Differentially Private Two-Dimension Sparse Data Publication under Consistency

Recently, differential privacy data publication has become an extremely important research topic in data security field. However, most of the differential privacy algorithms do not take sparse data publishing into consideration. The aim of this study is to present an effective differential privacy algorithm for two-dimensional sparse data publication, so as to boost the accuracy of range queries of the released data. The proposed approach in this paper includes two steps: 1) getting the sampling set of the original two-dimensional sparse dataset by adopting filter sampling algorithm, 2) building an incomplete quad tree based on the sampling dataset and adjusting the tree nodes´ noise values under consistency. Experimental analysis is designed by comparing the proposed algorithm and the traditional algorithms on the accuracy of range queries in the released data. Experimental results show that the proposed algorithm is effective and feasible.