data was also published.People’s privacy have been violated.The recognized awareness increasingly of theimportance of privacy and privacy protection. Therefore, the privacy protection of thereleased data in social network became an concern problem.Because of the complexity structure of the social network, a lot of information can beconsidered to be the privacy which need to protect. Most of the studies can only protect onekind of privacy. Online social networks related to the individual privacy of network, and thusthe network data can not be directly released. For the usability of social network data, we cantake certain protective measures to ensure the published data is available, and in order not todisclose the individual’s privacy under attack. Anonymous by the removal of individualproperties directly, when the attacker had the structured background knowledge, also canidentified the individual in the network. Proposed the social networks (α, k)-Anonymitymethods, considered the relationship between the nodes, using the clustering-based method,the node the relationship between attributes and node protected. The number of nodes ineach cluster at least k, each sensitive property associated of nodes in the clusteringpercentage is not higher than α. By setting the parameter to control the relationship betweenthe different information loss.Theoretical analysis and experimental results show that, the social networks (α, k)-Anonymity methods can effectively prevent the disclosure of the relationship between thenode and identification, in the case of the smallest possible loss of information privacy.