The Application Research Of Differential Privacy Data Release On The Precision Poverty Alleviation Big Data Platform

In recent years,the rapid development of computer and information technology has promoted the arrival of big data era.While big data era promote information communicate and affect lifestyle,our privacy security is confronted with a challenge.It makes the data more transparent that sharing and releasing of data which relate to users'privacy information.Direct releasing of data may lead to the risk of privacy disclosure.Therefore,it becomes crucial to protect privacy information in the process of data publishing.Differential privacy is a widely studied and recognized privacy protection model,which keeps data usability while meeting privacy protection budget.Currently,the privacy protection technologies mainly involve in annoymous protection,access control,data encryption and data disturbance.Differential privacy which protect data as well by adding noise to data is a kind of data perturbation technology.The dissertation focuses on the research of differential privacy data publishing algorithms in the non-interactive framework.According to the data characteristics of targeted poverty alleviation big data platform,the dissertation mainly study the application of the differential privacy algorithm in single dimensional data and multidimensional data.The contributions of this dissertation are as follows:Firstly,the characteristics of single dimension and multidimensional data of targeted poverty alleviation big data platform are analyzed.On the basic of the characteristics of single-dimension data,the NoiseFirst algorithm can achieve privacy protect of single dimension data in the method of histogram publishing.Secondly,the insufficient of differential privacy algorithm in multidimensional data utilization is analyzed.Directly adding noise can effect the accuracy of multidimensional data,therefore an algorithm that extract the main ingredients can improve the accuracy of data by adding noise to the more significant data.Finally,according to the characteristics of the multidimensional data of targeted poverty alleviation big data platform,the application of principal component analysis differential privacy algorithm in multidimensional data is realized.Analysic the experimental results of principal component analysis differential privacy algorithm and classical differential privacy algorithm.Algorithm performance,privacy protection degree and data availability were used to evaluate and verify the differential privacy algorithm of principal component analysis.