Title: Redundant association rules reduction techniques

Authors: Mafruz Zaman Ashrafi, David Taniar, Kate Smith

Addresses: Clayton School of Information Technology, Monash University, Clayton, Vic 3800, Australia. ' Clayton School of Information Technology, Monash University, Clayton, Vic 3800, Australia. ' Clayton School of Information Technology, Monash University, Clayton, Vic 3800, Australia

Abstract: To discover hidden correlations, association rule mining methods use two important constraints known as support and confidence. However, mining methods are often unable to find the best value for these constraints: large number of rules when these thresholds are low; very few rules when these thresholds are high. In addition, regardless of these above thresholds, mining methods produce many rules that have identical meaning or, redundant rules. Indeed such redundant rules seem as a main impediment to efficient utilisation of discovered rules, and should be removed. To achieve this aim, here we present several methods that identify those rules that are redundant and eliminate them.

Keywords: association rule mining; support; support thresholds; confidence; interest; mining methods; redundant association rules; data mining.

DOI: 10.1504/IJBIDM.2007.012945

International Journal of Business Intelligence and Data Mining, 2007 Vol.2 No.1, pp.29 - 63

Published online: 31 Mar 2007 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article