Output Privacy Protection With Pattern-Based Heuristic Algorithm

TL;DR

This paper introduces a pattern-based heuristic algorithm for privacy-preserving data mining that effectively protects mined patterns from inference attacks while maintaining data utility.

Contribution

It proposes the Pattern-based Maxcover Algorithm, a novel approach that enhances output privacy in data mining by minimizing data dissimilarity and preventing pattern retrieval.

Findings

The algorithm reduces dissimilarity between source and released data.

Protected patterns cannot be retrieved even at low support thresholds.

Experimental results demonstrate the algorithm's effectiveness.

Abstract

Privacy Preserving Data Mining(PPDM) is an ongoing research area aimed at bridging the gap between the collaborative data mining and data confidentiality There are many different approaches which have been adopted for PPDM, of them the rule hiding approach is used in this article. This approach ensures output privacy that prevent the mined patterns(itemsets) from malicious inference problems. An efficient algorithm named as Pattern-based Maxcover Algorithm is proposed with experimental results. This algorithm minimizes the dissimilarity between the source and the released database; Moreover the patterns protected cannot be retrieved from the released database by an adversary or counterpart even with an arbitrarily low support threshold.