Customer Reviews

Knowledge Representation: Logical, Philosophical, and Computational Foundations

5 star:		(1)
4 star:		(1)
3 star:		(1)
2 star:		(0)
1 star:		(0)

 

 

In this book, John Sowa conveys diverse and effective insights within the field of knowledge representation (KR). The frameworks he employs are deeply grounded philosphically. (Sowa's previous work on conceptual structures reactivated and extended the innovative work of Charles Sanders Peirce, which integrates logic and graph theory.) The work reviewed here surveys a wide range of KR issues from basic ontology to agency and processes. Chapter 6, Knowledge Soup, is widely recognized for framing and addressing some of the more demanding, and largely unresolved, challenges in the field.

Throughout the book, issues are explored in a coherent, readable way. Of course, KR implies the use of relevant formalisms, and readers with some background in AI research will be better prepared to absorb the book's insights. However, for students and scholars looking for an integrated overview, Sowa makes a unique contribution.

There is real conceptual meat in this book, it is obviously the fruit of decades of thinking on this vey hard topic.
My only qualm is that it may be a little outdated by now. As far as I know there are better solutions to the Frames problem than the one outlined in the book.
But as a place to start it is about as good as any for the time being, roughly on par with Russel and Norvig
though more detailed on some aspects.
Here is the table of contents, which I OCR'd.
Preface XI
CHAPTER ONE
Logic
1.1 Historical Background
1.2 Representing Knowledge in Logic
1.3 Varieties of Logic
1.4 Names, Types, and Measures
1.5 Unity Amidst Diversity
CHAPTER TWO
Ontology
2.1 Ontological Categories
2.2 Philosophical Background
2.3 Top-Level Categories
2.4 Describing Physical Entities
2.5 Defining Abstractions
2.6 Sets, Collections, Types, and Categories

CHAPTER THREE
Knowledge Representations
3.1 Knowledge Engineering
3.2 Representing Structure in Frames
3.3 Rules and Data
3.4 Object-Oriented Systems
3.5 Natural Language Semantics
3.6 Levels of Representation
CHAPTER FOUR
Processes
4.1 Times, Events, and Situations
4.2 Classification of Processes
4.3 Procedures, Processes, and Histories
4.4 Concurrent Processes
4.5 Computation
4.6 Constraint Satisfaction
4.7 Change
CHAPTER FIVE
Purposes, Contexts, and Agents
5.1 Purpose
5.2 Syntax of Contexts
5.3 Semantics of Contexts
5.4 First-Order Reasoning in Contexts
5.6 Encapsulating Objects in Contexts 5.7 Agents

CHAPTER SIX
Knowledge Soup
6.1 Vagueness, Uncertainty, Randomness, and Ignorance
6.2 Limitations of Logic
6.3 Fuzzy Logic
6.4 Nonmonotonic Logic
6.5 Theories, Models, and the World
6.6 Semiotics

CHAPTER SEVEN
Knowledge Acquisition and Sharing
7.1 Sharing Ontologies
7.2 Conceptual Schema
7.3 Accommodating Multiple Paradigms
7.4 Relating Different Knowledge Representations
7.5 Language Patterns
7.6 Tools for Knowledge Acquisition

APPENDIX A
Summary of Notations
A.1 Predicate Calculus
A.2 Conceptual Graphs

APPENDIX B
B.1 Principles of Ontology
B.2 Top-Level Categories
B.3 Role and Relation Types
B.4 Thematic Roles
B.5 Placement of the Thematic Roles
APPENDIX C
Extended Example
C.1 Hotel Reservation System
C.2 Library Database
C.3 ACE Vocabulary
C.4 Translating ACE to Logic

Answers to Selected Exercises
Bibliography
Name Index
Subject Index
Special Symbols

I thought the first three chapters did an excellent job of covering advances in knowledge representation. However chapter four is marred by an attempt to present what appears to be virtually every syntax used relating to processes. Somewhere in this gulf of complexity I think he has some basic concepts, but they are hard to reach. It's equivalent to reading a book on algorithms in which the author presents the algorithms in C, Cobol, Fortran, Basic, SAS, etc. Why not just present the concepts within the context of a MINIMUM of syntax? Still the book is worth reading and has good appendixes.