相异敏感度下最小信息损失增量优先的隐私保护方法  

A privacy preserving approach based on minimum information loss increment first for dissimilar sensitivity

作　　者：谢静[1,2] 张健沛[2] 杨静[2] 张冰[3] 

机构地区：[1]武汉纺织大学管理学院,湖北武汉430200 [2]哈尔滨工程大学计算机科学与技术学院,黑龙江哈尔滨150001 [3]哈尔滨理工大学软件学院,黑龙江哈尔滨150080

出　　处：《中南大学学报（自然科学版）》2015年第12期4548-4555,共8页Journal of Central South University:Science and Technology

基　　金：国家自然科学基金资助项目(61370083;61073043;61073041;61402126;71571139;71171153);高等学校博士学科点专项科研基金资助项目(20112304110011;20122304110012)~~

摘　　要：针对不同敏感值的隐私保护程度需求,提出一种敏感度计算方法,将敏感值进行等级划分,再对不同等级的敏感值设定不同的敏感度;给出一种隐私保护原则(ε,k)-sensitivity来控制等价类中敏感度的分布情况,使得等价类中高敏感度的元组不会过多而造成隐私泄露;提出一种最小信息损失增量优先算法(minimum information loss increment first,MILIF)来实现隐私保护的要求。研究结果表明:所提出的方法在降低少量时间和保持数据效用的前提下,充分提高了数据表抵御敏感性攻击的能力。In order to satisfy the different privacy protection requirements for different sensitive values, a method was proposed to calculate the sensitivity of sensitive value, which was divided into several levels with different sensitivities. A（ε, k）-sensitivity principle was proposed to control the distributions of sensitivity in equivalence class and the number of the high sensitivity tuples. A minimum information loss increment first algorithm was proposed. The results show that the proposed method can improve the ability of resisting sensitivity attack, on the premise of expending a little time and maintaining a high data utility.

关 键 词：隐私保护 敏感度 邻域 信息损失增量 

分 类 号：TP309.2[自动化与计算机技术—计算机系统结构]