Formulation, design and implementation of the MAGIC/ROC decision support system generator

Decision Support Systems (DSS) , which may be briefly defined as man-computer systems for facilitating managers' decision-making processes in semi-structured and unstructured situations, have received much attention in the recent MIS literature. These systems use data and models to perform data extraction, data aggregation, data analysis, estimation, optimization, simulation, etc., to assist managers in decision making. Generalized software systems integrating these decision support and data management capabilities, which are known as DSS generators, are needed at the current stage of MIS/DSS development. No existing software system has been reported as a generalized, powerful and "friendly" system, which provides full-range capabilities for easily building specific DSS in any application area. The development of DSS generators is a complicated task. In order to cleverly integrate a variety of decision support and data management capabilities into a well-designed, orderly whole, a conceptual model must be created as a foundation for developing such software systems.

The objectives of this dissertation are: (1) to formulate a comprehensive conceptual model for developing DSS generators; (2) to design a prototype DSS generator based on the conceptual model t o demonstrate the validity of the model; and (3) to implement an experimental version (selected portions) of the prototype DSS generator to validate the design.

The conceptual model conveys a concept of "macro" to extend the conventional meaning of "model" in a CSS environment, and a concept of three - level knowledge bases to enhance organizational communication and sharing of corporate knowledge. Because of the structural similarity between matrices and relations, both of which represent data in tabular form, the conceptual model suggests the use of the relational data model to enhance the integration of the model base management subsystem (MBMS) and the data base management subsystem (DBMS). Components of the conceptual model are: • a model base management subsystem • a data base management subsystem • model/data dictionaries • the linkage between the MBMS and DBMS • the linkage between the DSS generator and the local computing environment • the skeleton of three-level knowledge bases • a security system • a performance monitor • a user-system interface • a language system

The top-down design approach was employed to design the MAGIC/ROC* Decision support system Generator based on the conceptual model. The rationale was to include all features present in the conceptual model in a software system which is designed to be a DSS generator in order to achieve an orderly system. MAGIC/ROC advances software technology employed in MIS/DSS fields and sets up a framework of DSS generators for subsequent development. The experimental version of HAGIC/ROC was implemented via an existing software system, Statistical Analysis System (SAS). This "workable" DSS generator validates the design of MAGIC/ROC and the conceptual model formulated for developing DSS generators.