View article

In spite of what it says in the preface, this book is not for readers with a “general interest in expert systems.” The audience that will appreciate the book the most are expert systems researchers who believe in qualitative reasoning, have a detailed knowledge of Prolog, and preferably have a knack for matters of the human heart (such as cardiac arrhythmias). It is hard for someone without thorough knowledge of these areas to follow the reasoning in the book. The body of the work contains at least 100 pages of code and medical charts, and 60 pages more appear in the appendices; the latter are not well documented. The authors contend that the key to viable expert systems lies in the “deep” or qualitative knowledge that lets these systems reason from “first principles,” the building blocks of a domain. While first-generation expert systems (such as Mycin) relied on surface knowledge—the precompiled judgments …