Research On Social Network Data Publishing Algorithm Based On Differential Privacy

With the rapid development of mobile Internet technology,more and more individuals have participated in social network activities,which results in a large amount of information,the study of which can promote the development of society.Meanwhile,such information also contains a lot of user privacy information,so how to protect the privacy of social network data that need to be released is particularly important.As a privacy protection model with strict mathematical theory proof,differential privacy can be used to solve the problem that traditional privacy protection cannot resist the attack of background knowledge,which has attracted the academic attention of many scholars.Therefore,it is of great significance and prospect to study the combination of differential privacy and social network privacy protection.The paper conducts the following work on this idea:In order to solve the problem that the ternary closure algorithm cannot resist the background knowledge attack,combined with the differential privacy technology,the paper proposes a differential privacy protection algorithm for uncertain graphs based on ternary closure.The algorithm first traverses the vertex pair of the original social network graph and adds edges to the vertex pair whose shortest path between vertices is 2(the weight of the default edge is 1)according to the ternary closure principle.The added edges form a triangle with the two edges of the shortest path and are recorded to the triangle set.After the traversal,the edges that do not belong to the triangle set in the social network graph are added to the determined edge set.Then the edges of the set of triangles are added with the probability of the existence of a Laplace distribution,and the existence probability of the edges in the set of determined edges is given as 1.Finally,a new social network graph is generated according to the probability of the existence of each edge in the social network graph.It had been theoretically proven that the proposed algorithm satisfies the differential privacy and solves the problem that the ternary closure algorithm cannot resist the background knowledge attack.In order to compare and analyze the privacy protection of the algorithm,the paper designs an experiment based on information entropy and the experiment results show that the privacy protection of the proposed algorithm is higher than that of the ternary closure algorithm.In order to solve the problem of excessive noise added in the differential privacy algorithm based on random projection,the paper proposes a differential privacy protection algorithm based on singular value decomposition(SVD).The algorithm first uses random projection to reduce the dimension of high-dimensional social network data,conducts singular value decomposition on the dimensionality reduction data,adds Gaussian noise to the singular value matrix,and finally generates the matrix to be released by inverse singular value decomposition.The singular value matrix used by the proposed algorithm is a matrix with values only on the main diagonal,and the number of values is the rank of the matrix.Compared with adding Gaussian noise directly to dimensionality reduction data,adding Gaussian noise to the singular value matrix can effectively reduce the amount of noise.It has been theoretically proven that the proposed algorithm satisfies differential privacy.The paper designs the Euclidean distance difference experiment and spectral clustering experiment to analyze the data availability of the algorithm,and the experiment results show that the data availability of the proposed algorithm is higher than that of the differential privacy algorithm based on singular value decomposition.The paper analyzes the possibility of privacy disclosure in the process of social network data release,and integrates the differential privacy algorithm into the process of social network data release so as to solve the problem that the ternary closure algorithm cannot resist the background knowledge attack and improve the data availability of the differential privacy algorithm based on random projection,which provides some solutions and ideas for privacy protection in the process of social network data release.