Efficient Management of Non Redundant Rules in Large Pattern Bases: a Bitmap Approach

François Jacquenet (1) , Christine Largeron (1) , Cédric Udréa (1)

1 LHC - Laboratoire Hubert Curien

Knowledge Discovery from Databases has more and more impact nowadays and various tools are now available to extract efficiently (in time and memory space) some knowledge from huge databases. Nevertheless, those systems generally produce some large pattern bases and then the management of these one rapidly becomes untractable. Few works have focused on pattern base management systems and researches on that domain are really new. This paper comes within that context, dealing with a particular class of patterns that is association rules. More precisely, we present the way we have efficiently implemented the search for non redundant rules thanks to a representation of rules in the form of bitmap arrays. Some experiments show that the use of this technique increases dramatically the gain in time and space, allowing us to manage large pattern bases.

Domaines: Intelligence artificielle [cs.AI]

François Jacquenet, Christine Largeron, Cédric Udréa. Efficient Management of Non Redundant Rules in Large Pattern Bases: a Bitmap Approach. 2006, pp.208-215. ⟨hal-00117475⟩