01-03-2017, 10:10 AM
The publication of anonymized data has received considerable attention from the research community in recent years. For numerical sensitive attributes, most existing privacy-preserving data publishing techniques are concentrated on micro-data with multiple categorical sensitive attributes or only a sensitive numeric attribute. However, many real-world applications may contain multiple numeric sensitive attributes. Directly applying existing privacy preservation techniques to attributes of a single numerical-sensitive and multiple-categorical-sensitive attribute often causes unexpected disclosure of private information. These techniques are particularly prone to proximity breach, which is a specific privacy threat of numerical sensitive attributes in data publishing. In this paper we propose a method of publishing data that preserves privacy, that is, MNSACM, which uses grouping and Multi-Sensitive Bucketization (MSB) ideas to publish micro-data with multiple numerical sensitive attributes. We use an example to show how effective this method is in protecting privacy when using multiple numeric sensitive attributes.