Research On Privacy Protection Method Based On K-Anonymity In The Social Network

With the development of Internet technology and the popularity of social media,the number of social users is increasing day by day.Users can share each other’s information,ideas,and interests,interact with other users,and gradually become a "social sensor",generating a large number of user-generated content(UGC).UGC is collected and collected by many social platforms,and users and their interactions are built into huge social networks.By studying these data,social platforms and related product providers can obtain useful information and provide better services for users.However,in this process,users’ privacy is often compromised.Therefore,before the release of social network data,it is necessary to conduct anonymous processing and ensure data availability on the premise of protecting user privacy.Aiming at the problems existing in the existing k-anonymous model,this thesis proposes an improved algorithm,and the specific work is as follows:1.Explain the background significance of social network privacy protection,summarize the importance of the social network model and social network privacy security,introduce common privacy attack means and social network privacy protection technology,and finally classify and summarize the existing social network privacy protection model.2.It mainly addresses the problem that the existing graph modification technology disturbs the structure of social networks too much,resulting in the low availability of anonymous data.The improved k degree anonymous social network privacy protection method(KDSNP)is proposed.This algorithm fuses nodes to form super-points according to the "attraction" of node inter degree,then divides the super-points to obtain candidate nodes,and then modifies the edges of candidate nodes according to the structural relationship between nodes to improve the availability of anonymous data.Tested on four real experimental data sets:Facebook,Feather-lastfm-social,Dolphins,and Polbooks.Average path length,average clustering coefficient,and transitivity are used to evaluate the performance of KDSNP algorithm.The experimental results show that KDSNP algorithm has little change to the original data in the experimental data set and improves the availability of anonymous data.3.Most of the existing privacy protection methods in social networks are based on clustering,which considers the relationship between the structure and attributes of nodes.Moreover,the differences in measurement between them usually lead to poor clustering effects and excessive data anonymity,leading to the decline of data availability.To solve these problems,a graph-clustering anonymity privacy protection algorithm with fused distance-attributes(GCA-DA)is proposed.This algorithm calculates the similarity of nodes in the figure according to the attributes and distances between nodes,merges similar nodes into multiple clusters,each of which contains at least k nodes,anonymizes the attributes of nodes contained in clusters,hides the node structure in clusters to ensure that user nodes are not identified,and only publicizes the statistical information of each cluster.This includes the number of nodes in the cluster,the number of edges,and the anonymous attribute.The performance of the algorithm is evaluated by means of clustering density,entropy,and information loss.The experimental results show that GCA-DA algorithm in improve the quality of clustering nodes,reducing the effectiveness of the node structure and attribute information loss.This thesis explores the privacy protection method of social networks based on the k-anonymity and designs a new method by using the existing theory and technology.The proposed method achieves a better clustering effect by improving and optimizing the node clustering conditions and uses the generalization method to protect user privacy and better balance the privacy and availability of anonymous data.