An Enhanced Algorithm for Dynamic Data Release Based on Differential Privacy

Abstract

Abstract Privacy preserving of dynamic data release is a research hotspot all the while. From the previous study, the release of dynamic data set has the problems of high computational complexity, poor data availability and more noise. To solve these problems, we present a method named Dynamic Data Release on Numeric (DP-DDRN) and a modified strategy based on dynamic similarity for optimal noise sets. Firstly, when the privacy budget remains the same, the DP-DDRN is to classify data set, compute the weight and dynamic sensitivity of each group, and then add Laplace noise according to the weight and the dynamic sensitivity. Secondly, the similarity between the processing results and the original data is calculated using the dynamic similarity algorithm; if the similarity is under a threshold, DP-DDRN algorithm will adjust strategy to reprocess the original data in loops until the threshold is reached, and then the processing results are kept and released. Experiments show that, our algorithm can protect the privacy of data with a low computational complexity and a higher data accuracy result.