摘要
为避免多维数值敏感属性数据发布中的近似猜测攻击,基于分解思想提出了一种有效的数据发布方法(l-MNSA).首先通过按敏感属性值均匀间隔分组的方法,提出针对单维数值敏感属性的l-SNSA算法;然后提出最小距离的思想,通过将敏感属性统一化并按最小距离均匀间隔分组,提出适用于多维数值敏感属性的l-MNSA算法.与以往仅针对单敏感属性的发布算法相比,该算法同时能对多维敏感属性提供较好的保护.实验结果表明,采用l-MNSA算法发布的数据,其组内最小差异与l-SNSA算法针对各维属性分别发布的结果相比,平均降低10%左右,算法时间复杂度仍为O(nlgn).该算法可以较好地均衡发布数据的安全性和可用性,是有效可行的.
Proximity breach is a privacy threat specific to numerical sensitive attributes in data publication.This paper tries to remedy the problem by introducing a novel principle called l-MNSA（l-multi numerical sensitive attribute） approach based on the idea of lossy join.Firstly,a data publishing algorithm concentrating on tables with only one numerical sensitive attribute,i.e.l-SNSA（l-single numerical sensitive attribute） algorithm,is proposed,in which the sensitive attribute is grouped by their values.Then,the idea of shortest distance is suggested.By unifying the sensitive attributes value and grouping them by their shortest distance,l-MNSA algorithm is proposed.Compared with previous algorithm for single sensitive attribute,l-MNSA can provide better protection to the multi numerical sensitive attributes.The results show that the minimum difference of data published by l-MNSA is reduced by 10% compasing to that of l-SNSA,meanwhile,the time complexity is O（nlgn）.The l-MNSA can better balance the published data＇s security and availability,being feasible and effective.
出处
《东南大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2010年第4期699-703,共5页
Journal of Southeast University：Natural Science Edition
基金
国家自然科学基金资助项目(60973023)
江苏省自然科学基金资助项目(BK2006095)
关键词
隐私保护
多敏感属性
数值型数据
数据发布
privacy preserving
multiple sensitive attributes
numerical data
data publishing
