语义相似和多维加权的联合敏感属性隐私保护  

Privacy protection method for composite sensitive attribute based on semantics similarity and multi-dimensional weighting

作　　者：徐龙琴[1] 刘双印[1,2] 

机构地区：[1]广东海洋大学信息学院,广东湛江524088 [2]中国农业大学信息与电气工程学院,北京10083

出　　处：《计算机应用》2011年第4期999-1002,共4页journal of Computer Applications

基　　金：国家星火计划项目(2007EA780068);广东省自然科学基金资助项目(7010116);广东省粤港关键领域重点突破项目(2010B020315025);广东省科技计划项目(2008B021300002);湛江市科技计划项目(2010C3113011)

摘　　要：针对现有k-匿名方法直接用于多敏感属性数据发布中存在大量隐私泄露的问题,提出一种基于语义相似和多维加权的联合敏感属性隐私保护算法。该算法通过语义相似性反聚类思想和灵活设置多敏感属性值的权值,实现了联合敏感属性值和语义多样性分组的隐私保护,并根据应用需要为数据提供不同的隐私保护力度。实验结果表明,该方法能有效保护数据隐私,增强了数据发布的安全性和实用性。In view of a large number of privacy disclosure issues when using k-anonymity method directly for multi-sensitive attribute data publishing,a joint privacy-sensitive properties preserving algorithm based on semantic similarity and multidimensional weighting was proposed.This algorithm realized security protection of the joint-sensitive property value and the semantic diversity of the privacy group with the help of the semantic similarity anti-clustering principle and counter-sensitive property value.According to different application needs,data privacy protection of different extent was provided.The experimental results show that this method can effectively protect data privacy and enhance data security and practicality.

关 键 词：隐私保护 联合敏感属性 语义相似度 多维加权 l-diversity 

分 类 号：TP309.2[自动化与计算机技术—计算机系统结构]