摘要
针对传统匿名算法采用相同的匿名强度实现k-划分,常导致所要发布数据的隐私保护程度与数据可用性之间失衡的问题,提出一种基于粗糙集属性重要度的多约束匿名化方法。根据准标识符属性重要度的差别,对准标识符属性维度进行自动划分,实现多约束匿名参数的设计,对具有不同维度的划分进行相应的匿名化操作。基于粗糙集理论和信息熵理论,设计了一种分类型数据可用性评估模型。从数据泛化后的信息损失、等价类对集合划分导致的信息熵改变两方面综合评估匿名化数据表的信息损失量。实验结果表明,采用该方法能够较好地实现数据的隐私保护和数据可用性之间的平衡。
To erase the imbalance phenomenon between the privacy protection and the utility of anonymized data caused by identifying all attributes having the same importance degree in the traditional algorithm, a multi-constraint anonymous method based on the attribute significance of the rough set was proposed, which took into account the influence caused by various quasi-identi- fier attributes. The dimension division was carried out automatically according to the quasi-identifier attributes significance and thereby the design of multi-constraint anonymous parameters was realized. After that, an anonymous operation was executed on the separate partition. Additionally, a model for evaluating the utility of anonymized data based on both the rough set theory and the information entropy theory was designed, which comprehensibly evaluated the information loss of anonymized data by considering the information loss of generated attribute values and the change of the information entropy caused by using equivalence classes to partition the data set. Experimental results show that the method better balances the privacy protection degree and the data availability.
出处
《计算机工程与设计》
CSCD
北大核心
2014年第8期2769-2772,2784,共5页
Computer Engineering and Design
基金
国家自然科学基金项目(61070139)
江西省自然科学基金项目(20114BAB201039)
关键词
数据发布
隐私保护
多约束
粗糙集
属性重要度
data publication
privacy preservation
multiple constrains
rough set
attribute significance
