Integrating Logical Inference with Numeric Optimization
Ho Geun Lee, Ronald M. Lee · Intelligent systems in accounting, finance and management/Intelligent systems in accounting, finance & management · 1994
Abstract OR and AI techniques have progressed separately without adequate interactions although they can benefit from each other's complementary strengths: OR in efficient mathematical computation and AI in domain‐specific knowledge representation. Constraint Logic Programming (CLP) is introduced as a problem‐ solving tool which combines these complementary strengths. CLP uses predicate logic for knowledge representation, thus providing stronger expressive powers than prepositional logic, which is employed by mathematical programming. In addition, CLP allows numeric optimization with a computational efficiency comparable to OR approaches. This integration of logical inference and mathematical computation fits well into a special class of decision problems that requires not only logical inference on domain‐specific knowledge but also numeric optimization through mathematical formulation. A production planning problem is introduced and solved by CLP. The results are compared with an implementation by mixed integer programming to show the advantages of the integration of OR and AI approaches over either one used alone.