An Enhanced Semi-apriori Algorithm for Mining Association Rules

Mining association rules in large database is one of data mining and knowledge discovery research issue, although many algorithms have been designed to efficiently discover the frequent pattern and association rules, Apriori and its variations are still suffer the problem of iterative strategy to discover association rules, that’s required large process. In Apriori and Apriori-like principle it’s known that the algorithms cannot perform efficiently due to high and repeatedly database passes. In this paper we introduced an enhanced SemiApriori algorithm for mining frequent itemset as well as association rule, the algorithm work as follow, in the first step we use the first two steps used in Apriori algorithm in order to eliminate the repeatedly databases passes, in the third step we start discovering the rest of frequent pattern. In order to minimize the execution time as well as generating large number of candidate sets, we implement an enhanced Semi-Apriori algorithm using a binary based data structure, which is illustrated in details in enhanced Semi-Apriori technique section. Extensive experiments had been carried out, through comparing an enhanced Semi-Apriori with Apriori algorithm, the result show that our technique outperform Apriori in terms of execution time.