面向关联规则挖掘的随机响应技术研究
首发时间:2020-06-04
摘要:在现有的基于随机响应的隐私保护数据挖掘研究中,对于二项数据集,当随机响应的扰动过程建模成二进制对称信道时,扰动数据与原始的真实数据相关性仍然比较大,存在原始数据被还原的可能性,从而导致隐私保护效果不尽理想。针对这一问题,本文提出了一种改进的基于随机响应的联合扰动算法,针对某些特定场景下,对敏感程度有明显差异的敏感属性设置联合扰动机制,在满足挖掘准确性的前提下,降低挖掘数据与原始数据的相关性,从而增强对高敏感度的敏感属性的隐私保护程度。理论分析和仿真结果表明,在满足了一定的挖掘准确性的前提下,相比现有的算法,改进的扰动算法在隐私保护效果上的性能更优。
For information in English, please click here
Research on Randomized Response Technology for Association Rule Mining
Abstract:In the existing research on privacy preserving data mining based on randomized response, for the binomial data set, the correlation between the disturbance data and the original data is still relatively large. There is the possibility of a reduction of original data, which can lead to the privacy protection effect is not ideal. In order to solve this problem, this paper proposes an improved joint disturbance strategy based on randomized response. For some specific scenarios, a joint disturbance mechanism is set for sensitive attributes that have obvious differences in sensitivity, and on the premise of satisfying the accuracy of mining, the correlation between the mining data and the original data is reduced, so as to enhance the privacy of the high sensitivity of sensitive attributes. The theoretical analysis and simulation results show that, on the premise of satisfying the accuracy of mining, the improved perturbation algorithm has better performance in privacy protection compared with the existing algorithm.
Keywords: Data Mining Privacy Preserving Randomized Response Joint Disturbance
基金:
引用
No.****
动态公开评议
共计0人参与
勘误表
面向关联规则挖掘的随机响应技术研究
评论
全部评论0/1000