Developing knowledge-based systems: reorganizing the system development life cycle

Reviewer: Jane Fedorowicz

The authors describe a case study in which they were involved: the development of a knowledge-based system at Blue Cross/Blue Shield of South Carolina. They present a brief overview of the systems development life cycle (SDLC) they followed and a description of its use at Blue Cross/Blue Shield. The proposed SDLC is reasonable for the knowledge-based environment. Whether it is the best practical strategy to employ for these systems remains to be seen. Because of the diversity of the types of applications and technologies used in this fast-changing field, this methodology may be only one of many that we find in practice. The authors ask four questions at the beginning of the paper and answer them at the end: Can knowledge engineering be done by average people__?__ Do we need high-priced software engineers__?__ Do we need new software__?__ Do we need a new system development philosophy__?__ (Their answers are yes, no, yes, and yes.) The arguments supporting these answers are fairly cursory, and could have been much stronger. In fact, the level of technical knowledge needed to read this paper surpasses that of the typical systems analyst, the person whom the authors suggest to replace the knowledge engineer in the process. The authors could have fairly easily included more discussion of the knowledge acquisition problems analysts face, at the expense of some of the technical issues that they present only briefly. On the other hand, those with a good foundation in the knowledge-based systems area will be frustrated at the brevity of the case study; an expanded version could be very useful for current and future knowledge-based systems analysts. Finally, the acronym SDLC (“Systems Development Life Cycle”) is expanded incorrectly in the opening paragraph as “Synchronous Data Link Controls.” It is later used correctly in the text.