A Distributed Privacy-Preserving Association Rules Mining Scheme Using Frequent-Pattern Tree

Association rules mining is a frequently used technique which finds interesting association and correlation relationships among large set of data items which occur frequently together. Nowadays, data collection is ubiquitous in social and business areas. Many companies and organizations want to do the collaborative association rules mining to get the joint benefits. However, the sensitive information leakage is a problem we have to solve and privacy-preserving techniques are strongly needed. In this paper, we focus on the privacy issue of the association rules mining and propose a secure frequent-pattern tree (FP-tree) based scheme to preserve information while doing the collaborative association rules mining. We show that our scheme is secure and collusion-resistant for n parties, which means that even if n-1 dishonest parties collude with a dishonest data miner in an attempt to learn the associations rules between honest respondents and their responses, they will be unable to success.