Browse Prior Art Database

DIMENSIONS OF REPRESENTATION

IP.com Disclosure Number: IPCOM000128901D
Original Publication Date: 1975-Dec-31
Included in the Prior Art Database: 2005-Sep-20
Document File: 23 page(s) / 73K

Publishing Venue

Software Patent Institute

Related People

DANIEL G. BOBROW: AUTHOR [+3]

Abstract

Workers in cognitive science have worried about what people know, and how to represent such knowledge within a theory.1 Psychologists such as Paivio (1974) and Pylyshyn (1973) have argued, for example, over t~ro alternative forms for visual memory in humans. llle style of their arguments, which we return to at the ent of this paper, is to set up opposing characterizations and to argue about which one has more ";natural. properties with respect to observed phenomena.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 5% of the total text.

Page 1 of 23

THIS DOCUMENT IS AN APPROXIMATE REPRESENTATION OF THE ORIGINAL.

©; Copyright 1975 by Xerox Corporation; used with permission

DIMENSIONS OF REPRESENTATION

By DANIEL G. BOBROW

CSL 75-5 JULY 1975

A set of questions is presented concerning representations of knowledge. The questions are organized in terms of a framework in which knowledge of a world-state is derived by mapping the world to a knowledge base. The dimensions of representation are defined in terms of design issues which must be faced or finessed in any representation. Issues considered include correspondence between operations on the world and on the knowledge base, organization of the mapping process, inference which would make explicit knowledge which would otherwise be implicit, philosophy and mechanisms of access to elements of the knowledge base, pattern matching in knowledge processing, types of self-awareness an understander system might have, and use of multiple representations. These dimensions are illustrated with examples from the literature. The differences between analogical and propositional representations are illuminated by consideration along the multiple dimensions of representation.

Key Words and Phrases: artificial intelligence, representation, knowledge theory, understander systems

CR Categories:

3.6,3.36,3.42

XEROX PALO ALTO RESEARCH CENTER 3333 COYOTE HILL ROAD / PALO ALTO / CALIFORNIA 94304

TABLE OF CONTENTS

I. Introduction.....1 A. Representation and mapping.....1 B. Three simple visual
representations.....4

II. Domain and Range.....6 A. Units and relations.....6 B. Exhaustiveness.....7 C. Verbal
mediation.....7

III. Operational Correspondence.....10 A. Updating and consistency.....10 B. History and
planning.....11 C. Continuity.....12 D. Psychological modeling.....13

IV. The Mapping Process.....13 A. Constraints on world states.....13 B. Procedural declarative
tradeoffs.....15

V. Inference.....16 A. Formal inference techniques.....16 B. Computational inference.....18 C.
Meta-inferential techniques.....l9 D. Preferred inferences.....21

VI. Access.....22 A. Philosophy of association.....22 B. Access mechanisms.....23

VII. Matching.....25 A. Uses for matching.....25 B. Forms of matching.....26

Dec 31, 1975

Page 1

Page 2 of 23

DIMENSIONS OF REPRESENTATION

VIII. Self-Awareness.....28 A. Knowledge about facts.....28 B. Knowledge about process.....29

IX. Conclusion.....29 A. Multiple representations.....29 B. Analog representations.....31

References.....34

I. INTRODUCTION

Workers in cognitive science have worried about what people know, and how to represent such knowledge within a theory.1 Psychologists such as Paivio (1974) and Pylyshyn (1973) have argued, for example, over t~ro alternative forms for visual memory in humans. llle style of their arguments, which we return to at the ent of this paper, is to set up opposing characterizations and to argue about which one has more "natural. properties with respect to observed phenomena.

I claim that a more appr...