Research And Application Of Personalized Privacy Protection Technology Based On Diversity Clustering

In the era of big data,more and more users establish connections through the Internet,forming a social network with large data and complex structure,which provides powerful information resources for data analysis and data mining research,but also provides convenience for network attackers.Privacy protection technology can prevent network attackers from re-identifying attacks on target nodes and prevent privacy information from being stolen.Since different users have different privacy protection needs and preferences,it is necessary to define their privacy protection needs,limit the degree of protection and use of privacy information,and provide personalized privacy protection.This paper studies personalized privacy protection technology from two aspects of social network graph structure and specific data records.According to the Connection Fingerprint(CFP)of the graph structure,this paper calculates the privacy protection requirement of the node,which is used to identify the different privacy protection preferences of the nodes.At the same time,combined with the diversity clustering(l-diversity),this paper studies personalized privacy protection technology for graph structure information.There are many kinds of sensitive attributes in the nodes of social network.This paper calculates the need of personalized privacy protection for nodes and data records by inheritance classification tree corresponding to each attribute,and studies the personalized privacy protection technology for data records of graph nodes.In this paper,the design idea and implementation process of the algorithms are analyzed in detail,and a comparative experiment is carried out using standard data sets.At the same time,it is applied to a medical device supply chain system to verify the effectiveness and availability of the technology.Specifically,the main work of this paper is as follows.(1)For the graph structure information in social networks,this paper studies a personalized privacy protection method based on l-diversity.The method sets the personalized privacy protection requirement according to CFP of private nodes in the social network,and clusters them based on the l-diversity model.At the same time,a small number of virtual structures are added according to the individual privacy protection needs of each equivalent class,to form a publishable social network graph.(2)For the data records formed by nodes in social network graph,this paper studies a privacy protection method based on l-diversity for multi-dimensional sensitive attributes.The method sets the personalized privacy protection requirement for each data record according to the inheritance classification tree of each sensitive attribute,and then constructs a multi-dimensional sensitive attribute record set for l-diversity clustering,finally obtains a publishable data set.(3)In view of the two privacy protection methods proposed in this paper,the corresponding algorithms are implemented respectively.At the same time,a number of comparative experiments are designed from the aspects of time efficiency,data missing rate and change rate.The results show that the two methods can achieve privacy protection while ensuring the validity of post-publication social network data.Finally,the two methods are applied to a medical device supply chain system to protect the personalized privacy of the medical device supply chain network formed by the system,which not only improves the user experience,but also ensures the data security of the supply chain network.