Research On Social Network Privacy Preserving Based On K-isomorphism Algorithm

With the rapid development of Web2.0and popularization of the3G networktechnology, the increase in the number of mobile Internet users, social network websitesusher the rapid development of the optimal period, such as friend’s networks, SNS,micro-blogs. There are many users who registered in these social network sites, as theusers of autonomy and interaction, so the user will naturally be the creator of theinformation, or share the communicators. The number of social network’s data becomesvery large. Due to the scientific research and application of the need, social networkdata will be released by the data owner publicly. In addition, because mangy socialnetwork sites created, these create a very valuable market environment. Many socialnetworks based on statistical analysis of the research are developed and made wideapplication in the business sectors, the emergence of new social network in data miningtechnology generates economic value. But if these data can’t be reasonably used by agood manner, the user will pose a threat to privacy and information security.The traditional privacy protection technology can not directly be applied in higherdimensions of the social network data, and itself is complex, also lack adequate privacyprotection method that causes the user privacy information leakage. While thepublishers must ensure that the important information of personal data isn’t disclosed,so publishers need to treat the data issued by the anonymous processing. By using ofgraph theory in the data structure, a improved optimization algorithm based on k-isomorphic to the social network privacy protection methods is proposed, the originalgraph data is effectively divided into k sub-graph. While reducing the cost of anonymity,increased with the number of edges and removed are approximately equal to ensure therelease of graph data is k-isomorphic, which effectively prevents an attacker based onbackground knowledge of structured attack. In dealing with the dynamic social networkdata released in the privacy of users of multiple information disclosure issues, combinedwith the attacker based on background knowledge of the structure of the attacks, thedynamic social network privacy protection method is used. In each release algorithm ofk-isomorphic effectively divides the original graph into k-isomorphic sub-graphs, thengeneralize of the node ID, and when node added or removed to prevent an attacker withmultiple correlation between the release of private information identifying the user.Algorithm is divided into two main functional parts, the first structure to be k- isomorphic with social networks, and gets k-isomorphic sub-graph which is isomorphicto each other; the dynamic social network is dealt with k-isomorphic in the second part,and the node ID is generalized. For non-dynamic social networks, k-isomorphic isdirectly used to treat social network; the dynamic social network is firstly dealt with k-isomorphism, then the node ID is generalized, and anonymous social network ispublished. Finally, the improved optimization algorithm performance is tested, realdataset is used in experiment. The results show that the k-isomorphism algorithm ismore efficient than the original algorithm, and also significantly reduce information lossand improve the quality of anonymity, to protect user’s private information.