Formal Ontology in Information Systems

187 Pages·2004·18.95 MB·English

by  Achille Varzi, Laure Vieu

Frontiers in Artificial Intelligence and
Formal Ontology in
Applications
Information Systems
FAIA covers all aspects of theoretical and applied artificial intelligence research in the form
of monographs, doctoral dissertations, textbooks, handbooks and proceedings volumes. The
FAIA series contains several sub-series, including "Information Modelling and Knowledge
Bases" and "Knowledge-Based Intelligent Engineering Systems". It also includes the
biannual ECAI, the European Conference on Artificial Intelligence, proceedings volumes,
Proceedings of the Third International Conference
and other ECCAI - the European Coordinating Committee on Artificial Intelligence -
sponsored publications. An editorial panel of internationally well-known scholars is (FOIS-2004)
appointed to provide a high quality selection.
Series Editors:
J. Breuker, R. Dieng, N. Guarino, R. Lopez de Mantaras, R. Mizoguchi, M. Musen
Edited by
Volume 114
Achille C. Varzi
Recently published in this series
Department of Philosophy, Columbia University, New York, USA
Vol. 113. J. Vitria et al. (Eds.), Recent Advances in Artificial Intelligence Research and Development
Vol. 112. W. Zhang and V. Sorge (Eds.), Distributed Constraint Problem Solving and Reasoning in Multi
and
Agent Systems
Vol. 111. H. Fujita and V. Gruhn (Eds.), New Trends in Software Methodologies, Tools and Techniques -
Proceedings of the Third SoMeT _ W04 Laure Vieu
Vol. 110. R. Lopez de Mantaras and L. Saitta (Eds.), ECAI 2004-Proceedings of the 16th European
Conference on Artificial Intelligence
Institut de Recherche en Informatique de Toulouse, CNRS, France, and
Vol. 109. E. Onaindia and S. Staab (Eds.), ST AIRS 2004 - Proceedings of the Second Starting AI
Researchers' Symposium Laboratory of Applied Ontology, ISTC-CNR, Trento, Italy
Vol. 108. V. Stefanuk and K. Kaijiri (Eds.), Knowledge-Based Software Engineering-Proceedings of the
Sixth Joint Conference on Knowledge-Based Software Engineering
Vol. 107. F. Darses et al. (Eds.), Cooperative Systems Design - Scenario-Based Design of Collaborative
Systems
Vol. 106. D. Bourcier (Ed.), Legal Knowledge and Information Systems - JURIX 2003: The Sixteenth
Annual Conference
Vol. 105. Y. Kiyoki et al. (Eds.), Information Modelling and Knowledge Bases XV
Vol. 104. A. Abraham et al. (Eds.), Design and Application of Hybrid Intelligent Systems
Vol. 103. B. Tessem et al. (Eds.), Eighth Scandinavian Conference on Artificial Intelligence - SCAI'03
Vol. 102. C. Turchetti, Stochastic Models of Neural Networks
Vol. 101. G.L. Torres et al. (Eds.), Advances in Intelligent Systems and Robotics - LAPTEC'03
Vol. 100. I. Aguil6 et al. (Eds.), Artificial Intelligence Research and Development
Vol. 99. D. Sue, Machine Reconstruction of Human Control Strategies
Vol. 98. H. Fujita and P. Johannesson (Eds.), New Trends in Software Methodologies, Tools and
Techniques - Proceedings ofLyee_W 03: The Second International Workshop on Lyee
Methodology /OS
Vol. 97. H.U. Hoppe et al. (Eds.), Artificial Intelligence in Education -Shaping the Future of Learning
through Intelligent Technologies Press
Vol. 96. S. Handschuh and S. Staab (Eds.), Annotation for the Semantic Web
Amsterdam• Berlin •Oxford• Tokyo• Washington, DC
ISSN 0922-6389
v
© 2004, The authors mentioned in the table of contents
All rights reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted,
Preface
in any form or by any means, without prior written permission from the publisher.
ISBN 1 58603 468 5
Library of Congress Control Number: 2004113876
Just as ontology developed over the centuries as part of philosophy, so in recent years on
tology has become intertwined with the development of the information sciences. Re
searchers in such areas as artificial intelligence, formal and computational linguistics, bio
Publisher
medical informatics, conceptual modeling, knowledge engineering and information re
IOS Press
trieval have come to realize that a solid foundation for their research calls for serious work
Nieuwe Hemweg 6B
in ontology, understood as a general theory of the types of entities and relations that make
1013 BG Amsterdam
The Netherlands up their respective domains of inquiry. In all these areas, attention has started to focus on
fax: [phone] the content of information rather than on just the formats and languages in terms of which
e-mail: [email protected] information is represented. A clear example of this development is provided by the many
initiatives growing up around the project of the Semantic Web. And as the need for inte
grating research in these different fields arises, so does the realization that strong principles
Distributor in the UK and Ireland Distributor in the USA and Canada for building well-founded ontologies might provide significant advantages over ad hoc,
!OS Press/Lavis Marketing !OS Press, Inc. case-based solutions. The tools of Formal Ontology address precisely these needs, but a real
73 Lime Walk 4502 Rachael Manor Drive effort is required in order to apply such philosophical tools to the domain of Information
Headington Fairfax, VA 22032
USA Systems. Reciprocally, research in the information science raises specific ontological ques
Oxford OX3 7 AD
fax: +I 703 323 3668 tions which call for further philosophical investigations.
England
fax: [phone] e-mail: [email protected] The purpose of the International Conference on Formal Ontology in Information Sys
tems is to provide a forum for genuine interdisciplinary exchange in the spirit of a unified
ontological analysis effort. This volume collects the proceedings of the third edition of the
conference, held in Turin (Italy), November 4-6, 2004. As with the previous editions
(Trento, Italy, 1998; Ogunquit, Maine, 2001), the primary focus is on theoretical issues, but
contributed papers deal also with methodological proposals and concrete applications. The
list of topics below, taken from the original call for papers, is indicative of the target ad
dressed:
Foundational Issues
• Kinds of entity: particulars vs. universals, continuants vs. occurrents, abstracta
vs. concreta, dependent vs. independent, natural vs. artificial
• Formal relations: parthood, identity, connection, dependence, constitution,
subsumption, instantiation
• Vagueness and granularity
• Identity and change
• Formal comparison among ontologies
• Ontology of physical reality (matter, space, time, motion, etc.)
• Ontology of biological reality (genes, proteins, cells, organisms, etc.)
• Ontology of mental reality and agency (beliefs, intentions and other mental attitudes;
emotions, etc.)
LEGAL NOTICE • Ontology of social reality (institutions, organizations, norms, social relationships,
The publisher is not responsible for the use which might be made of the following information. artistic expressions, etc.)
• Ontology of the infom1ation society (information, communication, meaning
PRINTED IN THE NETHERLANDS negotiation, etc.)
• Ontology and Natural Language Semantics
• Ontology and cognition
vi
vii
Methodologies and Applications
• Top-level vs. application ontologies
• Ontology integration and alignment; role of reference ontologies
Conference Committees
• Ontology-driven information systems design
• Requirements engineering
• Knowledge engineering
• Knowledge management and organization Organizing Committee
• Knowledge representation; Qualitative modeling Conference Chair
• Computational lexica; Terminology
Nicola Guarino Laboratory for Applied Ontology, Institute of Cognitive Sciences and
• Information retrieval; Question-answering Technologies, CNR, Trento, Italy
• Semantic web; Web services; Grid computing
• Domain-specific ontologies, especially for: Linguistics, Geography, Law, Library Program Chairs
science, Biomedical science, E-business, Enterprise integration, etc. Achille C. Varzi Department of Philosophy, Columbia University, New York, USA
Laure Vieu Research Institute for Computer Science, CNRS, Toulouse, France, and Laboratory
for Applied Ontology, Institute of Cognitive Sciences and Technologies, CNR,
Out of the 68 papers submitted in response to this call, 28 were selected by the Program
Trento, Italy
Committee, with the help of some additional reviewers. With few exceptions, all papers
have been refereed by three experts. On behalf of the Organizing Committee, we would like Local Chairs
to thank the members of the Program Committee for their work and for their constructive Maurizio Ferraris Department of Philosophy, University of Torino, Italy
suggestions, as well as the additional referees for the good job they have done. We are also Leonardo Lesmo Department of Computer Science, University of Torino, Italy
very grateful to the two invited speakers, Amie L. Thomasson and Peter Gardenfors, for the
Publicity Chair
enthusiasm with which they accepted our invitation. Finally, we would like to thank the Lo
Jos Lehmann Laboratory for Applied Ontology, Institute of Cognitive Sciences and
cal Chairs, Maurizio Ferraris and Leonardo Lesmo, the Publicity Chair, Jos Lehmann, and
Technologies, CNR, Rome, Italy
the Web Masters Guido Boella and Pierre Maurice, who all contributed to making FOIS
2004 a success. Web Masters
Guido Boella Department of Computer Science, University of Torino, Italy
Pierre Maurice Research Institute for Computer Science, INRIA, Toulouse, France
Achille C. Varzi
Laure Vieu Program Committee
Bill Andersen OntologyWorks Inc., Odenton, USA
Nicholas Asher Department of Philosophy, University of Texas at Austin, USA
Nathalie Aussenac-Gilles Research Institute for Computer Science, CNRS, Toulouse, France
John Bateman Department of Applied English Linguistics, University of Bremen, Germany
Brandon Bennett Division of Artificial Intelligence, University of Leeds, UK
Andrea Battani Department of Philosophy, University of Bergamo, Italy
Joost Breuker Department of Computer Science and Law, University of Amsterdam, The
Netherlands
Roberto Casati Jean Nicod Institute, CNRS, Paris, France
Werner Ceusters European Centre for Ontological Research, Saarland University, Saarbriicken,
Germany
Tony Cohn Division of Artificial Intelligence, University of Leeds, UK
Robert Colomb School of Computer Science and Electrical Engineering, University of Queensland,
Australia
Ernest Davis Department of Computer Science, New York University, USA
Martin Dorr Institute of Computer Science, FORTH, Heraklion, Greece
Carola Eschenbach Department for Informatics, University of Hamburg, Germany
Jerome Euzenat INRIA Rh6ne-Alpes, Grenoble, France
Christiane Fellbaum Cognitive Science Laboratory, Princeton University, USA, and Berlin Brandenburg
Academy of Sciences and Humanities, Berlin, Germany
Maurizio Ferraris Department of Philosophy, University of Torino, Italy
Antony Gaitan School of Engineering and Computer Science, University of Exeter, UK
Aldo Gangemi Laboratory for Applied Ontology, Institute of Cognitive Sciences and
Technologies, CNR, Rome, Italy
Peter Giirdenfors Lund University Cognitive Science, Sweden
Pierdaniele Giaretta Department of Philosophy, University of Padova, Italy
viii ix
Michael Gruninger Institute for Systems Research, University of Maryland, College Park, USA, and
National Institute for Standards and Technology, USA
Nicola Guarino Laboratory for Applied Ontology, Institute of Cognitive Sciences and
Contents
Technologies, CNR, Trento, Italy
Heinrich Herre Institute oflnformatics, University of Leipzig, Germany
Jacques Jayez ENS-Humanities, Lyon, France
Ingvar Johansson Institute for Formal Ontology and Medical Information Science, Saarl and
Preface
University, Saarbriicken, Germany v
Hannu Kangassalo Department of Computer and Information Sciences, University of Tampere, Finland Conference Committees vii
Fritz Lehmann USA
Leonardo Lesmo Department of Computer Science, University of Torino, Italy
Bernardo Magnini Centre for Scientific and Technological Research, ITC, Trento, Italy Invited Talks
David Mark Department of Geography, State University of New York, Buffalo, USA
William E. McCarthy Department of Accounting and Information Systems, Michigan State University,
Methods of Categorization
East Lansing, USA 3
Amie L. Thomasson
Robert Meersman Department of Computer Science, Free University of Brussels, Belgium
Chris Menzel Department of Philosophy, Texas A&M University, College Station, USA
Friederike Moltmann Department of Philosophy, Stirling University, UK How to Make the Semantic Web More Semantic 17
Philippe Muller Research Institute for Computer Science, University of Toulouse III, France Peter Giirdenfors
John Mylopoulos Department of Computer Science, University of Toronto, Canada
Leo Obrst The MITRE Corporation, McLean, USA
Massimo Poesio Department of Computer Science, University of Essex, UK 1. Categories
Ian Pratt-Hartmann Department of Computer Science, University of Manchester, UK
James Pustejovsky Department of Computer Science, Brandeis University, USA
Steffen Schulze-Kremer German Resource Center for Genome Research, Berlin, Germany Individuals, Universals, Collections: On the Foundational Relations of Ontology 37
Barry Smith Department of Philosophy, State University of New York, Buffalo, USA, and Thomas Bittner, Maureen Donnelly and Barry Smith
Institute for Formal Ontology and Medical Information Science, Saarland
University, Saarbrilcken, Germany A Formal Theory of Substances, Qualities, and Universals 49
John Sowa USA Fabian Neuhaus, Pierre Grenon and Barry Smith
Veda Storey Department of Computer Information Systems, Georgia State University, USA
Mike Uschold The Boeing Company, Seattle, USA
How to Be a Bicategorialist
Achille C. Varzi Department of Philosophy, Columbia University, New York, USA 60
Jonathan Simon
Laure Vieu Research Institute for Computer Science, CNRS, Toulouse, France, and Laboratory
for Applied Ontology, Institute of Cognitive Sciences and Technologies, CNR,
Trento, Italy
YairWand Management Information Systems Division, University of British Columbia, 2. Perspectives
Vancouver, Canada
Chris Welty IBM Watson Research Center, Hawthorne, USA
Beyond Concepts: Ontology as Reality Representation
Roel Wieringa Computer Science Department, University ofTwente, The Netherlands 73
Barry Smith
Additional Referees
Top-Level Ontology: The Problem with Naturalism 85
Guido Boella, Stefano Borgo, Massimiliano Carrara, Maarten M. Fokkinga, Jos Lehmann, Frank Loebe, Simon K. Milton
Claudio Masolo, Hannes Michalek, Vittorio Morato, Domenico Pisanelli, Giuseppe Spolaore, Pascal van Eck
Ontology, Society, and Ontotheology
95
Sponsors Joseph A. Goguen
Laboratory for Applied Ontology, Institute of Cognitive Sciences and Technologies, CNR, Trento, Italy
Institut de Recherche en Informatique de Toulouse, CNRS & Universite de Toulouse III, France
Dipartimento di Informatica, Universita di Torino, Italy 3. Methodology
Dipartimento di Filosofia, Universita di Torino, Italy
Comune di Torino, Italy Relative Definability in Formal Ontologies
Regione Piemonte, Italy 107
Brandon Bennett
Modal Rigidity in the OntoClean Methodology
119
William Andersen and Christopher Menzel
x xi
Identity and Modality in OntoClean 128 7. Mind and Action
Massimiliano Carrara, Pierdaniele Giaretta, Vittorio Morato, Marzia Soavi and
Giuseppe Spolaore Towards a Computational Ontology of Mind 287
Roberta Ferrario and Alessandro Oltramari
Specification of Ontologies in CASL 140
Klaus Lilttich and Till Mossakowski Supervenience in Content-Based Image Retrieval 298
Walter ten Brinke, David McG. Squire and John Bigelow
4. Semantics and Cognition An Ontological Formalization of the Planning Task 305
Dnyanesh Rajpathak and Enrico Motta
Formalizing Conceptual Spaces 153
Martin Raubal
8. Ontological Analysis
Dynamic Typing for Lexical Semantics - A Case Study: The Genitive Construction 165
Nicholas Asher and Pascal Denis Ontological Foundations of Biological Continuants 319
Stefan Schulz and Udo Hahn
Feature-Based vs. Property-Based KR: An Empirical Perspective 177
Massimo Poesio and Abdulrahman Almuhareb Searching for a Time Ontology for Semantic Web Applications 331
Mariano Fernandez-Lopez and Asuncion Gomez-Perez
The Ontology-Epistemology Divide: A Case Study in Medical Terminology 185
Olivier Bodenreider, Barry Smith and Anita Burgun Philosophical Scrutiny for Run-Time Support of Application Ontology Development 342
Mariana Casella dos Santos, James Matthew Fielding, Christoffel Dhaen and
Werner Ceusters
5. Social Reality
Analyzing OWL Using a Philosophy-Based Ontology 353
An Agent-Oriented Ontology of Social Reality 199 Pa/ash Bera and Yair Wand
Guido Boella and Leendert van der Torre
Core Concepts of Law: Taking Common Sense Seriously 210 Author Index 363
Joost Breuker and Rinke Hoekstra
The Place of Language within a Foundational Ontology 222
John Bateman
6. Space, Time and Causation
Towards a Generic Foundation for Spatial Ontology 237
John Bateman and Scott Farrar
Relative Places 249
Maureen Donnelly
A Four-Dimensionalist Mereotopology 261
John G. Stell and Matthew West
Causality and Causation in DOLCE 273
Jos Lehmann, Stefano Bargo, Claudio Masolo and Aldo Gangemi
Invited Talks
Formal Ontology in Information Systems 3
A. C. Varzi and L. Vieu (Eds.) ·
!OS Press, 2004
Methods of Categorization
Amie L. THOMASSON
University ofM iami
Abstract. A system of basic ontological categories has a number of important po
tential uses both within and outside of philosophy. Unfortunately, however, past
philosophical work on categories provides an embarrassment of riches, since so
many systems have been proposed with so little agreement about what the categories
are or even what form a system of categories should take. I argue that we can only
make headway on the project of categorization by stepping back to consider what
methods we should use in developing a system of categories, and by what criteria we
should evaluate proposed systems of categories. I consider three promising methods
and argue that the third in particular may provide a useful route to developing an
adequate and useful system of categories.
Introduction
It's not surprising that many people find a basic system of ontological categories desirable.
Within philosophy, providing a system of ontological categories has the potential to be one
of the most useful tasks in metaphysics. One of the central tasks in metaphysics is to draw
out an ontology-a view about what things, more particularly, what kinds of things, exist.
And there are great controversies about whether or not one should accept that there are
things like numbers, properties, merely possible entities, or even ordinary artifacts, or if in
stead one should treat talk of these as a mere manner of speaking about more fundamental
entities (such as those postulated by physics). Since a system of categories is supposed to
provide an inventory of all of the highest-level kinds or genera that are, or even might be,
exemplified, it could help enormously in our attempts to formulate an ontology. For if we
have a list of highest categories of things there might be, we can then take them one by one,
attempting to answer in tum whether or not there are entities of each highest kind, giving us
a systematic approach to ontology. If that list of categories is complete and finite, we then
even have a finite procedure for answering the question 'what exists', by way of answering
the question 'do Ks exist' for each highest kind K.
Another important task of metaphysics is to address specific questions about the 'on
tological nature' of entities of certain kinds, asking, e.g., what sorts of things minds, per
sons, works of art, or governments are. For example, what is a work of literature? Is it a
physical object (pages with print), an abstract sequence of sentences, an idea in the mind of
the author or reader, or something else? In answering these questions, we typically presup
pose a certain range of ontological categories that could serve as possible answers to the
question. But, as I have argued elsewhere [1] [2], if our list of categories is not exhaustive,
we might overlook possible answers and fail to get an adequate answer.
Nor are the uses of systems of categories limited to those within metaphysics or even
within philosophy, as witnessed by the interest in top-level ontological categories within
information science and the biomedical sciences. And it is not surprising that those from
other fields with an interest in categories might hope to gain some insight from philosophy,
given the long history of philosophical work on categories from Aristotle onwards.
4 A.L. Thomasson I Methods of Categorization A.L. Thomasson I Methods ofC ategorization 5
Unfortunately, however, while turning to one philosopher for help may be useful, there are in the world. So his interest was in drawing out a system of categories in what we
turning to many is more problematic. For although a great deal of work on categories has might call the 'realist' spirit-as categories of real entities in the world (not categories of
been done-with major figures in the history of philosophy such as Aristotle, Kant, and thought or oflanguage).
Husserl, as well as a number of contemporary authors, laying out systems of categories-as Nonetheless, his method for approaching this project was via language. Aristotle be
I will show via a brief historical overview in § 1, the systems of categories proposed differ gins with the first category of substance, where a primary substance is, e.g., an individual
widely among themselves, and there is no consensus about which of these diverse set of man or horse, and the species (man, animal) are secondary substances. Names for primary
categories we should choose. In fact, there is not even any consensus about what the right substances are the natural grammatical subjects in sentences, indeed, "all the other things
form of answer to the question "What are the highest categories" would be-whether it are either said of the primary substances as subjects or in them as subjects" [3, 2b4]. The
should take the form of a porphyrian tree with a single highest category at the top, sub ancient Greek term "kategoriai" described what could be said against someone in a court of
suming a number of genuses, which in tum subsume species, and so on; or should top out in law, and Aristotle adopts this name for his treatise since all the other categories are dis
a simple list of highest kinds; whether a single list of categories can be provided, or if we cerned by distinguishing different things that could be said of (or "in") a substance, by
need multi-dimensional lists or charts of categories, and so on. noting the different sorts of questions that can be asked about something, and the sorts of
The result has been an increasing sense of embarrassment about categories in philoso answer that would be appropriate [4, 78-9]. E.g., ofa substance, we can ask "how long is it"
phy, and skepticism that any set of categories could rightly lay claim to being the 'correct' and get a quantity for an answer, "what is it like" and get a quality for an answer, "what is it
system, or even to being preferable to the others. For how can we choose non-arbitrarily doing" and get an action for an answer, and so on.
among the various systems of categories offered, or even non-arbitrarily figure out what By using this method, Aristotle famously arrived at a list of ten highest categories (see
form a system of categories ought to take, e.g. whether it should be tree-shaped or list-like, Table 1).
etc.? If we have no principled way to develop a system of categories or choose among
them, the philosophical exercise of offering systems of categories begins to look point Table I: Aristotle's Categories
less-and the systems offered appear more like byproducts of an individual philosopher's
presuppositions than insights into basic kinds. But if that is so, then the benefits we hoped ·Substance (e.g. man, horse)
·Quantity (e.g. four-foot, five-foot)
that a system of categories could provide begin to look unattainable, and it begins to look
·Quality (e.g. white, grammatical)
like philosophers would better spend their time elsewhere than in trying to formulate a sys · Relation (e.g. double, half)
tem of basic ontological categories. ·Place (e.g. in the Lyceum, in the market-place)
It seems to me that at least a large part of the source of this embarrassment lies in the ·Date (e.g. yesterday, last year)
fact that, in the drive to lay out systems of categories, philosophers have often paid insuffi · Posture (e.g. is lying, is sitting)
· State (e.g. has shoes on, has armor on)
cient attention to questions of methodology. I will argue that if we hope to make progress in
· Action (e.g. cutting, burning)
choosing among competing systems of categories or formulating a new and better system of · Passion (e.g. being cut, being burned)
categories that could provide some of the benefits mentioned above, we must begin by
stepping back from the task of enumerating categories to consider more deeply the meth Notice that these are all same-level highest categories-Aristotle argued that there could
odological questions of how we are to draw out a system of categories, and what criteria we not be a highest genus covering all of these categories [5, 998b22-3], so the original para
can use in evaluating proposed category systems. I will begin with the latter task in §2, digm system of categories takes the form of a simple list.
asking what criteria of adequacy a system of categories must meet if it is to help us with the Worries inevitably arise, however, in using categories of language (here, of question
metaphysical jobs described above, and what should count as virtues (or vices) in a syst~m and answer) as the key to revealing what are supposed to be categories ofreal entities in the
of categories. I will then go on in §§3-5 to discuss three proposals for methods of drawmg world. For, it might be asked, what guarantees that the categories of our language, or the
out systems of categories that can meet these criteria of adequacy: by feature and negation sorts of question and answer deemed appropriate by the rules of our language, map onto, or
(§3), by detecting absurdities (§4), and by criteria of identity and existence (§5). While each have anything at all to tell us about, the categories of items there really are in the world?
of these methods is promising in certain ways, and brings some advantages, I will focus on This is an issue that will arise again for any method of discerning ontological categories via
the virtues of the last of these, arguing that it provides a principled way of answering long linguistic categories, and I will return to it below.
standing questions about categories, and can yield a system of categories that meets the
needed standards of adequacy. That, in turn, can give us hope that, despite the daunting 1.2 Kant
history of philosophical category systems, we needn't abandon philosophical discussion of
categories, or the hopes of reaping the benefits that a well-worked out system of categories While Aristotle uses forms of language as the clue to distinguishing categories, Immanuel
can provide. Kant uses forms of thought (more specifically, of judgment) as the clue to his categories.
And while Aristotle leaves unanswered the question of how we can be sure that the catego
ries of language map onto categories of object, Kant tries to show that the categories of
1 Some Past Category Systems
thought (of the understanding) do have something to tell us about the categories of objects
known, because our thought imposes structure on the objects it can know. According to
1.1 Aristotle
Kant we can know nothing about mind-independent things in themselves; the only objects
In the Western philosophical tradition, the earliest known work on categories is Aristotle's we can hope to know are those of the phenomenal world of appearances-objects
treatise of the same name, the goal of which was to discern the most general kinds of entity considered-as objects of possible representations. What can serve as the object of a possible
6 A.L. Thomasson I Methods ofC ategorization A.L. Thomasson I Methods of Categorization 7
representation is, according to Kant, constrained by our possible ways of representing; Ingvar Johansson, who takes on the project of offering "a realist theory of categories
more specifically, what can serve as an object of possible thought' or cognition is con regarded as real aspects of being" [6, 2], claims to distinguish his list of nine major catego
strained by the forms of our understanding. ries by the method of 'successive abstraction'. So, for example, observing a particular man,
Socrates, we could move from 'Socrates is human' to 'Socrates is a mammal' to 'Socrates
Table 2: Kant's Categories is an animal' to finally arrive at 'Socrates is a substance'. Similarly, observing a particular
shade of violet we could move from 'dusty lavender' to 'purple' to 'color' finally to 'prop
Unity erty'. Johanssen's final list of categories consists of nine highest-level categories, some of
/
which are further subdivided (see Table 3).
Quantity Plurality
'-....._
Totality
Table 3: Johansson's Categories
Reality
/
Quality Negation
Space-time
'-....._
Limit
State of Affairs
Inherence and Subsistence (substance and accident) __- Substance
Quality --
Relation Causality and Dependence (cause and effect) Property
Community (reciprocity) External Relation
Grounded Relation
Possibility
Inertia
Modality Existence
Spontaneity
< Presentational
Necessity
Tendency
Real
Intentionality
So for Kant, although we can't know what the categories are of objects in themselves, Representational
we can find out what categories govern human understanding. Since these impose catego Fictional
ries on the phenomena known, we can use the categories of the understanding as the clue to
discovering the categories of all objects we could possibly have knowledge of (these, of
But while the method of successive abstraction is supposed to give us a way to extract
course, are still phenomena, not things in themselves).
a system of categories directly from the world, without mediation of language or thought,
Kant's method of generating a system of categories is thus based on starting from list
there are grounds for suspecting that this is not the case. For in picking out the individual
ing possible forms of judgment one might make (ostensibly) about objects. Following Ar
man Socrates, from whom we can then abstract first to the category 'human' and so on, are
istotle, Kant notes that judgments can be classified according to their quantity, quality, re
we not already presupposing a category 'man' to pick out this very individual Socrates
lation, or modality (each of which has three alternative possible values). He uses these
categories of judgment as the clue to arriving at categories that apply to all possible objects (from whom we can abstract), rather than, say, picking out the color of his skin, the lump of
matter composing him, the region of space-time he occupies, and so on? If so, the method
of cognition (see Table 2). According to Kant, while these can't be considered categories of
of successive abstraction might after all presuppose a set of conceptual or linguistic catego
independent things-in-themselves, anything we could possibly think about can be character
ized according to its quantity as a unity, plurality, or totality, and so on. ries and not provide a direct route for reading categories off of reality itself. I will return to
this issue below.
So Kant solves the problem of how categories of judgment could tell us about catego
Supposing for the moment that the method of successive abstraction is workable, one
ries of objects, but at the cost of making the relevant objects mere phenomena, and denying
might wonder why, if using that method, we could not abstract further from each of these to
that we could have any knowledge of independent things-in-themselves, or what categories
get a single, most-general category of 'entity' or some such. Although few contemporary
these might fall into.
philosophers are explicit about the methods they use to distinguish their categories, it may
be tacit appeal to such an abstractive method that that leads many contemporary philoso
1. 3 Contemporary Category Systems phers to propose category systems that top out in a single highest category, giving the sys
tems the form ofporphyrian trees. Thus, e.g., Roderick Chisholm's categories reach up to a
Given the difficulties of ensuring that categories of language or thought are correlated with single highest category of 'entia' (see Table 4).
real categories of mind-independent entities in the world, one might wonder why historical Joshua Hoffman and Gary Rosenkrantz similarly offer a tree with 'entity' at the top,
philosophers like Aristotle and Kant have approached the project of categories in this medi but 'concrete' and 'abstract' (rather than 'contingent' and 'necessary') marking the primary
ated way, rather than trying to acquire a system of categories more directly by study of the division (see Table 5).
world. This is what at least one contemporary philosopher proposes to do.
8 A.L. Thomasson I Methods of Categorization A.L. Thomasson I Methods of Categorization 9
Table 4: Chisholm's Categories cipled, and efficient way. Second, it ought to help us answer more specific questions about
what sorts of things works of art, minds, governments, etc., are, by providing a more com
Entia plete range of alternatives so that we do not miss any potential solutions.
To serve these purposes well, it seems that a system of categories must at least meet
two minimal criteria of adequacy [7, 75]. First, a category system should strive to be ex
Contingent Necessary haustive, providing a complete list of highest kinds, so that there is a category for every-
. thing there might be. Then we can use it to generate a complete assessment of what there is,
and ensure that no options are missed in seeking to categorize various specific types of
States Individuals States Nonstates
thing. Second, the categories provided (at any given level) should be mutually exclusive, so
that we avoid redundancy (and retain efficiency), and ensure that whatever there is can be
Events Boundaries Substances Attributes Substance uniquely located in exactly one category [8, 162].
No doubt this is still not good enough to uniquely select a system of categories from
among those proposed, much less to generate one ex nihilo. But we need not presuppose
Table 5: Hoffman and Rosenkrantz's Categories that a category system is the sole true, correct system for it to be useful in developing a
systematic ontology and in providing alternative possible classifications for entities of vari
Entity ous types. And even if these criteria alone don't generate a system of categories, they may
give us an idea of what methods might be useful for generating systems of categories that
---------------
could meet these criteria of adequacy.
Abstract Concrete
Prope~osition ~\~ 3 Methods of Deriving Categories
/ \ \
T•rope
~,! ~
Relation 3.1 The Feature-Negation Method
Event Time Privation
Limit
Place There is one obvious way of ensuring that both of the above criteria are met: draw out cate
Collection
Substance gories based in whether an entity has or does not have a certain kind of feature. That is, for
.......--......
any feature F we care to name, we can get a division into two categories by distinguishing
Material Spirit those things that are F (i.e. such that Fx) from those that are not F (i.e such that ~ (Fx)).
Object Two widely accepted logical principles ensure that this method gives us a division into
mutually exclusive and exhaustive groupings. The Principle of Non-Contradiction is the
Even such a brief survey as this is probably sufficient to illustrate the wide variations in principle that, for any sentence P, it is not the case that (P and ~P). Thus, applied to simiple
the content and form of category systems that have been offered. And that in tum may suf subject-predicate sentences, it ensures that, for any individual x and predicate F, it is not the
fice to create a sense of category fatigue. Faced with so many different systems of catego case that (Fx and ~ (Fx)), thereby guaranteeing that the categories are mutually exclusive,
ries, how can we even retain the faith that some correct--or at least adequate-system of with each entity belonging in at most one category .1 The Law of the Excluded Middle is the
categories can be found that help us with the philosophical tasks I described at the start? principle that, for any individual x and predicate F, (Fx or ~(Fx)). So this principle ensures
And if philosophers can't even agree among themselves about what the right categories are, that each entity finds its place in at least one category, and thus that the categories distin
what the right method is for distinguishing categories, or even what form a category system guished are jointly exhuastive.
should take, how could they possibly be of help to those in other fields where work on Applying this method results in two highest-level categories: things that are F and
categories is important? those that are not. It is not obvious that any further branches (marking subdivisions in our
top-level categories) could be made in this way, since (for many other features G) things
that are, and are not, F, might equally well be, or not be, G. So, for example, (as I have ar
2 The Need for Criteria gued elsewhere [1]), if we drew a first division between things that are and are not concrete
(located in space and time), and as the second, things that are and are not artifacts (products
To confront this category fatigue (which leads directly to category skepticism), we first of human creation), we would get things that are artifacts on both sides of the original di
need a set of criteria by means of which we can evaluate different proposed systems of vide (e.g. with chairs and statues on the concrete side, nations and symphonies on the ab
categories. If a system of categories is supposed to be a useful tool for metaphysical in stract side).
quiries, then, like other tools, we can develop criteria to evaluate it based at least in part on Nonetheless, we can get more complex systems of categories using this method by
how well it can succeed at doing its jobs. As I described it in the introduction, a good cate placing different features to be considered along different dimensions, so that the resulting
gory system ought to do at least two jobs for philosophy: First, it ought to provide us with a category system takes the form of a matrix rather than a tree. Thus, for example, by consid-
framework that will enable us to answer the question "What exists?" in a systematic, prin-
1 At least, at a time. Or we can avoid this caveat by indexing predicates to times.