An instance of the clp(x) scheme which allows to deal with temporal reasoning problems

In many applications oftemporal reasoning is necessary to express metric and symbolic temporal constraints among temporal objects whether they are points or intervals. In order to cope with these requirements different formalisms have been issued., those that allow to express symbolic temporal constraints by one hand, and others involvingmetric temporal constraints. AJthough this formalism are suitable to represent just sorne kind of problems, in many cases, it is necessary to handle and represent in the same framework both metric and symbolic constraints among temporal objects, whether they are point or interval. . Starting from the previous schemes, different formalisms to integrate metric and symbolic temporal constraints have been issued. A common limitation of these proposals is that none of them allows to represent disjunctive constraints involving a metric component and a symbolic one. This type of eonstraints arises for example in scheduling problems, where an activity must be performed beforé or after another activity, but considering the setting time of the used resources [lbáñez,92b]. Besides in.many planning applications, the formulation ofthe problem itself, must be expressed as logic formulas with a periQd of time associated. Therefore, a temporal reasoning system oriented to planning should be able to express both the logic and the temporal part in a same frame. Unfortunately, none of the approaches to integrate symbolic and tnetric temporal constraints allows to express the logic part of the problem. The main aim of this paper is to de.fine a temporal tool which allows to express and unify metric and symbolic temporal constraints among temporal objects (intervals and points). The temporal model proposed in this paper is based on intervals. However, as opposed to other formalisms, the duration of the intervals may be zero, and therefore temporal points are incIuded. In other words, the concept of temporal interval used in the literature (where the duration is strictly greater than zero), is generalized. Starting from the temporal model, a new operational framework oriented to the resolution oí the problems rather than focused to the representation oftemporal reasoning problems is defined. TIte proposed . operational frarn.ework was designed as a new instance oí the CLP(X) scheme [lbáñez,93] in which the computational domain is formed from temporal objects. Conceptually, the variables of the CLP(Temp) language have associated a finite set of pairs of value5 representing temporal intervals. Eje: 3er Workshop sobre Aspectos teóricos de la inteligencia artificial Red de Universidades con Carreras en Informática (RedUNCI)