一种抗相似性攻击的匿名保护方法  

An anonymity protection method against similarity attack

作　　者：高帅 徐雅斌[1,2,3] 武装[2] GAO Shuai;XU Yabin;WU Zhuang(Beijing Key Laboratory of Internet Culture and Digital Dissemination Research,Beijing Information Science&Technology University,Beijing 100101,China;Computer School,Beijing Information Science&Technology University,Beijing 100101,China;Beijing Advanced Innovation Center for Materials Genome Engineering,Beijing Information Science&Technology University,Beijing 100101,China)

机构地区：[1]北京信息科技大学网络文化与数字传播北京市重点实验室,北京100101 [2]北京信息科技大学计算机学院,北京100101 [3]北京信息科技大学北京材料基因工程高精尖创新中心,北京100101

出　　处：《北京信息科技大学学报（自然科学版）》2020年第4期63-68,共6页Journal of Beijing Information Science and Technology University

基　　金：网络文化与数字传播北京市重点实验室项目(ICDDXN004);北京材料基因工程高精尖创新中心项目(C18601);信息网络安全公安部重点实验室开放课题项目资助(C18601)。

摘　　要：针对因敏感属性值语义相似而遭受到相似性攻击问题,提出一种抵抗相似性攻击的(k,e)算法。首先,根据敏感属性值构建敏感属性值语义层次树,计算敏感属性值相异度e,并根据敏感属性值相异度进行聚类;在此基础上,再按准标识符间距离进一步聚类,为每一类选取出k条距离最近的记录;最后,对每一类的准标识符进行泛化。对比实验结果表明,(k,e)算法具有较小的信息损失度和较低的隐匿率。For the problem of similarity attack due to semantic similarity of sensitive attribute value,a(k,e)algorithm is proposed in this paper.First,a semantic hierarchy tree of sensitive attribute value is constructed,and then the heterogeneity degree e of sensitive attribute value is calculated.Clustering is carried out according to the heterogeneity degree of sensitive attribute value.On this basis,further clustering is conducted according to the distance between quasi-identifiers,and k records with the closest distance are selected for each category.Finally,the quasi-identifiers of each category are generalized.The experimental results show that this method has lower information loss degree and suppression ratio.

关 键 词：相似性攻击 匿名保护 隐私保护 聚类算法 

分 类 号：TP302[自动化与计算机技术—计算机系统结构]