Research On Privacy Protection Algorithm Based On K-anonymous Personalized Data

In recent years,with the popularity of the Internet and the rise of the mobile Internet,the commercial value of big data has been applied to all aspects of society,and it has had a profound impact on human society.It also makes the collection,analysis and mining of information data more convenient and accurate.However,in the process of researching the sharing,mining and knowledge discovery of data information,it is accompanied by the leakage of sensitive private information.How to protect personal sensitive data from being leaked is an urgent problem to be solved in the current society.That is,when data is distributed or shared,it needs to be considered from two aspects: a.Personal privacy in the data will not be leaked.b.In the process of knowledge discovery such as data mining,the anonymous data still has high practicability and usability.While avoiding data privacy leakage,ensuring the authenticity and efficiency of data is the main research direction of anonymous privacy protection.However,privacy protection has personalized demand,and different individuals have different definitions of privacy.Even for different individuals,the same data has different degrees of privacy protection.How to personalize privacy protection for different individuals has become a hot topic in the field of anonymous privacy protection.This paper is based on the personalized privacy protection demand,protect personal information from being leaked and reduce data loss after anonymization,and analyze and research the personalized anonymous privacy protection of data.The main work is as follows:(1)Combines traditional (a,k)-anonymous privacy protection algorithms with (p,k)-anonymous privacy protection algorithms.This paper proposes a personalized (p,a,k)-anonymous privacy protection algorithm.The algorithm classifies the attribute values of sensitive attributes according to the sensitivity of user-defined sensitive attributes,and uses the sensitive values of each level in the equivalence class.Different anonymous methods are used to achieve personalized privacy protection for sensitive attributes,while effectively defending against (a,k)-anonymous models and (p,k)-anonymous models with skewed attacks and probabilistic attacks resulting in privacy leakage.Experiments show that our personalized privacy protection algorithm can better protect personal privacy than other personalized privacy protection algorithms,and can greatly reduce the loss of data and greatly improve the availability of data.At the same time,there is an approximate better time cost.(2)The traditional method for generating the equivalence class of the anonymous model is easy to form an equivalence class that is highly correlated on the sensitive attribute,which is easy to trigger the similarity attack.The second generation of the equivalence class causes a large amount of information loss.The weighted clustering-based personalized anonymous privacy protection algorithm uses the weighting method to align the identifier attributes to perform similarity clustering in the process of generating equivalence classes,so that the tuples in the clustered clusters are quasi-identified.The maximum similarity of the attribute values,and the inverse clustering of the sensitive attributes,so that the tuples in the clusters are most different in the sensitive attribute values,and then each of the generated clusters is personalized and anonymous.Implement personalized constraints on sensitive attributes.The experiment result show that This algorithm can protect personal privacy to the greatest extent,and greatly reduces the loss of information in the generalization process.In summary,this paper focuses on the personalized anonymous privacy protection algorithm.By studying the needs of user personalized privacy protection,it provides a protection method for data information.Compared with the traditional anonymous algorithm,the proposed algorithm can better protect the data.Personal privacy is not compromised,while greatly reducing the loss of information on anonymous data and increasing the availability of data.