Amzi! inc.--Tools for Building Expert Systems in Prolog
==============================================================================

Expert Systems in Prolog is a book and disk combination.  The book is by 
Dennis Merritt, published by Springer-Verlag.  The disk is available direct 
from Amzi! inc.

Expert Systems in Prolog (XSiP) shows how to build a number of prototype 
expert system shells, describing the technical issues and design decisions 
involved in each.  Sample expert systems are implemented using each of the 
shells.  Full source code for the shells is available on an unrestricted 
basis.

Chapter summaries:

1- Introduction--Introduces expert system shells and the features that are 
found in them.

2- Using Prolog's Inference Engine--Shows how to use Prolog's built-in 
inference engine as the basis for a simple expert system shell.  It adds a 
user interface that is more expert-system-like than Prolog-like.  Use of the 
shell is illustrated with a bird identification expert system.

3- Backward Chaining with Uncertainty--Shows how to build an EMYCIN-like 
inference engine with certainty factors.  It adds a user interface and a 
knowledge representation language using English-like syntax implemented with 
Prolog's DCG support.  Use of the shell is illustrated with a car diagnostic 
expert system.

4- Explanation--Adds "how" and "why" explanation capabilities to both the 
EMYCIN-like shell and the native Prolog shell.  These explain how a conclusion 
was reached, or why a particular chain of questions is being asked.

5- Forward Chaining--Shows how to implement a basic OPS-like forward chaining 
(production) system.  It uses Prolog terms for knowledge representation and 
includes the OPS conflict resolution strategies.  The use of the shell is 
illustrated with an expert system that configures living room furniture.

6- Frames--Adds a frame-based (often called objects by AI vendors) knowledge 
representation scheme with inheritance and demons.  The power of the frames is 
illustrated by using them to re-implement the bird identification expert 
system without rules.

7- Integration--Builds an integrated system that uses the forward-chaining 
engine, frame-based knowledge representation, and native Prolog.  The three 
together form a powerful tool set for building a large variety of practical 
expert systems.  The concept is illustrated by re-implementing the furniture 
configuration expert system with a much cleaner knowledge representation.

8- Performance--Shows how to use indexing schemes to speed up backward-
chaining systems, and how to implement a Rete network for improved performance 
in forward-chaining systems.

9- User Interface--Describes an object-oriented windowing user interface 
developed in Prolog.  It allows menu and window interfaces to be added to the 
various other programs described in the book.  It uses a full message passing 
and hierarchical structure for describing and communicating with the windows.

10- Two Hybrids--Describes two actual cases studies of expert systems built to 
install and tune a mainframe database system.  Both show the benefits of 
custom building expert systems, as the knowledge representation language of 
the systems is designed to map closely to the way the "experts" describe the 
knowledge.  The inference engine is then customized to work using the 
customized knowledge representation.

11- Prototyping--Shows how to build "quick-and-dirty" prototypes of expert 
systems for proof-of-concept work, again by modeling the knowledge 
representation first.  The example used is a sales advisor that helps a sales 
person position a complex technical product for a prospect.

12- Rubik's Cube--For fun, presents a working expert system that will help you 
unscramble that old Rubik's cube lying on your desk.

Appendices--Full source code for all of the expert-system shells described in 
the book.

Optional XSiP disk--Includes running versions of all of the prototypes, all in 
Edinburgh syntax.  They were developed with a variety of Prologs, although 
most of the implementations are in Cogent Prolog.  With minor modifications 
(usually in the I/O area) they can be ported to any other Prolog.  The disk 
also includes an additional expert system for filling out U.S. tax forms along 
with an article describing how it was built.

Building Expert Systems in Prolog, Dennis Merritt, Springer-Verlag ISBN 0-387-
97016-9 ISBN 3-540-97016-9.  Book is available either direct from Amzi!, or 
your local technical book store.  Disk is available from Amzi!

"I was very impressed.  I found the shells described in your book very useful 
for some of the work I'm trying to accomplish."--a reader's comment