Research On Privacy Preservation Technology In Database Application

With the development of database application, people pay more and more attention to thesecurity of database, especially when the database application involves privacy information whichis related to peoples’ vital interests. In the application of data publishing, the data are published tothe public directly, so the protection of privacy is much more difficult. The main researches ofprivacy preservation data publishing concentrate on two points: one is principles and standards thatdatasets should obey to ensure the security, the other is methods for satisfying the principles.This thesis focuses on the privacy leakage problem in the data publishing and discusses theprivacy preservation technology. The main contributions of this thesis are as follows:Firstly, we summarize and analyse the anonymous technologies which are the most importantprivacy preservation technologies in the data publishing, including classic anonymity principlesand new development of anonymity principles.Secondly, we focus on the privacy preservation of datasets with numerical sensitive attributes;propose (Δ t,e)-anonymity principle based on steady development idea to defend approximateguess attack, and an equivalence group partition algorithm which can immensely satisfy thisprinciple.Thirdly, we focus on the small number of equivalence group in the anonymous table that doesnot obey steady development principle, and propose a scaling idea for those sensitive values whichare not “good”, and prove that the sensitive values in the equivalence group can satisfy the(Δt,e)-anonymity principle and difficult to be backward reasoning which means having a goodprivacy protection ability while we choose appropriate scaling paramters, objects and methods. Wealso prove that our methods having good data availability through comparing the approaches basedon generalization, permutation, and scaling in their expression of query and statistics. On the basis,we also propose methods to achieve better data availability for the use of more-trustable recipient.Finally, we realize these algothrims and compare our scaling algothrim with generalization,permutation algothrims about the data availability. Then we analyse our scaling algorithm’s timecomplexity and privacy protection degree, and prove the feasibility of our approach.