Closed sets based discovery of small covers for association rules
Résumé
In this paper, we address the problem of the understandability and usefulness of the set of discovered association rules. This problem is important since real-life databases lead most of the time to several thousands of rules with high confidence. We thus propose new algorithms based on the Galois closed sets to limit the extraction to small informative covers for exact and approximate rules, and small structural covers for approximate rules. Once frequent closed itemsets - which constitute a generating set for both frequent itemsets and association rules - have been discovered, no additional database pass is needed to derive these covers. Experiments conducted on real-life databases show that these algorithms are efficient and valuable in practice.
Fichier principal
Closed_Set_Based_Discovery_of_Small_Covers_for_Association_Rules_Pasquier_et_al._BDA_1999.pdf (300.61 Ko)
Télécharger le fichier
Origine | Fichiers produits par l'(les) auteur(s) |
---|