Efficient Mining of Multiple Fuzzy Frequent Itemsets

Traditional association-rule mining or frequent itemset mining only can handle binary databases, in which each item or attribute is represented as either 0 or 1. Several algorithms were developed extensively to discover fuzzy frequent itemsets by adopting the fuzzy set theory to the quantitative databases. Most of them considered the maximum scalar cardinality to find, at most, one represented item from the transformed linguistic terms. This paper presents an MFFI-Miner algorithm to mine the complete set of multiple fuzzy frequent itemsets (MFFIs) without candidate generation. An efficient fuzzy-list structure was designed to keep the essential information for mining process, which can greatly reduce the computation of a database scan. Two efficient pruning strategies are developed to reduce the search space, thus speeding up the mining process to discover MFFIs directly. Substantial experiments were conducted to compare the performance of the proposed algorithm to the state-of-the-art approaches in terms of execution time, memory usage, and node analysis.