A DENSITY-BASED MICRO AGGREGATION TECHNIQUE FOR PRIVACY-PRESERVING DATA MINING

S. K. Das, B. Borah

Abstract: Microaggregation is an effective means of protecting the microdata in the statistical databases. Microaggragation protects the microdata by partitioning them into groups of at least k records in each group and substituting the records in each group with the centroid of the group. An optimal microaggregation can be achieved by minimizing the information loss incurred from the aggregation. This paper presents a density-based microagregation method for protecting the numeric data employing the densitybased notion of clustering. In this work we provide a microaggragation method which reduces the risk of microdata disclosure and incurs the minimum information loss with incresed data utility. In addition to it the paper also shows an experimental compariosn with the existing heuristics for microaggregation

Keywords: Privacy-preserving, microdata, density-based, microaggregation, clustering

DOI: https://doi.org/10.15623/ijret.2013.0214019