Social Network Data Release Based On Node Differential Privacy

With the rapid development of science and technology and the popularization of social network services,various social platforms can collect a large number of user personal data and information through the user side for data mining and analysis.Although mining and analyzing user data can make people's life more convenient,but it also brings the risk of privacy disclosure.How to protect the privacy of users while publishing user data has become an increasingly important issue.Differential privacy is a privacy protection model with strict theoretical verification and mathematical proof.In order to publish social network safely,differential privacy is gradually applied to the field of social network data release.At present,the application of differential privacy in the field of social network data publishing is mainly divided into two directions: one is the statistical characteristics of differential privacy social network publishing,the other is the distribution of differential privacy composite social network.This paper proposes the following two parts according to the application direction of differential privacy in the field of social network publishing.(1)This paper proposes a weighted histogram publishing method for node strength of social networks which satisfies the differential privacy of nodes.Firstly,in order to satisfy the difference privacy of nodes without losing the utility of data,this paper proposes a projection method to transform the original social network into bounded social network,which reduces the sensitivity by constraining the degree of nodes and the weight of the edges.Then,in order to select the appropriate threshold and grouping parameters,the quality function is designed and selected by the exponential mechanism.Then,this paper proposes a new method of node strength histogram publishing based on substitution and density,which improves the availability of data while publishing the strength distribution of social network nodes.Finally,the experiment on four data sets proves that the proposed method improves the accuracy of the histogram of node intensity distribution under the condition of satisfying the differential privacy of nodes.(2)This paper proposes a weighted social network publishing method based on node differential privacy.Firstly,in order to reduce the sensitivity and reduce the noise added,this paper proposes a projection method which is constrained by node degree and triangle number.Then,the node attributes are formed by obtaining the statistical characteristics of the original social network,which is used as the parameter of the composite weighted social network publishing method.Then,a new method of social network publishing is proposed,which combines the attributes and weights of nodes.The method synthesizes an initial social network by degrees in node attributes,and then adds or cuts edges to the initial social network according to the triangle number in the node attribute,so as to obtain the final social network.Finally,the weighted social network publishing method is verified on three data sets.The results show that the algorithm has some utility while satisfying the differential privacy conditions of nodes.