Abstract: Most algorithms of frequent itemsets (or maximum frequent itemsets) do not consider any domain knowledge. As a result they generate many irrelevant patterns. Therefore, finding constrained maximum frequent itemsets is a key in important data mining application such as discovery of constrained association rules, constrained strong rules, etc. Little work has been done on this problem. This paper presents an effective algorithm for mining constrained maximum frequent itemsets and its update, UCMFIA, based on a novel frequent pattern tree (FP-tree) structure that is an extended prefix-tree structure for storing compressed and crucial information about frequent patterns. Experiments show that the algorithm is effective.

Key words:

association rules, item constraint, maximum frequent itemsets, frequent pattern tree, incremental updating