A new closure algorithm based in logic: SLFD-Closure versus classical closures
- Autores: Ángel Mora Bonilla, Gabriel Aguilera Venegas, Manuel Enciso García-Oliveros, Pablo Cordero Ortega, Inmaculada Pérez de Guzmán Molina
- Localización: Inteligencia artificial: Revista Iberoamericana de Inteligencia Artificial, ISSN-e 1988-3064, ISSN 1137-3601, Vol. 10, Nº. 31, 2006, págs. 31-40
- Idioma: español
- Enlaces
-
Resumen
The field of application of closure systems goes from theoretical areas as algebra or geometry to practical areas as databases and artificial intelligence. In these practical areas, a kind of constraint named functional dependencies have an important role. Given a set of attributes $X$ and a set of functional dependencies $\Gamma$, the computation of the closure of $X$ for $\Gamma$, denoted as $X^+$ is abundantly used in artificial intelligence and database literature and is one of the key points in many problems: knowledge compilation, redundant constraint elimination, query optimization, the finding key problem, etc. We outline the main classical closure algorithms and we compare them with a novel algorithm named SLFD-Closure. We show an empirical study with the execution of the closure algorithms, and we establish that SLFD-Closure is the fastest.
