Generating optimization-based decision support systems

This paper discusses the implementation of optimization based DSSs. An approach is proposed that will enable OR/MS analysts to develop this kind of system much more efficiently than is possible today. The aim is to develop systems with modern GUIs, that interact with DBMSs, and that can be built with little programming effort. Many of the tools required for this approach were developed to support SML, which was chosen to represent the models because it has ample expressive power, completely specified syntax and semantics, total structure/data independence, and lends itself to the surface/deep structure distinction, which makes the approach feasible. The approach is being implemented in a research project funded by the Chilean Government and several firms and is being tested on problems taken from these firms.<<ETX>>