A Breadth-First Search Algorithm for Mining Generalized Frequent Itemsets Based on Set Enumeration Tree

Mining generalized association rules is one of important research area in data mining. If we use the traditional methods, it will meet two basic problems, the first is low efficiency in generating generalized frequent itemsets with the items and levels of taxonomy increasing, and the second is that too much redundant itemsets´ support are counted. This paper proposes an improved Breadth-First Search method to mine generalized association rules. The experiments on the real-life data show that our method outperforms the well-known and recent algorithms greatly.