Research On High-dimensional Data Publishing Algorithm Based On Differential Privacy Protection Technology

Publishing and sharing of high-dimensional data have become an important basis for data analysis and utilization.However,there is a risk of privacy leakage in publishing highdimensional data directly.Therefore,differential privacy technology should be used for data privacy processing before publishing high-dimensional data.In this paper,differential privacy technology is applied to the privacy publishing process of high-dimensional data.There are some problems,such as adding noise directly to high-dimensional data,making the publishing results unusable,unable to choosing the personalized privacy budget allocation method according to the actual privacy protection needs,and unable to optimize the contradiction between data privacy protection intensity and availability Solution algorithm.The main contents are as follows.(1)This paper studies the Bayesian network model use for dimension transformation in the privacy publishing process of high-dimensional data,proposes MMPB algorithm,uses the maximum weight value and maximum support graph to limit the randomness of the Bayesian network structure and improves the accuracy of Bayesian network.Experimental results show that MMPB algorithm has a better publishing effect and lower time cost compared with existing algorithms.(2)This paper studies the waste problem of differential privacy budget in MMPB algorithm and designs MMPB-PPBA algorithm which can choose the allocation method of privacy budget according to the actual privacy needs.The algorithm sorts according to the importance of attribute fields,sets the ratio constant individually,and realizes the personalized choice of privacy budget allocation method.Through the experimental verification,MMPB-PPBA can reasonably allocate the privacy budget and according to the privacy needs of personalized privacy budget allocation.(3)MMPB algorithm cannot balance the contradiction between data security and availability,and MMPB-SG algorithm based on Stackelberg game is proposed.he dynamic game model between data release results and optimal attack strategy is constructed by setting the threshold of differential privacy budget,and the game equilibrium objective function is solved by the linear programming method.Experimental results show that MMPB-SG can effectively balance the contradiction between data utility and privacy and minimize the utility loss of published data while ensuring privacy protection.