Symbolic optimization in an object-oriented database using constraints

Departing from pure Datalog and simple constraints arising from the logical objects, a underling theory can be obtained which permits to study certain aspects of optimization. A logical access language for objects, called U-LOG 3, has been conceived and implemented along these lines. U-LOG 3 can performed two kinds of optimization: a symbolic optimization and an optimization of physical access to objects. The first one is performed through a confluent rewriting system and which consists upon the reduction of the sorts involved in the logical objects, exploiting the information contained in the signature. In this paper, we briefly discuss U-Log 3 underlying theory and symbolic optimization, after motivating our switch from feature logic to Datalog + Constraints.