Mining Minimal Non-redundant Association Rules Using Frequent Closed Itemsets

摘要: The problem of the relevance and usefulness extracted association rules is primary importance because, in majority cases, real-life databases lead to several thousands with high confidence among which are many redundancies. Using closure Galois connection, we define two new bases for union a generating set all valid support confidence. These characterized using frequent closed itemsets their generators; they consist nonredundant exact approximate having minimal antecedents maximal consequents, i.e. most relevant rules. Algorithms extracting these presented results experiments carried out on show that proposed useful, generation not time consuming.