Research On Personalized Privacy Preservation Method For Microdata Publishing

In the field of informatization,a large amount of personal data is collected and published every day.These individual related data are called microdata,for example,census data,personal consumption data,medical data and so on.If the personal microdata information can't be handled properly or publishing directly,the privacy issues will inevitably be brought to the individual.Thus,many researchers start to study the privacy preservation problems of microdata publishing.In addition,considering the difference of privacy protection requirements for individuals in real life,the study of personalized privacy preservation is a hotspot in the field of data publishing.Based on the personalized privacy preservation needs,this paper focuses on the research of personalized privacy preservation methods in the field of microdata publishing.The main contributions of this paper are as follows:(1)A personalized(?,l)-diversity k-anonymity privacy preservation model is proposed.Be aimed at the shortcomings of traditional k-anonymity model and l-diversity anonymity model,considering the lack of personalized anonymity in these two models,on the basis of k-anonymity and l-diversity model,a personalized(?,l)-diversity k-anonymity model is proposed to solve the existing problems.In the proposed model,the sensitive attribute values are divided into several categories according to their sensitivities,each category is assigned with corresponding constraints.By constructing a generalization hierarchy tree for sensitive attributes,specific individuals are allowed to set up privacy preservation level for their sensitive attribute value,and personalized privacy protection rules are formulated to provide personalized privacy preservation services for dataset.Then,according to the proposed personalized privacy preservation model,a clustering based personalized(?,l)-diversity k-anonymity algorithm is presented.Finally,it is proved by experiments that the proposed personalized privacy preservation method has a stronger privacy protection ability while providing personalized services effectively.(2)A personalized p-sensitive k-anonymity model that against similarity attack is proposed.Aiming at the problem that the traditional p-sensitive k-anonymity model and(?,k)-anonymity model can't resist the similarity attack for they ignored the semantic relations between sensitive attribute values,a personalized p-sensitive k-anonymity model that orienting sensitive attribute values is presented.According to the semantic correlation of sensitive attribute values,the sensitive values are divided into different semantic groups,and different frequency constraints are set for sensitive values in different groups.In the proposed model,the dataset is required to satisfy the p-sensitive k-anonymity model first,and then achieve the personalized constraints on the sensitive values.Then,the algorithm of personalized p-sensitive k-anonymity model is presented.It is proved by experiments that personalized p-sensitive k-anonymity model can provide stronger privacy protection for dataset in the case of little difference in the information loss from traditional p-sensitive k-anonymity model and(?,k)-anonymity model.