Constraint satisfaction problems have been extensively studied by researchers in both the operations research (OR) and artificial intelligence (AI) areas. The research aimed at integrating the two approaches so that some of their limitations can be removed is described. Specifically, a knowledge-based system that formulates and maintains OR models for manufacturing planning purposes is presented. Domain-specific knowledge allows synthesis of various qualitative relationships into mathematical relationships and its identification of various dependencies between symbolic and mathematical models. The modeler component engages in a search process to identify the simplest model that can be formulated. A truth maintenance system, specifically designed to support modeling for planning, allows the user to explore various scenarios to arrive at an appropriate plan. Preliminary experiments indicate that human planners are able to formulate models that are equivalent to those formulated by experienced OR modelers for various planning problems.
展开▼
