Abstract: An optimized privacy preservation method for OLAP data is presented First the method simulates the original result data of query using polynomial regression Then two efficient rules are adopted to optimize the simulated result Kullback-Leibler divergence is adopted to measure the information utility which is hard to evaluate Experimental results indicate that this method gets the satisfying efficiency on the balance of the privacy and the utility

Key words: privacy preservation, OLAP, regression

摘要： 提出一种对于联机分析处理（OLAP）数据的隐私保护方法。首先采用多项式回归的方法对查询数据进行初步模拟，在初步模拟基础上提出了两类有效的规则进行优化处理。该方法在隐私保护的同时保持了较好的信息有效性。同时利用Kullback-Leibler信息量来描述OLAP模拟数据的信息有效性，解决了有效性难以评价的问题。实验表明该方法在隐私保护和信息有效性两个方面均取得了满意的结果。

关键词: 隐私保护, 联机分析处理, 多项式回归