An intelligent system for formulating linear programs

摘要

The research and system development work described in this paper is aimed at overcoming some of the problems associated with the development of large, complex linear programming problems. The most overwhelming problem is that of size. It is not uncommon for large planning and policy analysis problems to have tens of thousands of constraints and activities. Matrix generator systems have been designed to help in this process. However, the amount of manual labor involved is still very great and the formulation process is subject to errors which are difficult to detect. We provide an overview of a system which uses artificial intelligence and database techniques to help a knowledgeable user formulate large linear programs. The system automates many of the tedious processes associated with large-scale modeling and provides a top-down development environment with a number of different forms of problem representation.