SECURE ASSOCIATION RULE MINING IN HORIZONTALLY DISTRIBUTED DATABASES USING THRESHOLD-C AND UNIFY PROTOCOLS

Abstract

Abstract: Data mining is the drastic and fast growing area today which is used to extract knowledge from huge data collections but often these collections are separated among various parties. Privacy accountability may prevent the parties from honestly giving out the data and some sort of information about the data. In this project we propose a protocol for secure association rule mining in horizontally distributed databases. The existing integral protocol is that of Kantarcioglu and Clifton well known as K&C protocol. This protocol is based on an unsecured distributed version of the named as Fast Distributed Mining (FDM) algorithm of Cheung et al. The main constituents in our protocol are two novel secure multi-party algorithms one that process the union of private subsets that each of the interacting players hold and another that check whether an element held by one player is included in a subset held by another. This protocol suggests enhanced privacy with respect to the former protocols. In addition, it is not complex and is prominently more powerful in terms of communication cost, communication rounds and computational cost.