A conceptual architecture for generalized decision support system software

Abstract

A conceptual architecture is proposed, and a prototype system that validates the proposed architecture's technical feasibility is presented. The relational data model is used in the proposed approach, and the way in which its tabular structure may be exploited to provide a unified logical view of data handling and model management is shown. Also, the concept of macro is used to extend the conventional meaning of model to include composite queries that often arise in a DSS environment. A primitive knowledge base consisting of macros, data, and associated dictionaries for both are included in the architecture. Organized into a three-level hierarchy, the knowledge base consists of systemwide, group, and individual macros and data. Also included in the architecture as a language system and modules for monitoring system performance, maintaining data and macro security, and linking the DSS to other parts of the local computing environment. A prototype system known as Magic/Roc is described.