Hiding emerging patterns with local recoding generalization

Abstract

Establishing strategic partnership often requires organizations to publish and share meaningful data to support collaborative business activities. An equally important concern for them is to protect sensitive patterns like unique emerging sales opportunities embedded in their data. In this paper, we contribute to the area of data sanitization by introducing an optimization-based local recoding methodology to hide emerging patterns from a dataset but with the underlying frequent itemsets preserved as far as possible. We propose a novel heuristic solution that captures the unique properties of hiding EPs to carry out iterative local recoding generalization. Also, we propose a metric which measures (i) frequentitemset distortion that quantifies the quality of published data and (ii) the degree of reduction in emerging patterns, to guide a bottom-up recoding process. We have implemented our proposed solution and experimentally verified its effectiveness with a benchmark dataset. © 2010 Springer-Verlag Berlin Heidelberg.