A key issue in mining association rules is to find out all frequent itemsets, therefore how to mine frequent itemsets quickly has been hot in current research. Mining algorithms of the maximal frequent itemsets based on FP-trees necessitate not only the multiple generations of large numbers of FP-trees, but also the multiple traversals of these FP-trees, thus taking much time. Against the above shortcomings, we propose an FP-tree-based algorithm MMFI optimized with array and matrix for mining the maximal frequent itemsets. It not only reduces the quantity of the FP-trees constructed, but also saves the time in traversing the FP-trees. Finally, we have compared the algorithm MMFI with the algorithm FP-MAX, the results of our experiment have shown that this algorithm is working efficiently.
展开▼