Research On Differential Privacy Protection For Data Release

In recent years,incidents of leakage of private information occur frequently.Nowadays,how to protect the collective and personal privacy has become the focus of the field of information security.Differential privacy protection only needs to add a small amount of noise to protect the privacy of users,which has gradually become an important direction of information security research.At present,the biggest bottleneck of differential privacy is how to distribute the private parameter ? reasonably and improve the availability of data under the condition that the private information is not revealed.In this paper,the pretreatment methods of static data set are improved.Study of adding noise and distribution of private parameters are as follows:First of all,differential privacy information is to release data under the condition that the data is distorted after adding noise to the data.In this paper,contrapose the problems that the differential privacy protection of histogram data has too much noise in static data set and the availability of data is too low,density-based clustering algorithm is used to preprocess the data and technique of differential privacy protection is applied to protect privacy.Contrapose the problem that DP-DBSCAN is sensitive to the input of parameters,the existing OPTICS algorithm is applied to differential privacy protection,based on which DP-OPTICS differential privacy protection algorithm is proposed.It improves the error of pretreatment of data.In particular,the differential privacy protection algorithm based on OPTICS clustering has been improved in dealing with data of low frequency,which improves the usability of data.Secondly,the way of adding noise is altered.A way of heteroscedasticity is applied to add noise to the data that has been preprocessed,which solves the problem of excess of accumulation of noise,while assuming the possibility of successfully breaking the private information under the maximum background,setting the upper limit of privacy parameters to effectively balance the relationship between the availability of data and the safety of private information within the possibility of compromising privacy protectionFinally,summarize the work that has been done and the direction of future efforts.