The Research On Data Publication Algorithms Satisfying Differential Privacy

Nowadays many service providers have accumulated myriad user data.Sharing data avoids data precipitation,but it involves user privacy.Thus,the issue of privacy protection in data release has gradually attracted the attention of academia and industry.Differential privacy protection model has been applied to varied fields due to its excellent performance.This paper studies the high-dimensional data release algorithm of differential privacy protection model.The aim of this paper is to improve the accuracy of the published data on the basis of ensuring that the release algorithm satisfies the differential privacy protection model,which can not effectively deal with the problem of high-dimensional data.In this paper,we improve the data release technology in differential privacy model.When the predicate is a range query on each attribute,it provides a more accurate query result.The core of this algorithm is to apply the wavelet transform to the data before adding noise.The two data types,ordinal and nominal data,are given the corresponding processing methods.And then the method is extended to multidimensional data.The algorithm proposed in this paper improves the privacy and usability of data release.In order to make the quality of the constructed Bayesian network higher,this paper presents a surrogate function of mutual information function.There is more total amount of mutual information under the same privacy budget in the case of differential privacy.So that we get a more accurate way to measure the information content between each pair of attributes.In the experimental part,through the comparison with the existing differential privacy model,it is proved that the two improved approaches proposed in this paper are presented in the application scenario of the multidimensional data set.In terms of improve the efficiency of the algorithm and the effectiveness of data.