According to the preface, this book is intended as an introduction to expert systems for the widest possible audience. I am afraid this goal is too ambitious to be realized. The book is not suitable for a computer novice on her or his own because it assumes too much background knowledge about computing, and I do not think it is suitable as the sole text in an expert systems course because it is much too skimpy on both the theoretical and the experimental background of current expert systems technology. It is a readable guide to the use of a particular expert systems shell, VP-Expert. If used with supplementary material by a knowledgeable instructor, it could be a useful text in an expert systems course. It would also be a reasonable introduction to expert systems methods for an experienced professional working alone.
Writing for novices is difficult because so much needs explanation. The authors explain the ENTER key, for example, but not the terms “floppy disk” and “hard disk” (chapter 2). The first page of chapter 3 uses the terms “variable” and “knowledge base variable” without explanation. They will be perfectly intelligible in context to a programmer but very possibly not to a novice. A source of confusion for most readers, not just novices, is a tendency to use terms and concepts before explaining them. The authors give an example of a rule before explaining the format (p. 29); they discuss how to edit induction tables before explaining what an induction table is. Sometimes the desire to remain accessible to novices has made discussions needlessly opaque for the computer sophisticated. In discussing the syntax of rules, for example, the reader must deduce the fact that the “IF” part must be in conjunctive normal form.
Most of the examples in the first part of the book are very good. (Requiring the student to type in the whole of a knowledge base printed in the book in order to explain the editor seems to be a bit of overkill, though.) The examples in the final chapters are a bit skimpy, however. Working out a more complicated case with confidence factors would have been helpful. Examples of the use of external databases and spreadsheets could also have been expanded.
The book seems to be quite free of typos and similar flaws. There are a few inconsistencies. On page 29, the authors state that the only character not permitted in a DISPLAY statement is a double quote, but on page 42, curly brackets and the tilde are also excluded. On page 40, the discussion of the display does not agree with the example display shown. Because the book is meant to be used in hands-on mode, a reader should see easily enough from the computer’s behavior what is really meant.
The preface states that VP-Expert runs on IBM PCs or compatibles with 512K of memory and at least two floppy drives. By experiment, it does not work on an IBM PS/2 (unsurprisingly) nor on a number of PC configurations with monochrome monitors; for example, an IBM XT with an ATI EGAWONDER graphics card or an IBM PC Convertible with an attached monochrome monitor. The difficulty with these configurations is that some of the color to black-and-white mapping results in invisible displays. The program did run on a PC Convertible using the LCD monochrome screen. (The disks are not copy-protected, so one can copy them onto 31-2--inch media for use on laptops.) Of course, software problems are not the fault of the authors of the book.
In summary, the book, with its software, provides a reasonable practical introduction to the backward chaining, rule-oriented expert systems paradigm as exemplified by the VP-Expert shell.
