Synthesizing algorithm for mining composite-frequent item sets

It is very important to get the frequent item set in the associate rule mining. In order to fast obtain the frequent item set from a database that includes multiple values, the definition of transaction database was extended. By the tree concept, a special tree was built in which every node is formed by item and item's count. On the foundation of apriori algorithm and artificial intelligent search, FABCTA (fast algorithm by candidate transaction tree and apriori) was presented to solve the frequent item set in small branches of tree. By the test on real data, FABCTA is more efficient than apriori algorithm.