A knowledge-based software development environment for scientific model-building

Abstract

The authors describe a knowledge-based software development environment for scientific modeling. The SIGMA system provides scientist-users with a high-level data flow language for specifying scientific models. This language is intended to be both easier to use and easier to understand than low-level scientific code. Thus, the language and the software tool should facilitate sharing among users. SIGMA makes use of an extensive store of background domain knowledge to ease the specification acquisition process and to maintain the consistency of an evolving specification. In addition, SIGMA uses its background knowledge to aid in the execution of specified models. The availability of background knowledge enables the system to automatically infer portions of the scientific model specification that normally would be tediously specified by the user. The approach is illustrated with examples from the scientific domain of planetary atmospheric modeling. >