A Temporal Constraint Logic Model and Its Applications

Abstract

Temporal knowledge representation and reasoning are important for solving planning, scheduling, discourse understanding and other problems. Many temporal logic models have been proposed for this purpose, but most of them have some disadvantages such as insufficient representability or requirement of a huge amount of time to maintain consistency of represented temporal knowledge. In this paper, we propose a temporal constraint logic model (TCLM). This model,in which representation and consistency maintenance are unified, uses temporal constraint representation. The temporal constraint representation is constructed with event-interval relation, interval constraints, and time-point constraints. Using this representation, both uncertain knowledge and exact knowledge about time can be represented in compact and computationally efficient forms. Consistency in the time knowledge is represented as a network, and maintained using a constraint propagation algorithm. The network consists of nodes which imply time points (corresponding to both ends of intenval) and arcs which imply constraints on time points. The temporal knowledge is dynamically translated into time point constraints and the network is constructed. Whenever new knowledge is added and the network is reconstructed, the constraint propagation algorithm is used to keep consistency. Based on TCLM, a temporal reasoning system (PERIOD) is implemented using ESP. In order to confirm the ability of TCLM, some scheduling and planning problems are solved by PERIOD system.