Fast algorithm for mining fuzzy association rules

This paper proposes a novel approach for mining fuzzy association rules. It employs the adjacency list structure used in Apriori-Q algorithm to store items existing in database, which is convenient for the generation of frequent 2-itemsets. Besides, it uses vectors to represent texts, and frequent k-itemsets (k > 2) are discovered on the vector table. Since our method only requires two scans of database, the process of rule mining can be speeded up. In addition, it is not necessary to store candidate itemsets, then saving the store space. Experimental results show that this approach is feasible, and in comparison with the original algorithms, the mining efficiency is obviously improved.