基于约束聚类的k-匿名隐私保护方法  

K-anonymity method based on restrained clustering for privacy preservation

作　　者：吴梦婷 孙丽萍 刘援军 胡朝焱 赵延年 罗永龙 WU Meng-ting;SUN Li-ping;LIU Yuan-jun;HU Zhao-yan;ZHAO Yan-nian;LUO Yong-long(School of Computer and Information,Anhui Normal University,Wuhu 241000,China)

机构地区：[1]安徽师范大学计算机与信息学院,安徽芜湖241000

出　　处：《计算机工程与设计》2021年第3期607-613,共7页Computer Engineering and Design

基　　金：国家自然科学基金面上基金项目(61672039、61972439)。

摘　　要：针对现有的匿名方案往往较少考虑离群数据的敏感问题以及信息损失与时间效率的最优化问题,提出一种基于约束聚类的k-匿名隐私保护方法。通过K近邻思想划分初始集群,根据设定的阈值δ将集群进行重新划分,划分过程始终遵循信息损失最小化原则,得到每个等价类元组数都在k与2k之间,过程中分类考察准标识符属性并充分考虑离群点对聚类结果的影响,有效降低匿名过程中的信息损失。实验结果表明,该方法有效节省了执行时间并降低了信息损失。The existing anonymity schemes fail to consider the sensitive problems of outlier data,the optimization problem of information loss and time efficiency.Therefore,a k-anonymity privacy preservation method based on constraint clustering was proposed.The initial cluster was divided based on K-nearest neighbor idea.And the cluster was subdivided according to the set threshold valueδ.Partitioning process followed the principle of minimizing information loss.The number of tuples of each equi-valent class was between k and 2k.In the process of classification,quasi identifier attributes were inspected in group and the impact of outliers on the clustering results was fully considered,which effectively reduced the information loss in the process of anonymity.Experimental results show that the proposed method is effective in reducing the running time and information loss.

关 键 词：隐私保护 数据发布 约束聚类 K-匿名 信息损失 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]