A Fast Algorithm for Deriving Frequent Itemsets

Abstract

When mining frequent itemsets (abbr. FIs) from dense datasets, too many itemsets are generated and results in the mining task from a large amount of execution time and high memory consumption. Frequent closed itemset (abbr. FCI) is a lossless and concise representation of FIs. Mining FCIs can not only greatly reduce the execution time and memory consumption, but also retain the complete information all of FI. Although many studies have proposed different mining FCI algorithms, but they have less developed methods that can effectively derive all FIs from FCIs. Form this point of view, this study proposes a novel efficient algorithm named DFI-List for efficiently deriving FIS from FCIs. The algorithm adopts the methodology of depth-first-search and divide-and-conquer to derive all FIs from FCIs. DFI-List efficiently derives all the FIs with vertical index structure called Cid List and uses SC Table to quickly find the support count of the derived FI. Experimental results show that the execution speed and memory consumption of the proposed algorithm with the proposed strategy is better than of the state-of-art algorithm.