基于等距变换的聚类挖掘敏感信息保护方法

摘要

经典基于旋转的数据转换(RBT)算法需要预先设定安全度值,而目前并无有效规则量化该值.为此,提出随机选取等距变换角度的方法,在一个计算合理的数据区间内随机选取安全度阈值,使原始数据集经过数据转换后保持任意数据点在空间中的距离不变.理论分析和实验结果表明,该算法易于实现,转换后的数据集较原始数据集发生改变,且每次对数据的转换都是随机的,攻击者不能推导出原始数据,确保算法在完成数据变换的同时较好地保护敏感信息.%To solve the disadvantage of classic Rotation-based Transformation(RBT) algorithm which is ineffective of quantification security degree that has to be preliminary set up, a method of selecting randomly isometric transformation angles is presented. This method randomly selects the security degree in a reasonable range of data set. It can insure that the spatial distance for any two points in the new data set is the same as in the raw data set after the raw data set is transformed into the new data set The theoretical analysis and experimental results show that, the presented algorithm is simple and easy to implement. And the transformation of the data is random every time and the new data set is obviously different from the raw data set. The attacker can not trace the original data sets, and it can preserve the private information.