Differential Privacy Histogram Distribution Algorithm Based On Gradient Drop Threshold

As an important part of computer technology,Internet play a more and more important role in people's life.The Internet space has become the another living space of every people,in this space,people social,do shopping,sharing,learning,have a very nice experience.However,with the increase in the time peolple spending on the Internet,people's property security,information security is bind to the network without notice.In this peaceful Internet environment which seems like one,crimes for the user information,users property happened more and more frequently.In recent years,by the prism event happened,people began to pay more attention to information security on the Internet,personal privacy security,the strategy for the protection of user privacy born and development fast.The object of this paper is about the differential privacy,which is a kind of strategy used on protecting user privacy.This strategy is using aggainst the background knowledge attack.Considered that most attention on the differential privacy not include the data itself,most popular strategy on prevent data leakage is adding noise,whitch lead to a condition that the data relased don't have a good accuracy,finally become poor usability.Aiming at the defects of traditional histogram data in the release of excessive error,reference to the structure_first strategy,design more reasonable,more friendly to the data itself,whitch greatly protect the validity of the data..The algorithm consist multiple steps to achieve different purpose,this paper introduces the algorithm very carefully,analysis every steps' purpose,and summed up the shortcomings of some of existing algorithms.In the experiment,two different kind of data were collected as the input of experiment.Finally,through the experiments and thoughtfully analysis,the original validity of the experimental and the defects of the algorithm are proved.