Interscience Open Access Journals
The a priori algorithm for association rule mining is time consuming in the initial stages of the algorithm since usually large number of candidate itemsets is generated. To cope with this, several alternatives are suggested using hash functions. The authors provide a method to arrive at hash functions for 2-itemsets which can be extended to find hash function for 3- itemset also. When these hash functions used on frequent 1- itemsets, it will be more efficient in terms of space utilization than using the hash functions on set of all itemsets. For this they have used a mapping on 1-itemsets and provided an algorithm to use these mappings to find frequent 2-itemsets and 3-itemsets.