Research On Histogram Data Publishing Method Based On Weighted Privacy Budget Allocation

In recent years,with the uninterrupted development and progress of network information technology applications,various information systems have stored and accumulated various types of rich data.However,the data set contains a large amount of personal privacy.If certain protection measures are not taken during the process of data release,as the data set is released and shared,personal information may be leaked,and data privacy protection problems will increase.The more people pay attention.In order to be able to provide data researchers with more effective data while protecting personal privacy information,a data publishing method that can ensure data privacy and data availability has important theoretical value and practical significance.Related researchers have proposed a variety of data release methods based on differential privacy,but most of them are release methods for static data and data streams,and there are relatively few research methods for updated dynamic data release.In addition,in the differential privacy dynamic data release,a reasonable allocation of a limited privacy budget can improve the availability of released data.Aiming at the two major problems of poor data availability and unreasonable privacy budget allocation,this paper proposes a histogram data publishing method based on weighted privacy budget allocation.This method uses Wasserstein distance to calculate the similarity of data published at two adjacent time points.By comparing the Wasserstein distance d with the threshold T,a new histogram publishing strategy is proposed,and a corresponding dynamic noise addition mechanism is provided.The Poisson distribution weighting method is used to allocate the privacy budget.After the distance similarity is detected,the privacy budget is for data that needs to be noisy to avoid premature exhaustion of the privacy budget.Experiments show that the method proposed in this paper meets differential privacy,and a reasonable budget allocation strategy not only reduces the overall error,but also ensures that the data availability is improved while reducing the noise error.