一种基于逆聚类的个性化隐私匿名方法

摘要

针对不同个体对隐私保护的不同需求,提出了一种面向个体的个性化扩展l-多样性隐私匿名模型.该模型在传统l-多样性的基础上,定义了扩展的l-多样性原则,并通过设置敏感属性的保护属性来实现个体与敏感值之间关联关系的个性化保护需求.同时,还提出了一种个性化扩展l-多样性逆聚类(PELI-clustering)算法来实现该隐私匿名模型.实验表明:该算法不仅能产生与传统基于聚类的l-多样性算法近似的信息损失量以及更小的时间代价,同时也满足了个性化服务的需求,获得更有效的隐私保护.%For achieving the different privacy preservation requirements of each individual,this paper presents a personalized extension /-diversity privacy anonymous model orienting individuals.This model proposes an extension /-diversity principle based on the traditional /-diversity,and realizes the requirement of personalized protection of relationship between individual and sensitive value by setting up guarding attributes on sensitive attributes.In the meantime,this paper also proposes a personalized extension l-diversity inverse clustering algorithm (PEU-clustering) to implement the privacy anonymous model presented in this paper.The experiments show that the proposed algorithm in this paper not only meets the requirements of personalized service,but also produces similar information loss to the traditional clustering-based /-diversity algorithm with less time cost,which achieves more effective privacy preservation.