(k,l)-多样性数据发布研究

Achieving(k,l)-Diversity in Privacy Preserving Data Publishing

发布未经处理的数据会导致身份泄露和敏感属性泄露,通过概化准标识符可以达到隐私保护的目的,但信息损失过大。针对该问题提出一种基于聚类的(k,l)-多样性数据发布模型并设计算法予以实现。通过使用概率联合分布度量数据对象的离散属性和连续属性相似性,提高了数据的效用。详细论述了簇的合并、调整和概化策略,结合参数k和l提出隐私保护度概念,指出了基于聚类的最优化(k,l)-多样性算法是NP-难问题,并分析了算法的复杂度。理论分析和实验结果表明,该方法可以有效减少执行时间和信息损失,提高查询精度。

In order to avoid disclosure of individual identity and sensitive attribute,reduce the information loss when da- ta release, a clustering-based algorithm to achieve（k, l）-diversity（CBAD）in data publishing was presented. The discrete attributes and continuous attributes mixed in the data set were fully taken into account while clustering. The probability distribution was used as metrics to measure similarity between the data objects. We solved the confusion of the informa- tion loss and the distance between data objects, pointed out that the clustering-based optimization（k,/）-diversity algo- rithrn is NP-hard problem, proposed the concept of privacy protection degree with parameter k and l, and analysed the complexity of the algorithm. Theoretical analysis and experimental results show that the method can effectively reduce the execution time and information loss, improve query precision.