Privacy-Preserving Collaborative Association Rule Mining

Abstract

Abstract. This paper introduces a new approach to a problem of data sharing among multiple parties, without disclosing the data between the parties. Our focus is data sharing among parties involved in a data mining task. We study how to share private or confidential data in the following scenario: multiple parties, each having a private data set, want to collaboratively conduct association rule mining without disclosing their private data to each other or any other parties. To tackle this demanding problem, we develop a secure protocol for multiple parties to conduct the desired computation. The solution is distributed, i.e., there is no central, trusted party having access to all the data. Instead, we define a protocol using homomorphic encryption techniques to exchange the data while keeping it private.

Secure Extraction of Association Rule from Distributed Database

“…Secure clustering using the EM algorithm was implemented by Lin et al [4] over horizontally distributed data. In The problem of distributed association rule mining was studied in [5], and [6] but here the data was distributed vertically, where each party holds a different set of attributes. Also the work of [7] considered this problem in the horizontal setting, but they considered large-scale systems in which, on top of the parties that hold the data resources there are also managers which are computers that assist the resources to decrypt messages.…”

Privacy-preserving Data Mining

“…Vaidya & Clifton (2002) presented the component scalar product protocol for privacy-preserving association rule mining over vertically partitioned data in the case of two parties; Wright & Yang (2004) applied homomorphic encryption to the Bayesian networks induction for the case of two parties. Zhan et al, (2007) proposed a cryptographic approach to tackle collaborative association rule mining among multiple parties.…”

Preemptive Measures against Malicious Party in Privacy-Preserving Data Mining

“…The criteria to evaluate the "goodness" of each possible splits, such as information gain and entropy or gini, are securely computed. To mine association rules, various secure scalar product protocols with different complexity and levels of secu-rity [21,28,29] have been proposed to securely compute the support and confidence to determine if an association rule is frequent.…”