Enhancing SWF for Incremental Association Mining by Itemset Maintenance

Abstract

Incremental association mining refers to the maintenance and utilization of the knowledge discovered in the previous mining operations for later association mining. Sliding window filtering (SWF) is a technique proposed to filter false candidate 2-itemsets by segmenting a transaction database into partitions. In this paper, we extend SWF by incorporating previously discovered information and propose two algorithms to boost the performance for incremental mining. The first algorithm FI_SWF (SWF with Frequent Itemset) reuses the frequent itemsets of previous mining task to reduce the number of new candidate itemsets that have to be checked. The second algorithm CI_SWF (SWF with Candidate Itemset) reuses the candidate itemsets from the previous mining task. Experiments show that the new proposed algorithms are significantly faster than SWF.