Research On Privacy Preserving Clustering Mining Method

With the development of data mining technology and the large appearance of mining tools, people's confidentiality requirements about their privacy are becoming more urgent. On the one hand, data mining can bring tremendous social benefits; On the other hand, the misuse of the data also causes a serious of threat to people's privacy. In this case, how to mine data under the premise of preserve personal privacy, has become an urgent problem to address. At present, the problems of privacy preserving clustering are less studied, and the used methods are very single. Clustering mining is an important method of analyzing and managing questions. It is often used in market segmentation, customer classification, pattern recognition, Web document classification, the cell design of manufacturing systems and other important areas.Through the in-depth research and analysis of the existing privacy protection cluster mining methods, we find that geometric data transformation methods are most simple to apply without affecting the accuracy of mining results, but have a lower degree of privacy preserving. In order to solve the low privacy of the existing methods of geometric data transformation, this paper presents the method of plane reflection based geometric data transformation and the random response method of geometric transformation separately.The method of plane reflection based geometric data transformation is as follows: first we choose a straight line in the plane, then match the attributes into pairs to form points on the plane, and compute the symmetry point of each point. The result is the transformed data. The test proved that this method is simple and has a higher degree of privacy protection than translation, scaling, rotation and other geometric data transformation methods.To further enhance the degree of privacy preserving, this paper raises the random responses method of geometric transformation. The random response method of geometric transformation combines the random response technology with the geometric transform algorithm. According to the different random number generated by random number generator, different geometric transformation method is selected, which has played a dual effect of privacy protection. Our experiment proves that this method has a higher degree of privacy preserving, and it is feasible and efficient.