Mining rare association rules in a distributed environment using multiple minimum supports

Abstract

Distributed data mining of association rules is an area of data mining which intends to find association rules over items geographically across the network. Several researches have been performed in this field as applications have started to exploit distributed databases. Discovering rare association rules is a new area of distributed mining research. In this paper, an algorithm for discovering rare association rules in distributed environment is proposed. It utilized the idea of using statistic percentile to produce multiple minimum supports to mine rare association rules. Finally, the proposed algorithm has been implemented and evaluated by comparing with the Optimized Distributed Association rule Mining (ODAM) algorithm and the Apriori with Multiple Support Generating by statistic Percentile threshold (Apriori MSG-P) algorithm. The result shows that the proposed algorithm can discover more rare association rules with an optimized communication cost.