STR et compression de contraintes tables

Nebras Gharbi (1) , Fred Hemery (1) , Christophe Lecoutre (1) , Olivier Roussel (1)

1 CRIL - Centre de Recherche en Informatique de Lens

Over the recent years, many ltering algorithms have been developed for table constraints. STR2, one of the most e cient algorithms, is based on the technique of simple tabular reduction, meaning that it maintains dynamically the list of supports in each constraint table during inference and search. However, for some speci c problems, the approach that consists in representing in a compact way tables by means of multi-valued deci- sion diagrams (MDD) overcomes STR2. In this paper, we study the possibility of combining simple tabular reduction with a compression form of tables based on the detection of recurrent patterns in tuples.

Domaines: Intelligence artificielle [cs.AI]

Nebras Gharbi, Fred Hemery, Christophe Lecoutre, Olivier Roussel. STR et compression de contraintes tables. 9ièmes Journées Francophones de Programmation par Contraintes (JFPC'13), 2013, Aix-en-Provence, France. pp.143-146. ⟨hal-00869930⟩