Research On Data Privacy Preserving Method For Clustering Based On Neighborhood Correlation

In recent years, how to ensure the data which is motivated by the need to both protect privileged information and enable its use for research after disturbing has become more difficulty. And how to meet the need during the publishing process is also a challenging and urgent problem. However,the traditional perturbation methods can not keep a balance between data privacy preservation and data clustering results.Although substantial research had been conducted on data privacy protection safety and the availability of data clustering methods which found that by combining neighborhood correlation clustering and data disturbing method can improve the results,only a few works were received good results on data privacy protection safety. To solve this problem, this paper presents a NCCQDP algorithm which based on NCDP algorithm to insure both individual privacy and information usefulness for clustering analysis. This paper attempts to improve the sensitive data privacy preservation safety and data clustering usability accuracy by sacrificing the space complexity and keeps them balancing from two angles to obtain the expecting result.NCCQDP algorithm includes three steps: firstly, selecting the cluster center based on data of the neighborhood correlation density; secondly,rough clustering on data points record set; at the last, data disturbance of the points record set based on the circle of four quadrant.Finally, experimental results on synthetic data sets of NCCQDP algorithm have shown significant improvements in clustering accuracy and individual privacy in comparison with the NCDP algorithm and RBT algorithm. And the results of NCCQDP algorithm keeps balance between data clustering accuracy and data individual privacy. And the direction of the further exploration is discussed.