Research On Multi-Social Network Data Publishing Privacy-Preserving Algorithm

Currently,social networking sites are indispensable in people's daily work and entertainment.Social networking sites such as Facebook,Twitter,and Weibo are widely used.The increasing number of users and the number of visitors make social network data more complex,and the privacy protection of data publishing becomes more and more important.The data of the social network is a graph structure established by the individual as the vertex and the friend relationship.After the data is published,there is a privacy leak problem caused by a malicious opponent carrying different background knowledge,and the leaked privacy includes the target being attacked.The vertices or edges,the sensitive attributes of the vertices,or the weight information of the edges.How to establish a privacy attack model and design a targeted solution to the possible privacy breaches and protect the private information in the data publish is the focus of research in the field of privacy protection of social network data publish.Aiming at the privacy leakage problem of re-recognition of vertex identity caused by attack with background knowledge of multiple social networks,this paper defines a heuristic multi-social network attack model-combination attack model,which proposes multiple social network data.The vertex value of the concentrated target can be combined as the background knowledge of the attacker.The attacker obtains the candidate set composed of the attack target from different social network data sets according to the combination degree,and matches the non-sensitive attribute information of the vertex between the sets to the attack target.The vertices are re-identified.In order to solve this privacy attack,this paper proposes a combination degree(d_x,d_y)-k anonymity algorithm,which maximizes the availability of raw data and the degree of composition data through binary combination degree clustering and single vertex clustering.The value is assimilated so that when a privacy attack is initiated on any binary combination in the group graph,no less than k candidate targets are obtained,and the purpose of protecting the attacked target vertex is achieved.The algorithm is evaluated by two sets of artificial data sets.The experimental results show that the privacy protection algorithm effectively prevents the combination attack and better protects the usability of the graph data.Aiming at the problem of sensitive label leakage caused by vertex re-identification attack in real social network data,this paper defines a combination graph-neighbor label matching attack model based on group graph,and combines the target degree with the neighborhood based on the heuristic combination attack model.The label is used as the background knowledge of the attacker to obtain the candidate vertices.The sensitive label matching result alone exposes the sensitive information of the attacked target.In order to solve this privacy attack,this paper proposes a group-sensitive label generalization L diversity algorithm,which reduces the probability of sensitive tags being recognized by designing a group-sensitive label generalization tree,and makes the attacker popularize the sensitive label L diversity algorithm.According to the background knowledge,the number of sensitive labels of the candidate vertices obtained by the background knowledge and the number of sensitive tags obtained by the matching are not less than L,and the purpose of protecting the sensitive information of the attacked target is achieved.The algorithm is evaluated by using three sets of data with different ratios.The results show that the privacy protection algorithm effectively prevents the sensitive label privacy attack composed by combination degree-domain label matching and the availability of better maintenance graph data.