Mining and Ranking Association Rules in Support, Confidence, Correlation, and Dissociation Framework

摘要

Existing methods in association rule mining based on traditional support-confidence framework generates huge number of frequent patterns and association rules often ignoring the dissociation among items. Moreover these procedures are unable to order the rules by comparing them to find which one is better than whom. We have introduced a new algorithm for mining frequent patterns based on support and dissociation and thereafter generating rules based on confidence and correlation. The association rules have been ranked based on a composite index computed from the four measures. The experimental results obtained after implementation of the proposed algorithm justify our approach.