Hierarchical Privacy Protection Algorithm For Social Network Based On Community Division

With the development of network and communication technology,social network is increasingly becoming an important platform for people to share information and expand social relations.The data produced and disseminated in social networks reflect real social relationships and include a lot of personal information.These data have been widely used in the community structure analysis,group behavior analysis,public opinion monitoring and many other fields which have high scientific and commercial value.Because of the need of data sharing and academic research,a large number of social networks data were collected and published.The privacy disclosure problem should not be underestimated.How to balance the privacy of information and the availability of data has become a difficult problem to be solved in social network data publishing.The existing social network privacy protection algorithm usually adopts a unified generalization standard which does not consider the differential requirement for node privacy protection.The homogenized privacy protection method causes the lack of protection for the core nodes or the over protection for the noncore nodes.Therefore,this paper proposes a hierarchical privacy protection algorithm for social network based on community division,which realizes hierarchical privacy protection for nodes in different structural characteristics in social network.The main work is as following:Aiming at the problem of stability in social network community division,this paper proposes a propagation algorithm that integrates the influence factors of nodes.It comprehensively estimates the influence of nodes and updates labels in order of influence from high to low,so as to improve the quality of community division.The experiments on Pokec data sets showed that comparing with LPA algorithm,Q was enhanced by 6.42%.Comparing with the LPAm algorithm,Q was enhanced by 4.28% and the quality of community division was better.In the view of equalization privacy protection in social networks,a hierarchical privacy protection algorithm based on node influence and community division was proposed in this paper.The different influence nodes were generalized with different intensity anonymous algorithms to realize the need of personalized privacy protection.This paper proposed a new(?)anonymous algorithm which considers the distribution of sensitive attribute to satisfy the higher privacy protection level of the local core nodes.The experiments on Pokec data sets showed that comparing with k-degree-l-diversity algorithm,the loss rate of information was reduced by 3.54%,the number change of boundary was reduced by 6.32%,the average path length was increased by 5.58% and the quality of the published data was higher.Simulation results and theoretical analysis showed that the cost of degree anonymous was smaller than traditional anonymous algorithm.The local core nodes have been protected adequately and the data loss rate was reduced.The data availability was improved and the structure property of the graph was preserved.