View article

Frequent Itemset Mining (FIM) is an important technique of analyzing data that can extract associations inside data. In the race of calculating frequent items efficiently using certain (precise) or uncertain (probabilistic or fuzzy) transactional database on single or distributed system, various sequential and parallel algorithms have been developed. Apriori, Count Distribution, UApriori and Uncertain Data Mining can be viewed as first and basic algorithms for calculating frequent items respectively on precise, distributed, probabilistic uncertain and fuzzy uncertain transactional databases. We proposed a new method for calculating frequent items on a fuzzy uncertain transactional database named as FuzzyApriori. Then, Introduced a new method to calculate frequent items on distributed system as Approximate Frequent Itemsets for a combination of certain and fuzzy uncertain transactional database as FuzzyMasterApriori. The performance of proposed FuzzyApriori is compared with UApriori, and the performance of proposed FuzzyMasterApriori is compared with the MasterApriori by carrying out different experiments on synthetic data and FIMI repository data. We got significantly improved results, as well as experiments, show the high scalability of proposed algorithms.