No abstract

A distributive data mining algorithm with association rules is presented in this paper, and the performance of layer matching algorithm is compared with Apriori's by examples. The established algorithm of mining large itemsets mainly makes use of the thought of layer matching, it is divided to build up a kind of layering structure, and the data are divided finally. Each independent calculating process relatively related to the generating algorithm of the whole data mining large itemsets is separated, and different match algorithms are adopted to build local large itemsets. Result shows that the introduced algorithm has better mining speed and performance than Apriori algorithm by numerical cases, the number of systematic I/O is reduced greatly, the memory expenses are moderate. and it can avoid missed mining effectively.

No abstract