Building an expert system involves eliciting, analyzing, and interpreting the knowledge that a human expert uses when solving problems. Expe rience has shown that this process of "knowledge acquisition" is both difficult and time consuming and is often a major bottleneck in the production of expert systems. Unfortunately, an adequate theoretical basis for knowledge acquisition has not yet been established. This re quires a classification of knowledge domains and problem-solving tasks and an improved understanding of the relationship between knowledge structures in human and machine. In the meantime, expert system builders need access to information about the techniques currently being employed and their effectiveness in different applications. The aim of this book, therefore, is to draw on the experience of AI scientists, cognitive psychologists, and knowledge engineers in discussing particular acquisition techniques and providing practical advice on their application. Each chapter provides a detailed description of a particular technique or methodology applied within a selected task domain. The relative strengths and weaknesses of the tech nique are summarized at the end of each chapter with some suggested guidelines for its use. We hope that this book will not only serve as a practical handbook for expert system builders, but also be of interest to AI and cognitive scientists who are seeking to develop a theory of knowledge acquisition for expert systems.

In June of 1983, our expert systems research group at Carnegie Mellon University began to work actively on automating knowledge acquisition for expert systems. In the last five years, we have developed several tools under the pressure and influence of building expert systems for business and industry. These tools include the five described in chapters 2 through 6 - MORE, MOLE, SALT, KNACK and SIZZLE. One experiment, conducted jointly by developers at Digital Equipment Corporation, the Soar research group at Carnegie Mellon, and members of our group, explored automation of knowledge acquisition and code development for XCON (also known as R1), a production-level expert system for configuring DEC computer systems. This work influenced the development of RIME, a programming methodology developed at Digital which is the subject of chapter 7. This book describes the principles that guided our work, looks in detail at the design and operation of each tool or methodology, and reports some lessons learned from the enterprise. of the work, brought out in the introductory chapter, is A common theme that much power can be gained by understanding the roles that domain knowledge plays in problem solving. Each tool can exploit such an understanding because it focuses on a well defined problem-solving method used by the expert systems it builds. Each tool chapter describes the basic problem-solving method assumed by the tool and the leverage provided by committing to the method.

Readings in Knowledge Acquisition and Learning collects the best of the artificial intelligence literature from the fields of machine learning and knowledge acquisition. This book brings together the perspectives on constructing knowledge-based systems from these two historically separate subfields of artificial intelligence.

The Knowledge Level In Expert Systems: Conversations and Commentary deals with artificial intelligence, cognitive science, qualitative models, problem solving architectures, construction of knowledge bases, machine learning integration, knowledge sharing or reusability, and mapping problem-solving methods. The book tackles two opposing dogmas: first, that control is generic so is in the inference engine; and two, deep and surface knowledge are different so deep knowledge belongs in a performance system. The text also explains how to use SPARK, a selection method, in approaching the task features that can be used to select or construct the problem-solving method suitable for the task. An alternative method to SPARK starts with an analysis of the domain model and a classification using primitive inference steps. The book also adds that expert problem solving is a form of qualitative modeling that connects other expert systems and engineering. The text then describes very large knowledge bases, particularly, the volume of which knowledge bases can be integrated with expert systems, coherence maintenance, and use/neutral representation of knowledge. Task analysis and method selection focuses on SPARK; how theories about the relation between task features and expert system solutions can be empirically validated. The book also enumerates the benefits and limitations of a generic task approach, and how various modules with their specific internal architectures can be integrated. Programmers, computer engineers, computer technicians, and computer instructors dealing with many aspects of computers such as programming, networking, engineering or design will find the book highly useful.

This book presents a practical view of the knowledge acquisition process, its methodologies and techniques, in order to enable readers to develop expert systems knowledge bases more effectively. It strikes a balance between presenting (1) summaries of research in the field of knowledge acquisition and (2) methodologies and techniques that have been applied and tested on numerous programs in various contexts. Written for novice knowledge engineers or others tasked with acquiring knowledge for the systematic development of expert systems. The presentation of the material does not presume a background in either computer science or artificial intelligence.