Research On Privacy Protection Model Of Dynamic Data Set Re-Publication

Data privacy protection technology has an important role and significance in the realworld. Among them, privacy protection in dynamic data set re-publication has been morewidely used, followed by the corresponding research in privacy protection model have beenfurther developed. However, there are some defects and shortcomings in the existing dynamicdata set re-publication of the privacy protection model.This article focuses on the M-Distinct model analysis. First, M-Distinct models hadconsidered the difference between sensitive attribute update probabilities. However, in theconcrete realization of the process, because of its random selection of such sensitive attributesof the updated set of candidate of the sensitive attributes that being updating, so this kind iftreatment does not completely solve the problem of attribute links, and the probability ofattack. If an attacker had gained two or more consecutive anonymous versions and somebackground knowledge, it would deduce the privacy leak by the different probabilities inupdating of the continuous release of any version of the anonymous relevant QI-Groupsensitive attribute values. For this reason, in this paper for sensitive attributes of each class setof the updated set of candidate, we create a expert systems for auxiliary candidates updatingand construct a tree for the corresponding sensitive attribute updating, in order to accuratelymeasure the updating properties and relationships in the updated set of candidate of this kindof sensitive attributes, make the equivalent probability updated between sensitive attributeswhich may lead to links and probabilities attacks using the limitation ofα_max, for areasonable allocation to the appropriate QI-Group, so that we can better address the disclosureof sensitive attributes that caused by attributes links and probabilities attacks. Second,M-Distinct model shows powerless in information leakage caused by permanent sensitiveattributes. In this paper, we considered that the difference of different attribute values,respectively deal with the non-sensitive attribute values assigned to the same QI-Group andthe record of permanent sensitive attribute values. Finally, proposed a dynamic data setre-publication model in privacy protection based on (m,α_max)-Distinct.Simulation result shows that the model can better solve privacy disclosure issues thatcaused by attribute links and probability attacks in dynamic data set re-publication, and also to some extent to keep data availability.