一种基于最小选择度优先的多敏感属性个性化l-多样性算法

Personalized l-Diversity Algorithm for Multiple Sensitive Attributes Based on Minimum Selected Degree First

数据发布中的隐私保护技术一直是数据挖掘与信息安全领域关注的重要问题.目前大部分的研究都仅限于单敏感属性的隐私保护技术,而现实生活中存在着大量包含多敏感属性的数据信息.同时,随着个性需求的不断提出,隐私保护中的个性化服务越来越受研究者的关注.为了扩展单敏感属性数据的隐私保护技术以及满足个性化服务的需求问题,研究了数据发布过程中面向多敏感属性的个性化隐私保护方法.在单敏感属性l-多样性原则的基础上,引入基于值域等级划分的个性化定制方案,定义了多敏感属性个性化l-多样性模型,并提出了一种基于最小选择度优先的多敏感属性个性化l-多样性算法.实验结果表明:该方法不仅可以满足隐私个性化的需求,而且能有效地保护数据的隐私,减少信息的隐匿率,保证发布数据的可用性.

Privacy preserving in data publishing （PPDP） is always an important issue in the fields of data mining and information security. So far, most of the research on privacy preserving technology is limited to single sensitive attribute, but there are a lot of data information which includes multiple sensitive attributes in the real life. In the meanwhile, more and more researchers are paying great attention to the personalized service in the process of privacy preserving as the requirements of personality put forward continuously. To expand the privacy preserving, technology for single sensitive attribute and to satisfy the requirement of personalized service, the personalized privacy preserving approaches for multiple sensitive attributes in the process of data publishing are studied. Based on the single sensitive attribute /-diversity principle, a personalized multiple sensitive attributes /-diversity model is defined by introducing a personalized customization scheme based on domain hierarchies partitions. In the meanwhile, a multiple sensitive personalized /-diversity algorithm based on the minimum selected degree first （MSFMPL-diversity） is presented. The experimental results show that the proposed method not only can satisfy the requirement of individual personalized privacy, but also can protect the data privacy effectively and reduce the information hidden rate, which ensures the usability of the publishing data.