A knowledge-based expert-system - development paradigm with empirical validation in the human services sector

Frank Stuart Wells III, Louisiana Tech University


This study (1) presents a paradigm, based on structured systems methodology, for the development of knowledge-based expert systems, and (2) validates the resultant methodology through application in the case management of the developmentally disabled. A procedure is described whereby use of modular, top-down analysis and bottom-up construction techniques facilitate the interchange between and among the knowledge engineer, the expert, management, and the user leading to prototype development. The resultant model is robust, flexible, and maintainable. Expansion of the prototype, both in breadth and depth, follows naturally using the same open architecture design.

Chapter 1 provides problem background and a review of current research in the area of consultative knowledge-based expert system development. Chapter 2 describes the development paradigm in detail, comparing and contrasting it with the tools, techniques, and processes employed in structured systems analysis, design, and development.

Chapters 3 and 4 present the application used during validation of the development model along with test results. The application, called CARE, for Client And Resource Evaluation, aids in the case management process by providing expert advice such as would be attainable from an experienced case manager. The system analyzes client demographic and clinical data to (1) determine eligibility for assistance, (2) recommend needed evaluations, (3) offer advice on optimal client services, and (4) match recommended services with available community resources. System output includes reasoning behind advice as well as the advice itself.

Although developed in the human services field, the development paradigm addresses problems common to the commercial business environment. The resultant methodology provides a guide for the creation of knowledge-based expert systems by employing and adapting the proven techniques of structured systems analysis, design, and development to the realm of knowledge engineering.