Mining bases for association rules using closed sets

Rafik Taouil (1) , Nicolas Pasquier (1) , Yves Bastide (1) , Lotfi Lakhal (1)

1 LIMOS - Laboratoire d'Informatique, de Modélisation et d'optimisation des Systèmes

Association rules are conditional implications between requent itemsets. The problem of the usefulness and the elevance of the set of discovered association rules is related to the huge number of rules extracted and the presence of many redundancies among these rules for many datasets. We address this important problem using the Galois connection framework and we show that we can generate bases or association rules using the frequent closed itemsets extracted by the Close or the A-Close algorithms.

Domaines: Algorithme et structure de données [cs.DS]

Rafik Taouil, Nicolas Pasquier, Yves Bastide, Lotfi Lakhal. Mining bases for association rules using closed sets. ICDE'2000 International Conference, Feb 2000, San Diego, United States. pp.307. ⟨hal-00467758⟩