一种基于SMC和RD的隐私保护挖掘算法

针对当前的隐私保护关联规则挖掘算法在应用中存在的一些不足,如：在算法效率与计算精确度之间顾此失彼,提出了基于安全多方计算方法与随机干扰方法相结合的PPARMSR挖掘算法.该方法结合了安全多方计算方法的精确性和随机干扰方法的高效率,使用项集随机干扰矩阵同时干扰多属性,且兼顾它们的相关性,提前对候选频繁项集的剪枝,减少安全多方计算的处理量.通过试验验证了PPARMSR算法在保持隐私度的同时,还能够在计算精确度与运行效率之间实现较好的折中效果.

A mining algorithm based on SMC and RD for the privacy protection

In view of the shortcomings of the current mining algorithm for the privacy preserving association rule,such as only taking into account the algorithm efficiency while neglecting the calculation precision,or vice versa,this paper proposes the mining algorithm of PPARMSR,which makes use of the advantage of the accuracy of the secure multi-party computation and that of the high efficiency of the random disturbance method.The mining algorithm uses a set of random interference matrix and interference on multiple attributes,and gives consideration to their correlation.The method of pruning the candidate frequent itemsets in advance reduces the processing amount of secure multi-party computation.Experimental results show that the PPARMSR algorithm maintains the privacy degree,at the same time,it is able to keep a balance between the calculation precision and the running efficiency.