Knowledge-Based Expert Systems

Reviewer: Joseph L. Podolsky

This article is an excellent introduction to and review of expert systems by a pioneer and leader in the field. Hayes-Roth provides the viewpoint of a commercial practitioner and an entrepreneur, which is built on a solid academic background. Hayes-Roth provides a good, albeit long, definition of expert systems and then places the systems into historical perspective. He concisely lists the types of problems that seem appropriate for expert systems and shows a very general methodology for applying “knowledge engineering” to those problems. Both the state-of-the-art and the state-of-commercialization of the art are placed in time frames. Distinctions are made between the design issues (knowledge engineering) and the construction technologies needed to realize the design (symbolic programming). The article provides us with an interesting diagram showing both the “demand pull” and “technology push” forces over time. Hayes-Roth makes the interesting point that knowledge engineering converts “inactive” (passive) knowledge found in books and the private, fragile knowledge held by experts into active, “inspectable” electronic form. Because of this, knowledge engineering will have direct impact on those industries based on “knowledge transfer.” These industries include the obvious ones, such as publishing, libraries, and schools, but also included are such professional knowledge industries as law and accounting. Interestingly, Hayes-Roth omits mention of computer application design and programming as another knowledge transfer industry likely to be affected by knowledge engineering. Yet the data processing industry may be one of the highest-leverage areas in which this technology can be applied. This is especially true because Hayes-Roth predicts that knowledge systems and conventional EDP will converge in the future. Thus, applying knowledge engineering to EDP will recursively help the progress of knowledge engineering itself. Hayes-Roth also predicts that, on the road to commercialization of this technology, we will see lots of fads and “speculative products.” Reading this article might make us all better consumers of knowledge engineering and expert systems; more importantly, this gentle introduction might give us a base on which to build our own knowledge and enable us to find ways of moving our own professions more comfortably into this aspect of the future.