一种带频繁项过滤机制的隐私保护新方法

针对差分隐私保护方法的隐私保护过度问题,提出了一种带频繁项过滤机制的隐私保护新方法,以提高数据发布结果的准确性。该方法首先对数据源进行预处理,即对非频繁项进行过滤,然后执行差分隐私保护算法。从理论上证明了带频繁项过滤机制的隐私保护方法达到差分隐私保护级别,而且实验结果表明,在相同的隐私保护度下,提出的方法数据发布准确性比当前差分隐私保护方法更高。

Novel method of frequent item filtering mechanism in privacy-preserving

In order to improve the accuracy of publishing dataset, this paper put forward a novel privacy preserving method that was based on frequent item filtering mechanism, which aimed at to solve the existing issue of over-protected privacy in differential-privacy preserving methods. This method solved the issue by preprocessing the data source, namely, filtering non-frequent items firstly, and then performing the differential-privacy algorithm. The protection level of a frequent item filtering mechanism of privacy preserving method was proved to be differential-privacy in this paper. Experiment results show that the accuracy of the publishing data that of proposed method is higher than the current differential-privacy methods under the same degree of privacy protection.