Research On Privacy Protection Methods In Social Networks Based On Uncertain Graphs

With the rapid development of the Internet,social software has been widely used in people's lives which has changed people's connection,interaction and information sharing ways.Social software also provide convenient services to people's daily life.The circle of interest or circle of friends formed by social software can be called social network,which reflects the relationship between people in real-life.At the same time,when people use social software to make friends and interact with friends,social software will record the user's personal information and produce large amounts of behavioral data.However,everything has two sides,when people are enjoying the convenience of social network,people's sensitive information were exposed.In recentyears,there have been all kinds of privacy leaks in people's lives,such as unlawfully defrauded by impersonating friends on QQ.Therefore,how to protect user's privacy information in social networks is a challenging problem in big data's environment.The privacy protection of social network data mainly about how to protect individuals and the relationship between them.So researchers propose different privacy preserving methods for different objects and scenarios.The traditional privacy protection method based on encryption can effectively protect personal sensitive information.But for social network data,these methods are not conducive to the value of the data itself and will consume a lot of computing resources.Aiming at the relationship between individuals in social networks,this paper uses uncertain graph method and differential privacy method of privacy preserving of social network,a new privacy preserving method of uncertain graph is proposed to protect the association relationship between individuals.Meanwhile,due to the problem of different algorithms have different privacy measure,this paper proposes a general measurement for algorithm's privacy which provides a quantitative method for different algorithm's privacy.Finally,in algorithm's data utility,we define some metrics to measure the data utility.The main research work of this paper is as follows:(1)This paper analyzes the privacy leakage of social network,summarizes the process of privacy protection about social network data,and also summarizes the privacy protection methods of social network data.At the same time,four privacy protection algorithms for uncertain graphs are compared in detail from several aspects.(2)Some general measure metrics of algorithm privacy and data utility are proposed.It is mainly divided into the following two aspects:? Different algorithms have different privacy metrics.In order to measure the privacy of different algorithms uniformly,a general privacy measure metric—edge entropy is proposed.The greater edge entropy,the better privacy protection effect.? In order to measure the data utility of the algorithm,the measurement metrics of the data utility NE,AD,DV,Ii and data feasibility metric Utility are defined.(3)A privacy preserving algorithm UGDP(Uncertain Graph based on Differential Privacy)is proposed which based on differential privacy.In order to protect the relationship between individuals in social networks,UGDP algorithm combined the characteristics of differential privacy and uncertain graph to realized the protection of friendly relationships between individuals.The experimental results show that in privacy,the UGDP algorithm satisfied edge-differential privacy,and a theorem and theorem proof are given.Meanwhile,it is concluded that the privacy preserving of the algorithm is better than(k??)-obfuscation algorithm by the calculation of edge entropy.In terms of data utility,the nodes importance between UGDP algorithm and(k??)-obfuscation algorithm is similar in social network,but the data utility and the feasibility of the data is lower.(4)A privacy preserving algorithm of uncertain graph is proposed based on differential privacy to enhance the utility of data.By the technique of noise interception,we optimize the UGDP algorithm and propose a UGDP-mod algorithm to enhance the utility of data.The experimental results show that in privacy,the algorithm also satisfied edge-differential privacy but the privacy of UGDP-mod algorithm is lower than UGDP and higher than(k??)-obfuscation algorithm.Meanwhile,in the aspect of data utility,the data utility of the algorithm is between(k??)-obfuscation algorithm and the UGDP algorithm,and the data is more feasible.Compared with the UGDP algorithm,this result accords with the relationship between the utility and privacy of the algorithm.And the UGDP-mod algorithm is the UGDP algorithm after enhancing the utility of data.