Using Genetic Algorithm for Extracting Association Rules

Abstract

The process of extracting interesting and unknown information from large database is called as association rule technology. The typical approach for solving association rule problem is Apriori Algorithm developed by Agrawal et al.[1993]. Most of the related existed works are improvements to this algorithm. The limitations of these algorithms are: (1) they required high storage space for saving the huge data resulting the generation of the frequent itemset, (2) they required encoding scheme where separate symbols are used for each possible value of an attribute of the itemset.In the present work, another trend of solution is proposed. First, we use Genetic Algorithm (GA) to define the maximal frequent itemset, so no huge storage requirement is needed. Also, we force the (GA) to work directly on database, so no encoding scheme is required. The calculations are based on our suggestion to use the variable length individual in the population.