A natural language discourse model to explain linear programming models and solutions

摘要

This paper presents a natural language (i.e., English) discourse model to explain a linear programming model and, possibly, its computed solution (which need not be optimal). Drawing from earlier syntactic translations, a semantic model is presented, based on economic input-output relations commonly found in large-scale linear programs. These are then shown how to form rule-driven explanations, which comprises a fundamental component of an intelligent mathematical programming system.