Research On Differential Privacy Based On Broadening Histogram With Constraint Error

Into the twenty-f irst century,the rapid development of the In ternet and large data technology bring a new experience to our lives,but enjoy the conve nience at the same time,but also so that we fall into the unknown danger.For example,our phone m ay be eavesdropped,when online shopping we left the identity information,at any time faced by some businesses or criminals use them to harm our interests and threaten our security.These are due to data leakage or release of data without taking into account the user's privacy.How to enjoy the shared network to provide users with a lot of information at the same time,but also to ensure that the user's sensitive information is not leake d,we will re-f ocus on privacy se curity.From a technica l perspective,scientists and technicians need to study new technologies to solve this problem.In this paper,differential privacy histogram publishing technology is studied in the context of differential privacy,the existing differential privacy histogram publishing technique does not pay attention to the characteristics of data,wh ich leads to poor accu racy and poor usability.In real life,the distribution of data presents many forms,especially in the histogram identification data,some of the histogram and adjacen t histogram data vary greatly,so the release of the histogram on a serious im pact on t he data accur acy.After analyzing th e characteristics of the uneven distribution of data,this paper uses the new strategy to publish the data.Three different algorithms are proposed for di fferent application backgrounds,na mely,A Differential Privacy Algorithm for Narrowing the Histogram with C onst,A Differential Privacy Algorithm for Narrowing the Histogram with Con st when equivalent query,A Differential Privacy Algorithm for Narrowing the Histogram with Const when range query.The three algorithms use the widened histogram that constrained the error to be C onst to solve the prob lem that the num ber of histograms is too large when using the equal-width histogram,and apply the narrowed histogram with Const to the differential privacy,and finally adopt the greedy strategy to realize the merging of the adjacent bucket,so as to guarantee the high accuracy of the published data,and protect the privacy data.The average error a nd the running time of the algorithm in this paper are dif ferent from those of the same algorithm.The results show that the new algorithms have higer accuracy in data distribution,to ensure the availability,and good protection of the user privacy.