基于属性分区的(α_i,k)-p隐私保护算法

(α_i,k)-p Privacy Preserving Algorithm Based on Attribute Partition

针对(αi,k)-匿名算法使用有损链接思想无法对用户身份进行保护的问题,引入属性分区置换概念,提出基于属性分区的(αi,k)-p匿名算法,对桶中QI属性采取分区、置换的方式保护用户身份信息。在人口真实数据集21956条数据上对两种算法进行敏感值保护和会员身份保护有效性对比实验。结果表明,敏感值泄露概率最高时只刚好超过0.05,接近传统方法的1/4;在会员身份保护方面,FOR值在0.7以上。相对于(αi,k)-匿名算法,该算法能更好地保护敏感值信息和会员身份信息。

The (αi,k)-anonymity algorithm proposed by Jinhua uses the idea of lossy links,and it can not provide the protection of user identity.In this paper,the idea of attribute partition replacement is introduced,and an (αi,k)- p anonymity algorithm based on attribute partition is proposed.QI attribute partition and replacement in bucket are adopted to protect user's identity information.This paper implements two algorithms for 21 956 data sets of real population,and compares the effectiveness of sensitive value protection and membership protection.The results show that the leakage probability of sensitive value is just over 0.05,which is close to 1/4 of the traditional method,and FOR value is above 0.7 in membership protection.Compared with (αi,k)-anonymous algorithm,the proposed algorithm can better protect sensitive value information and membership information.