ArticlePDF Available

Abstract

This survey provides a conceptual introduction to ontologies and their role in information systems and AI. The authors also discuss how ontologies clarify the domain's structure of knowledge and enable knowledge sharing
ONTOLOGIES
What Are Ontologies, and
Why Do We Need Them?
B. Chandrasekaran and John R. Josephson, Ohio State University
V. Richard Benjamins, University of Amsterdam
T
HEORIES IN AI FALL INTO TWO
broad categories: mechanism theories and
content theories. Ontologies are content the-
ories about the sorts of objects, properties of
objects, and relations between objects that are
possible in a specified domain of knowledge.
They provide potential terms for describing
our knowledge about the domain.
In this article, we survey the recent devel-
opment of the field of ontologies in AI. We
point to the somewhat different roles ontolo-
gies play in information systems, natural-
language understanding, and knowledge-
based systems. Most research on ontologies
focuses on what one might characterize as
domain factual knowledge, because knowl-
ede of that type is particularly useful in nat-
ural-language understanding. There is an-
other class of ontologies that are important
in KBS—one that helps in sharing know-
eldge about reasoning strategies or problem-
solving methods. In a follow-up article, we
will focus on method ontologies.
Ontology as vocabulary
In philosophy, ontology is the study of the
kinds of things that exist. It is often said that
ontologies “carve the world at its joints.” In
AI, the term ontology has largely come to
mean one of two related things. First of all,
ontology is a representation vocabulary, often
specialized to some domain or subject matter.
More precisely, it is not the vocabulary as such
that qualifies as an ontology, but the concep-
tualizations that the terms in the vocabulary
are intended to capture. Thus, translating the
terms in an ontology from one language to
another, for example from English to French,
does not change the ontology conceptually. In
engineering design, you might discuss the
ontology of an electronic-devices domain,
which might include vocabulary that describes
conceptual elements—transistors, operational
amplifiers, and voltages—and the relations
between these elements—operational ampli-
fiers are a type-of electronic device, and tran-
sistors are a component-of operational ampli-
fiers. Identifying such vocabulary—and the
underlying conceptualizations—generally
requires careful analysis of the kinds of objects
and relations that can exist in the domain.
In its second sense, the term ontology is
sometimes used to refer to a body of knowl-
edge describing some domain, typically a
commonsense knowledge domain, using a
representation vocabulary. For example,
CYC
1
often refers to its knowledge repre-
sentation of some area of knowledge as its
ontology.
In other words, the representation vocab-
ulary provides a set of terms with which to
describe the facts in some domain, while the
body of knowledge using that vocabulary is
a collection of facts about a domain. How-
ever, this distinction is not as clear as it might
first appear. In the electronic-device exam-
ple, that transistor is a component-of opera-
tional amplifier or that the latter is a type-of
electronic device is just as much a fact about
THIS SURVEY PROVIDES A CONCEPTUAL INTRODUCTION
TO ONTOLOGIES AND THEIR ROLE IN INFORMATION
SYSTEMS AND
AI. THE AUTHORS ALSO DISCUSS HOW
ONTOLOGIES CLARIFY THE DOMAIN
S STRUCTURE OF
KNOWLEDGE AND ENABLE KNOWLEDGE SHARING
.
20 1094-7167/99/$10.00 © 1999 IEEE IEEE INTELLIGENT SYSTEMS
.
its domain as a CYC fact about some aspect
of space, time, or numbers. The distinction
is that the former emphasizes the use of
ontology as a set of terms for representing
specific facts in an instance of the domain,
while the latter emphasizes the view of ontol-
ogy as a general set of facts to be shared.
There continues to be inconsistencies in
the usage of the term ontology. At times, the-
orists use the singular term to refer to a spe-
cific set of terms meant to describe the entity
and relation-types in some domain. Thus, we
might speak of an ontology for “liquids” or
for “parts and wholes.” Here, the singular
term stands for the entire set of concepts and
terms needed to speak about phenomena
involving liquids and parts and wholes.
When different theorists make different pro-
posals for an ontology or when we speak about
ontology proposals for different domains of
knowledge,we would then use the plural term
ontologies to refer to them collectively. In AI
and information-systems literature, however,
there seems to be inconsistency: sometimes we
see references to “ontology of domain” and
other times to “ontologies of domain,” both
referring to the set of conceptualizations for
the domain. The former is more consistent with
the original (and current) usage in philosophy.
Ontology as content theory
The current interest in ontologies is the lat-
est version of AI’s alternation of focus be-
tween content theories and mechanism the-
ories. Sometimes, the AI community gets
excited by some mechanism such as rule sys-
tems, frame languages, neural nets, fuzzy
logic, constraint propagation, or unification.
The mechanisms are proposed as the secret
of making intelligent machines. At other
times, we realize that, however wonderful the
mechanism, it cannot do much without a
good content theory of the domain on which
it is to work. Moreover, we often recognize
that once a good content theory is available,
many different mechanisms might be used
equally well to implement effective systems,
all using essentially the same content.
2
AI researchers have made several attempts
to characterize the essence of what it means
to have a content theory. McCarthy and
Hayes’theory (epistemic versus heuristic dis-
tinction),
3
Marr’s three-level theory (infor-
mation processing, strategy level, algorithms
and data structures level, and physical mech-
anisms level),
4
and Newell’s theory (Knowl-
edge Level versus Symbol Level)
5
all grap-
ple in their own ways with characterizing
content. Ontologies are quintessentially con-
tent theories, because their main contribution
is to identify specific classes of objects and
relations that exist in some domain. Of
course, content theories need a representa-
tion language. Thus far, predicate calculus-
like formalisms, augmented with type-of
relations (that can be used to induce class
hierarchies), have been most often used to
describe the ontologies themselves.
Why are ontologies
important?
Ontological analysis clarifies the structure
of knowledge. Given a domain, its ontology
forms the heart of any system of knowledge
representation for that domain. Without
ontologies, or the conceptualizations that
underlie knowledge, there cannot be a vocab-
ulary for representing knowledge. Thus, the
first step in devising an effective knowledge-
representation system, and vocabulary, is to
perform an effective ontological analysis of
the field, or domain. Weak analyses lead to
incoherent knowledge bases.
An example of why performing good
analysis is necessary comes from the field of
databases.
6
Consider a domain having sev-
eral classes of people (for example, students,
professors, employees,females, and males).
This study first examined the way this data-
base would be commonly organized: stu-
dents, employees, professors, males, and
female would be represented as types-of the
class humans. However, some of the prob-
lems that exist with this ontology are that stu-
dents can also be employees at times and can
also stop being students. Further analysis
showed that the terms students and employee
do not describe categories of humans, but are
roles that humans can play, while terms such
as females and males more appropriately rep-
resent subcategories of humans. Therefore,
clarifying the terminology enables the ontol-
ogy to work for coherent and cohesive rea-
soning purposes.
Second, ontologies enable knowledge
sharing. Suppose we perform an analysis and
arrive at a satisfactory set of conceptualiza-
tions, and their representative terms, for some
area of knowledge—for example, the elec-
tronic-devices domain. The resulting ontol-
ogy would likely include domain-specific
terms such as transistors and diodes; general
terms such as functions, causal processes,
and modes; and terms that describe behavior
such as voltage. The ontology captures the
intrinsic conceptual structure of the domain.
In order to build a knowledge representation
language based on the analysis, we need to
associate terms with the concepts and rela-
tions in the ontology and devise a syntax for
encoding knowledge in terms of the concepts
and relations. We can share this knowledge
representation language with others who
have similar needs for knowledge represen-
tation in that domain, thereby eliminating the
need for replicating the knowledge-analysis
process. Shared ontologies can thus form the
basis for domain-specific knowledge-repre-
sentation languages. In contrast to the previ-
ous generation of knowledge-representation
languages (such as KL-One), these lan-
guages are content-rich; they have a large
number of terms that embody a complex con-
tent theory of the domain.
Shared ontologies let us build specific
knowledge bases that describe specific situ-
ations. For example, different electronic-
devices manufacturers can use a common
vocabulary and syntax to build catalogs that
describe their products. Then the manufac-
turers could share the catalogs and use them
in automated design systems. This kind of
sharing vastly increases the potential for
knowledge reuse.
Describing the world
We can use the terms provided by the
domain ontology to assert specific proposi-
tions about a domain or a situation in a
domain. For example, in the electronic-
device domain, we can represent a fact about
a specific circuit: circuit 35 has transistor 22
as a component, where circuit 35 is an
instance of the concept circuit and transistor
22 is an instance of the concept transistor.
Once we have the basis for representing
propositions, we can also represent knowl-
edge involving propositional attitudes (such
as hypothesize,believe,expect, hope,desire,
and fear). Propositional attitude terms take
propositions as arguments. Continuing with
the electronic-device domain, we can assert,
for example: the diagnostician hypothesizes
or believes that part 2 is broken, or the
designer expects or desires that the power
plant has an output of 20 megawatts. Thus,
an ontology can represent beliefs, goals,
JANUARY/FEBRUARY 1999 21
.
hypotheses, and predictions about a domain,
in addition to simple facts. The ontology also
plays a role in describing such things as plans
and activities, because these also require
specification of world objects and relations.
Propositional attitude terms are also part of
a larger ontology of the world, useful espe-
cially in describing the activities and prop-
erties of the special class of objects in the
world called “intensional entities”—for
example, agents such as humans who have
mental states.
Constructing ontologies is an ongoing
research enterprise. Ontologies range in
abstraction, from very general terms that
form the foundation for knowledge repre-
sentation in all domains, to terms that are
restricted to specific knowledge domains. For
example, space, time, parts, and subpartsare
terms that apply to all domains; malfunction
applies to engineering or biological domains;
and hepatitis applies only to medicine.
Even in cases where a task might seem to
be quite domain-specific, knowledge repre-
sentation might call for an ontology that des-
cribes knowledge at higher levels of gener-
ality. For example, solving problems in the
domain of turbines might require knowledge
expressed using domain-general terms such
as flows and causality. Such general-level
descriptive terms are called the upper ontol-
ogy or top-level ontology. There are many
open research issues about the correct ways
to analyze knowledge at the upper level. To
provide some idea of the issues involved,
Figure 1 excerpts a quote from a recent call
for papers.
Today,ontology has grown beyond philos-
ophy and now has many connections to infor-
mation technology. Thus, research on ontol-
ogy in AI and information systems has had to
produce pragmatically useful proposals for
top-level ontology. The organization of a top-
level ontology contains a number of problems,
similar to the problems that surround ontol-
ogy in philosophy. For example, many ontolo-
gies have thing or entity as their root class.
However, Figure 2 illustrates that thing and
entity start to diverge at the next level.
For example, CYC’s thing has the subcat-
egories individual object, intangible, and rep-
resented thing; the Generalized Upper
Model’s
7
(GUM) um-thing has the subcate-
gories configuration, element,and sequence;
Wordnet’s
8
thing has the subcategories liv-
ing thing and nonliving thing, and Sowa’s
root T has the subcategories concrete, pro-
cess, object, and abstract. (Natalya Fridman
Noy’s and Carol Hafner’s article discusses
these differences more fully.
9
) Some of these
differences arise because not all of these
ontologies are intended to be general-pur-
pose tools, or even explicitly to be ontolo-
gies. Another reason for the differences is
that, in principle, there are many different
taxonomies.
Although differences exist within ontolo-
gies, general agreement exists between on-
tologies on many issues:
There are objects in the world.
Objects have properties or attributes that
can take values.
Objects can exist in various relations with
each other.
Properties and relations can change over
time.
There are events that occur at different
time instants.
There are processes in which objects par-
ticipate and that occur over time.
The world and its objects can be in dif-
ferent states.
Events can cause other events or states as
effects.
Objects can have parts.
The representational repertoire of objects,
relations, states, events, and processes does
not say anything about which classes of these
entities exist. The modeler of the domains
makes these commitments. As we move from
an ontology’s top to lower taxonomic levels,
commitments specific to domains and phe-
nomena appear. For modeling objects on
earth, we can make certain commitments. For
example, animals, minerals, and plants are
subcategories of objects; has-life(x) and con-
tains-carbon(x) are object properties; and
can-eat(x, y) is a possible relation between
any two objects. These commitments are spe-
cific to objects and phenomena in this do-
main. Further, the commitments are not arbi-
trary. For them to be useful, they should
reflect some underlying reality.
There is no sharp division between do-
main-independent and domain-specific on-
tologies for representing knowledge. For
example, the terms object, physical object,
device,engine, and diesel engine all describe
objects, but in an order of increasing domain
specificity. Similarly, terms for relations
between objects can span a range of speci-
ficity, such as connected, electrically-con-
nected, and soldered-to.
Subtypes of concepts. Ontologies generally
appear as a taxonomic tree of conceptual-
izations, from very general and domain-
independent at the top levels to increasingly
domain-specific further down in the hierar-
chy. We mentioned earlier that different
ontologies propose different subtypes of even
very general concepts. This is because, as a
rule,different sets of subcategories will result
from different criteria for categorization. Two,
among many, alternate subcategorizations of
the general concept object are physical and
abstract, and living and non-living. In some
cultures and languages, words for objects
have gender, thus creating another top-level
classification along the gender axis. We can
easily think of additional subcategorizations
based on other criteria. The existence of alter-
nate categorizations only becomes more acute
as we begin to model specific domains of
knowledge. For example, we can subcatego-
rize causal process into continuous and dis-
crete causal processes along the dimension
of how time is represented, and into mechan-
ical, chemical, biological, cognitive, and
social processes along the dimension of the
kinds of objects and relations involved in the
description.
In principle, the number of classification
criteria and distinct subtypes is unlimited,
because the number of possible dimensions
along which to develop subcategories can-
22 IEEE INTELLIGENT SYSTEMS
Figure 1. Call for papers for a special issue on temporal parts for
The Monist, An International Quarterly Journal of
General Philosophical Inquiry.
This quote suggests that ontology has always been an issue of deep concern in philoso-
phy and that the issues continue to occupy contemporary philosophers.
On the one hand there are entities, such as processes and events, which have temporal
parts.… On the other hand there are entities, such as material objects, which are always pre-
sent in their entirety at any time at which they exist at all. The categorical distinction between
entities which do, and entities which do not have temporal parts is grounded in common
sense. Yet various philosophers have been inclined to oppose it. Some … have defended an
ontology consisting exclusively of things with no temporal parts. Whiteheadians have favored
ontologies including only temporally extended processes. Quine has endorsed a four-dimen-
sional ontology in which the distinction between objects and processes vanishes and every
entity comprises simply the content of some arbitrarily demarcated portion of space-time.
One further option, embraced by philosophers such as David Lewis, accepts the opposition
between objects and processes, while still finding a way to allow that all entities have both
spatial and temporal parts.
.
not be exhaustively specified. Often, this fact
is not obvious in general-purpose ontologies,
because the top levels of such ontologies
commit to the most commonly useful sub-
types. However, domain-specific ontologies
can contain categorizations along dimensions
that are usually outside the general ontology.
Task dependence of ontologies. How task-
dependent are ontologies? Presumably, the
kinds of things that actually exist do not
depend on our goals. In that sense, ontologies
are not task-dependent. On the other hand,
what aspects of reality are chosen for encod-
ing in an ontology does depend on the task.
For example, in the domain of fruits, we
would focus on particular aspects of reality if
we were developing the ontology for the
selection of pesticides; we would focus on
other aspects of reality if we were develop-
ing an ontology to help chefs select fruits for
cooking. In ontologies for engineering appli-
cations, categorizing causal processes into
those that do, and that do not, produce dan-
gerous side effects might be useful. Design
engineers and safety analysts might find this
a very useful categorization, though it is
unlikely to be part of a general-purpose ontol-
ogy’s view of the causal process concept.
Practically speaking, an ontology is un-
likely to cover all possible potential uses. In
that sense, both an ontology for a domain and
a knowledge base written using that ontology
are likely to be more appropriate for certain
uses than others and unlikely to be sharable
across widely divergent tasks. This is, by now,
a truism in KBS research and is the basic
insight that led to the current focus on the rela-
tionship between tasks and knowledge types.
Presuppositions or requirements can be asso-
ciated with problem-solving methods for dif-
ferent tasks so that they can capture explicitly
the way in which ontologies are task-depen-
dent. For example, a method might have a pre-
supposition (or assumption
10
) stating that it
works correctly only if the ontology allows
modeling causal processes discretely. There-
fore, assumptions are a key factor in practical
sharing of ontologies.
Technology for ontology
sharing
There have been several recent attempts to
create engineering frameworks for construct-
ing ontologies. Michael R. Genesereth and
Richard E. Fikes describe KIF (Knowledge
Interchange Format), an enabling technology
that facilitates expressing domain factual
knowledge using a formalism based on aug-
mented predicate calculus.
11
Robert Neches
and his colleagues describe a knowledge-shar-
ing initiative,
12
while Thomas R. Gruber has
proposed a language called Ontolingua to help
construct portable ontologies.
13
In Europe, the
CommonKADS project has taken a similar
approach to modeling domain knowledge.
14
These languages use varieties of predicate
calculus as the basic formalism. Predicate
calculus facilitates the representation of
objects, properties, and relations. Variations
such as situational calculus introduce time
so as to represent states, events, and pro-
cesses. If we extend the idea of knowledge
to include images and other sense modali-
ties, we might need radically different kinds
of representation. For now, predicate calcu-
lus provides a good starting point for ontol-
ogy-sharing technologies.
Using a logical notation for writing and
sharing ontologies does not imply any com-
mitment to implementing a related knowl-
edge system or a related logic. We are simply
taking a knowledge-level
5
stance in describ-
ing the knowledge system, whatever the
means of implementation. In this view, we
can ask of any intelligent system, even one
implemented as a neural network, “What
does the system know?”
Use of ontologies
In AI, knowledge in computer systems is
thought of as something that is explicitly rep-
resented and operated on by inference pro-
cesses. However, that is an overly narrow
view. All information systems traffic in knowl-
edge. Any software that does anything useful
cannot be written without a commitment to a
model of the relevant world—to entities, prop-
erties, and relations in that world. Data struc-
tures and procedures implicitly or explicitly
make commitments to a domain ontology. It
is common to ask whether a payroll system
“knows” about the new tax law, or whether a
database system “knows” about employee
salaries. Information-retrieval systems, digi-
tal libraries, integration of heterogeneous
information sources, and Internet search
engines need domain ontologies to organize
information and direct the search processes.
For example, a search engine has categories
and subcategories that help organize the
search. The search-engine community com-
monly refers to these categories and subcate-
gories as ontologies.
Object-oriented design of software sys-
tems similarly depends on an appropriate
domain ontology. Objects, their attributes,
and their procedures more or less mirror
aspects of the domain that are relevant to the
application. Object systems representing a
useful analysis of a domain can often be
reused for a different application program.
Object systems and ontologies emphasize
different aspects, but we anticipate that over
time convergence between these technolo-
gies will increase. As information systems
model large knowledge domains, domain
ontologies will become as important in gen-
eral software systems as in many areas of AI.
In AI, while knowledge representation per-
vades the entire field, two application areas
in particular have depended on a rich body of
knowledge. One of them is natural-language
understanding. Ontologies are useful in NLU
in two ways. First, domain knowledge often
plays a crucial role in disambiguation. A well-
designed domain ontology provides the basis
for domain knowledge representation. In
addition,ontology of a domain helps identify
the semantic categories that are involved in
understanding discourse in that domain. For
this use, the ontology plays the role of a con-
cept dictionary. In general, for NLU, we need
JANUARY/FEBRUARY 1999 23
Thing
Intangible RepresentedIndividual object
CYC
Thing
NonlivingLiving
Wordnet
Um-Thing
Element SequenceConfiguration
GUM
Thing
Process AbstractObjectConcrete
Sowa's
Figure 2. Illustration of how ontologies differ in their analyses of the most general concepts.
.
both a general-purpose upper ontology and a
domain-specific ontology that focuses on the
domain of discourse (such as military com-
munications or business stories). CYC,Word-
net,
8
and Sensus
15
are examples of sharable
ontologies that have been used for language
understanding.
Knowledge-based problem solving is the
second area in AI that is a big consumer of
knowledge. KBPS systems solve a variety of
problems—such as diagnosis, planning, and
design—by using a rich body of knowledge.
Currently, KBPS systems employ domain-
specific knowledge, which is often sufficient
for constructing knowledge systems that tar-
get specific application areas and tasks. How-
ever, even in specific application areas,
knowledge systems can fail catastrophically
when they are pushed to the edge of the capa-
bility of the domain-specific knowledge. In
response to this particular shortcoming,
researchers have proposed that problem-
solving systems need commonsense knowl-
edge in addition to domain-specific knowl-
edge. The initial motivation for CYC was to
provide such a body of sharable common-
sense knowledge for knowledge-based sys-
tems. There is a similar need for developing
domain-specific knowledge. Thus, ontology-
based knowledge-base development provides
a double advantage. The ontologies them-
selves are sharable. With these ontologies,
we can build knowledge bases using the
structure of conceptualizations to encode
specific pieces of knowledge. The knowledge
bases that we develop using these ontologies
can be shared more reliably, because the for-
mal ontology that underlies them can help
clarify the representation’s semantics.
Information systems and NLU systems
need factual knowledge about their domains
of discourse. The inferences they make are
usually simple. Problem-solving systems, in
contrast, engage in complex sequences of
inferences to achieve their goals. Such sys-
tems need to have reasoning strategies that
enable them to choose among alternative rea-
soning paths. Ontology specification in
knowledge systems has two dimensions:
Domain factual knowledge provides
knowledge about the objective realities in
the domain of interest (objects, relations,
events, states, causal relations, and so
forth).
Problem-solving knowledge provides
knowledge about how to achieve various
goals. A piece of this knowledge might be
in the form of a problem-solving method
specifying—in a domain-independent
manner—how to accomplish a class of
goals.
Most early research in KBPS mixed fac-
tual and problem-solving knowledge into
highly domain-specific rules, called domain
knowledge. As research progressed, it be-
came clear that there were systematic com-
monalities in reasoning strategies between
24 IEEE INTELLIGENT SYSTEMS
Related work
The field of ontology attracts an interdisciplinary mix of researchers,
both from academia and industry. Here we give a selection of references
that describe related ontology work. Because the literature is vast, a com-
plete list is impossible. For an extensive collection of (alphabetically
ordered) links to ontological work, including proceedings and events, see
http://www.cs.utexas.edu/users/mfkb/related.html.
Special issues on ontology
N. Guarino and R. Poli, “The Role of Ontology in the Information
Technology,Int’l J. Human-Computer Studies, Vol. 43, Nos. 5/6,
Nov.-Dec. 1995, pp. 623–965.
G. Van Heijst, A.T. Schreiber, and B.J. Wielinga, “Using Explicit
Ontologies in KBS Development,Int’l J. Human-Computer Studies,
Vol. 46, Nos. 2/3, Feb.-Mar. 1997, pp. 183–292.
M. Uschold and A. Tate, “Putting Ontologies to Use,Knowledge
Eng. Rev., Vol. 13, No. 1, Mar. 1998, pp. 1–3.
Ontology development
J. Benjamin et al., “Ontology Construction for Technical Domains,
Proc. EKAW ’96: European Knowledge Acquisition Workshop, Lec-
ture Notes in Artificial Intelligence No. 1076, Springer-Verlag,
Berlin, 1996, pp. 98–114.
W.N. Borst and J.M. Akkermans, “Engineering Ontologies,Int’l J.
Human-Computer Studies,Vol. 46, Nos. 2/3, Feb.-Mar. 1997, pp.
365–406.
A. Farquhar, R. Fikes, and J. Rice, “The Ontolingua Server:A Tool
for Collaborative Ontology Construction,Int’l J. Human-Computer
Studies,Vol. 46, No. 6, June 1997, pp. 707–728.
A. Gomez-Perez,A. Fernandez, and M.D. Vicente, “Towards a
Method to Conceptualize Domain Ontologies,Working Notes 1996
European Conf. Artificial Intelligence (ECAI ’96) Workshop on Onto-
logical Eng., ECCAI, Budapest, Hungary, 1996, pp. 41–52.
T.R. Gruber, “Towards Principles for the Design of Ontologies Used
for Knowledge Sharing,Int’l J. Human-Computer Studies,Vol. 43,
Nos. 5/6, Nov.-Dec. 1995, pp. 907–928.
R. Studer, V.R. Benjamins, and D. Fensel, “Knowledge Engineering,
Principles, and Methods,Data and Knowledge Eng., Vol. 25, Mar.
1998, pp. 161–197.
M. Uschold and M. Gruninger, “Ontologies: Principles, Methods,
and Applications,Knowledge Eng. Rev., Vol. 11, No. 2, Mar. 1996,
pp. 93–155.
Natural-language ontology
J.A. Bateman, B. Magini, and F. Rinaldi, “The Generalized Upper
Model,Working Papers 1994 European Conf. Artificial Intelligence
(ECAI ’94) Workshop on Implemented Ontologies, 1994, pp. 34–45;
http://www.darmstadt.gmd.de/publish/komet/papers/ecai94.ps.
K. Knight and S. Luk, “Building a Large-Scale Knowledge Base for
Machine Translation,Proc. AAAI ’94, AAAI Press, Menlo Park,
Calif. 1994.
G.A. Miller, “Wordnet:An Online Lexical Database,Int’l J.
Lexicography, Vol. 3, No. 4, 1990, pp. 235–312.
P.E. Van de Vet, P.H. Speel, and N.J.I. Mars, “The Plinius Ontology of
Ceramic Materials,Working Papers 1994 European Conf. Artificial
Intelligence (ECAI ’94) Workshop on Implemented Ontologies,
ECCAI,Amsterdam, 1994, pp. 187–206.
Ontologies and information sources
Y. Arens et al., “Retrieving and Integrating Data from Multiple Infor-
mation Sources,Int’l J. Intelligent and Cooperative Information
Systems,Vol. 2, No. 2, 1993, pp. 127–158.
S. Chawathe, H. Garcia-Molina, and J. Widom, “Flexible Constraint
Management for Autonomous Distributed Databases,IEEE Data
Eng. Bulletin,Vol. 17, No. 2, 1994, pp. 23–27.
S. Decker et al., “Ontobroker: Ontology-Based Access to Distributed
and Semi-Structured Information,Semantic Issues in Multimedia
Systems, R. Meersman et al., eds., Kluwer Academic Publishers,
Boston, 1999.
.
goals of similar types. These reasoning
strategies were also characterized by their
need for specific types of domain factual
knowledge. It soon became clear that strate-
gic knowledge could be abstracted and
reused.
With few exceptions,
16, 17
the domain fac-
tual knowledge dimension drives the focus
of most of the AI investigations on ontolo-
gies. This is because applications to language
understanding motivates much of the work
on ontologies. Even CYC, which was origi-
nally motivated by the need for knowledge
systems to have world knowledge, has been
tested more in natural-language than in
knowledge-systems applications.
K
BPS RESEARCHERS REALIZED
that, in addition to factual knowledge, there
is knowledge about how to achieve problem-
solving goals. In fact, this emphasis on meth-
ods appropriate for different types of prob-
lems fueled second-generation research in
knowledge systems.
18
Most of the KBPS
community’s work on knowledge represen-
tation is not well-known to the general
knowledge-representation community. In the
coming years, we expect an increased focus
on method ontologies as a sharable knowl-
edge resource.
Acknowledgments
This article is based on work supported by the
Office of Naval Research under Grant N00014-96-
1-0701. We gratefully acknowledge the support of
ONR and the DARPA RaDEO program. Any opin-
ions, findings, and conclusions or recommenda-
tions expressed in this publication are those of the
authors and do not necessarily reflect the views of
ONR. Netherlands Computer Science Research
Foundation supported Richard Benjamins with
financial support from the Netherlands Organiza-
tion for Scientific Research (NWO).
References
1. D.B. Lenat and R.V. Guha, Building Large
Knowledge-Based Systems: Representation
and Inference in the CYC Project, Addison-
Wesley, Reading, Mass., 1990.
2. B. Chandrasekaran, “AI, Knowledge, and the
Quest for Smart Systems,IEEE Expert,Vol.
9, No. 6, Dec. 1994, pp. 2–6.
3. J. McCarthy and P.J. Hayes, “Some Philo-
sophical Problems from the Standpoint of
Artificial Intelligence,Machine Intelligence
Vol. 4, B. Meltzer and D. Michie, eds., Edin-
burgh University Press, Edinburgh, 1969, pp.
463–502.
4. D. Marr, Vision: A Computational Investiga-
tion into the Human Representation and Pro-
cessing of Visual Information, W.H. Freeman,
San Francisco, 1982.
5. A. Newell, “The Knowledge Level,Artifi-
cial Intelligence,Vol. 18, 1982, pp. 87–127.
6. R. Wieringa and W. de Jonge, “Object Iden-
JANUARY/FEBRUARY 1999 25
S. Luke et al., “Ontology-Based Web Agents,Proc. First Int’l Conf.
Autonomous Agents, ACM Press, New York, 1997, pp. 59–66; http://
www.cs.umd.edu/projects/plus/SHOE/ 1997.
S.T. Polyak et al., “Applying the Process Interchange Format (PIF) to
a Supply Chain Process Interoperability Scenario,Proc. 1998 Euro-
pean Conf. Artificial Intelligence (ECAI ’98) Workshop on Applica-
tions of Ontologies and Problem-Solving Methods, ECCAI, Brighton,
England, 1998, pp. 88–96.
G. Wiederhold, “Intelligent Integration of Information,J. Intelligent
Information Systems, Vol. 6, Nos. 2/3, 1996.
G. Wiederhold and M. Genesereth, “The Conceptual Basis for Medi-
ation Services,IEEE Intelligent Systems,Vol. 12, No. 5, Sept./Oct.
1997, pp. 38–47.
Ontologies and knowledge management
A. Abecker et al., “Toward a Technology for Organizational Memories,
IEEE Intelligent Systems, Vol. 13, No. 3, May/June 1998, pp. 40–48.
V.R. Benjamins and D. Fensel, “The Ontological Engineering Initia-
tive (KA)2,Formal Ontology in Information Systems, N. Guarino,
ed., IOS Press,Amsterdam, 1998, pp. 287–301.
M.S. Fox, J. Chionglo, and F. Fadel, “A Common-Sense Model of the
Enterprise,Proc. Industrial Eng. Research Conf., Inst. for Industrial
Engineers, Norcross, Ga., 1993, pp. 425–429.
Manual of the Toronto Virtual Enterprise, tech. report, Enterprise
Integration Laboratory, Dept. of Industrial Eng., Univ. of Toronto,
Toronto, 1995.
M. Uschold et al., “The Enterprise Ontology,Knowledge Eng. Rev.,
Vol. 13, No. 1, Mar. 1998.
Task and method ontologies
D. Fensel et al., “Using Ontologies for Defining Tasks, Problem-
Solving Methods, and Their Mappings,Knowledge Acquisition,
Modeling, and Management, E. Plaza and V.R. Benjamins, eds.,
Springer-Verlag, Berlin, 1997, pp. 113–128.
J.H. Gennari et al., “Mapping Domains to Methods in Suppport of
Reuse,Int’l J. Human-Computer Studies, Vol. 41, No. 3, Sept. 1994,
pp. 399–424.
A. Tate, “Roots of SPAR—Shared Planning and Activity Represen-
tation,Knowledge Eng. Rev., Vol. 13, No. 1, Mar. 1998, pp.
121–128.
Y.A. Tijerino and R. Mizoguchi, “Multis II: Enabling End-Users
to Design Problem-Solving Engines via Two-Level Task Ontolo-
gies,Proc. EKAW ’93: Seventh European Workshop on Know-
ledge Acquisition for Knowledge-Based Systems, Lecture Notes in
Artificial Intelligence No. 723, Springer-Verlag, 1993, pp.
340–359.
Ontology workshops
Applications of Ontologies and Problem-Solving Methods, ECAI ’98
(European Conf. AI), http://delicias.dia.fi.upm.es/WORKSHOP/
ECAI98/index.html
Building, Maintaining, and Using Organizational Memories, ECAI
’98, http://www.aifb.uni-karlsruhe.de/WBS/ECAI98OM/
Formal Ontologies in Information Systems (FOIS ’98), http://krr.irst.
itc.it:1024/fois98/program.html
Intelligent Information Integration, ECAI ’98, http://www.tzi.de/
grp/i3/ws-ecai98/
Sharable and Reusable Components for Knowledge Systems, KAW
’98 (Workshop on Knowledge Acquisition, Modeling, and Manage-
ment), http://ksi.cpsc.ucalgary.ca/KAW/KAW98/KAW98Proc.html
Ontological Engineering,AAAI Spring Symp. Series, Stanford,
Calif., 1997, http://www.aaai.org/Symposia/Spring/1997/sss-97.html
Problem-Solving Methods, IJCAI ’97 (Int’l Joint Conf. AI), http://
www.aifb.uni-karlsruhe.de/WBS/dfe/PSM/main.html
Ontological Engineering, ECAI ’96, http://wwwis.cs.utwente.nl:
8080/kbs/EcaiWorkshop/homepage.html
Practical Aspects of Ontology Development, AAAI ’96
Sharable and Reusable Ontologies, KAW ’96, http://ksi.cpsc.
ucalgary.ca/KAW/KAW96/KAW96Proc.html
Sharable and Reusable Problem-Solving Methods, KAW ’96, http://
ksi.cpsc.ucalgary.ca/KAW/KAW96/KAW96Proc.html
.
tifiers, Keys and Surrogates: Object Identi-
fiers Revisited,Theory and Practice of
Object Systems (TAPOS),Vol. 1, No. 2, 1995,
pp. 101–114.
7. J.A. Bateman, B. Magini, and F. Rinaldi, “The
Generalized Upper Model,Working Papers
1994 European Conf. Artificial Intelligence
(ECAI ’94) Workshop on Implemented Ontolo-
gies, 1994, pp. 34–45; http://www.darmstadt
.gmd.de/publish/komet/papers/ecai94.ps.
8. G.A. Miller, “Wordnet: An Online Lexical
Database, Int’l J. Lexicography, Vol. 3, No.
4, 1990, pp. 235–312.
9. N. Fridman Noy and C.D. Hafner, “The State
of the Art in Ontology Design,AI Magazine,
Vol. 18, No. 3, 1997, pp. 53–74.
10. D. Fensel and V.R. Benjamins, “The Role of
Assumptions in Knowledge Engineering,
Int’l J. Intelligent Systems, Vol. 13, No. 7,
1998, pp. 715–747.
11. M.R. Genesereth and R.E. Fikes, Knowledge
Interchange Format, Version 0.3, Knowledge
Systems Lab., Stanford Univ., Stanford, Calif.
1992.
12. R. Neches et al., “Enabling Technology for
Knowledge Sharing,AI Magazine, Vol. 12,
No. 3, 1991, pp. 36–56.
13. T.R. Gruber, “A Translation Approach to
Portable Ontology Specifications,Knowl-
edge Acquisition,Vol. 5, 1993, pp. 199–220.
14. G. Schreiber et al., “CommonKADS: A
Comprehensive Methodology for KBS
Development, IEEE Expert, Vol. 9, No. 6,
Dec. 1994, pp. 28–37.
15. K. Knight and S. Luk. “Building a Large-
Scale Knowledge Base for Machine Transla-
tion,Proc. Am. Assoc. Artificial Intelligence,
AAAI Press, Menlo Park, Calif., 1994.
16. D. Fensel et al., “Using Ontologies for Defin-
ing Tasks, Problem-Solving Methods and
Their Mappings,Knowledge Acquisition,
Modeling and Management, E. Plaza and V.R.
Benjamins, eds., Springer-Verlag, New York,
1997, pp. 113–128.
17. R. Mizoguchi, J. Van Welkenhuysen, and M.
Ikeda, “Task Ontology for Reuse of Problem
Solving Knowledge,Towards Very Large
Knowledge Bases, N.J.I. Mars, ed., IOS Press,
Amsterdam, 1995.
18. J.M. David, J.P. Krivine, and R. Simmons,
Second Generation Expert Systems, Springer-
Verlag, 1993.
B. Chandrasekaran is professor emeritus, a
senior research scientist, and the director of the
Laboratory for AI Research (LAIR) in the Depart-
ment of Computer and Information Science at
Ohio State University. His research focuses on
knowledge-based systems, causal understanding,
diagrammatic-reasoning, and cognitive architec-
tures. He received his BE from Madras University
and his PhD from the University of Pennsylvania,
both in electrical engineering. He was Editor-in-
Chief of IEEE Expert from 1990 to 1994, and he
serves on the editorial boards of numerous inter-
national journals. He is a fellow of the IEEE,
AAAI, and ACM. Contact him at the Laboratory
for AI Research, Ohio State Univ., Columbus, OH,
43210; chandra@cis.ohio-state.edu; http://www.
cis.ohio-state.edu/lair/.
John R. Josephson is a research scientist and the
associate director of the Laboratory for AI
Research in the Department of Computer and
Information Science at Ohio State University. His
primary research interests are knowledge-based
systems, abductive inference, causal reasoning,
theory formation, speech recognition, perception,
diagnosis, the logic of investigation, and the foun-
dations of science. He received his BS and MS in
mathematics and his PhD in philosophy, all from
Ohio State University. He has worked in several
application domains, including medical diagnosis,
medical test interpretation, diagnosis of engineered
systems, logistics planning, speech recognition,
molecular biology, design of electromechanical
systems, and interpretation of aerial photographs.
He is the coeditor with Susan Josephson of Abduc-
tive Inference: Computation, Philosophy, Tech-
nology, Cambridge Univ. Press, 1994. Contact him
at the Laboratory for AI Research, Ohio State
Univ., Columbus, OH, 43210; jj@cis.ohio-
state.edu; http://www.cis.ohio-state.edu/lair/.
Richard Benjamins is a senior researcher and lec-
turer at the Department of Social Science Infor-
matics at the University of Amsterdam. His
research interests include knowledge engineering,
problem-solving methods and ontologies, diagno-
sis and planning, and AI and the Web. He obtained
his BS in cognitive psychology and his PhD in arti-
ficial intelligence from the University of Amster-
dam. Contact him at the Dept. of Social Science
Informatics, Univ. of Amsterdam, Roetersstraat
15, 1018 WB Amsterdam, The Netherlands;
richard@swi.psy.uva.nl; http://www.swi.psy.uva.
nl/usr/richard/home.html.
26 IEEE INTELLIGENT SYSTEMS
March/April Issue
Intelligent Agents
Guest Editor Jim Hendler of DARPA will present articles discussing development techiques for and the
practical application of intelligent agents, which might be the solution to handling the data and
information explosion brought about by the Internet. Scheduled topics include
Agents for the masses: Is it possible to develop sophisticated agents simple enough
to be practical?
Extempo Systems’ interactive characters, who engage, assist, and entertain people
on the Web
AgentSoft’s efforts at commercializing intelligent agent technology
Intelligent Systems
will also continue its coverage of the ontologies track, started in this issue, and
the track on vision-based vehicle guidance, which began in the November/December 1998 issue.
IEEE Intelligent Systems
covers the full range of intelligent system developments for the AI
practitioner, researcher, educator, and user.
IEEE Intelligent Systems
... Ontologies are semantic models which describe concepts and their relationships by establishing a domain's concept [8]. They can differ from domain-independent, more general approaches to domain-specific focused approaches [73]. Diverse backgrounds might lead to different assumptions about the usefulness of categorizations. ...
Article
The so-called Energiewende is a complex task with a variety of stakeholders, regulations, technical infrastructure, and proposed solutions. Buildings are an important sector for increasing resource and energy efficiency, as in Germany around 35 % of end energy usage can be attributed to them. Digital applications can help reduce these emissions through more efficient planning, operating, renovation, or demolition. Depending on the task and the parties involved, the complexity of descriptions and data models can vary from simple energy efficiency labels used in the labeling process of buildings to complex individual data models used for simulation. However, these specific information systems are often non-transferable, are hard to compare in their restrictions and requirements, and hence increase the overall costs of digital solutions. A good understanding is needed to preserve economic and ecological benefits while maintaining privacy and security aspects. To compare digital applications a variety of solutions are identified, namely frameworks, taxonomies, and ontologies. While Frameworks focus more on technological aspects and are complex to use, they provide an in-depth understanding. Taxonomies can be used for a less infrastructure-dependent comparison and provide simple relationships. Ontologies provide relationships and definitions while being task-dependent. We conclude that comparing the underlying data models of digital applications is a complex task and dependent on the application and its infrastructure. However, a variety of tasks refers to the same tools and data. After discussing these approaches, we then give an overview of digital applications developed by German researchers. Last we give an insight on how to combine these aspects in our ongoing research. To summarize, in this paper we give an overview of the complexity involved in transferring digital solutions in the building sector, provide a method used for comparing applications and describe a solution to compare infrastructure and digital tools built for it.
... Meaning can be added to data by using ontologies or other semantic standards, i.e., well-de ned vocabularies which allow precise and machine-readable description of knowledge about a certain domain [11]. It may enable semantic interoperability, allowing systems to make the same interpretation of the data in terms of its formal de nition [12]. ...
Preprint
Full-text available
The outcomes of a clinical research directly depend on the correct definition of the research protocol, the data collection strategy and the data management plan. Furthermore, researchers often need to work within challenging contexts, such as in Tuberculosis services, where human and technological resources for research may be rare. The use of Electronic Data Capture systems, such as REDCap and KoBotoolbox, can help to mitigate such risks and to enable a reliable environment to conduct health research and promote results dissemination and data reusability. The proposed solution was based on needs pinpointed by researchers, considering the lack of an embracing solution to conduct research in low resources environments. The REDbox framework was built to enhance collection, management, sharing and availability of data in tuberculosis research, while providing a better user experience. The relevance of this article lies in the innovative approach to support TB research by combining existing technologies and tailored supporting features. REDbox was implemented as a valuable asset in nationwide cross-institutional Tuberculosis research projects in Brazil.
... They provide the infrastructure for intelligent applications [18]. Ontologies belong to the content theory branch of Artificial Intelligence (AI) [19], and they are central for building intelligent computational agents [20]. Ontologies can minimise ambiguity and misunderstanding between stakeholders as well as lay the foundations for high-level reasoning and decision making [18,21]. ...
Article
Full-text available
The cyber threat landscape is highly dynamic, posing a significant risk to the operations of systems and organisations. An organisation should, therefore, continuously monitor for new threats and properly contextualise them to identify and manage the resulting risks. Risk identification is typically performed manually, relying on the integration of information from various systems as well as subject matter expert knowledge. This manual risk identification hinders the systematic consideration of new, emerging threats. This paper describes a novel method to promote automated cyber risk identification: OnToRisk. This artificial intelligence method integrates information from various sources using formal ontology definitions, and then relies on these definitions to robustly frame cybersecurity threats and provide risk-related insights. We describe a successful case study implementation of the method to frame the threat from a newly disclosed vulnerability and identify its induced organisational risk. The case study is representative of common and widespread real-life challenges, and, therefore, showcases the feasibility of using OnToRisk to sustainably identify new risks. Further applications may contribute to establishing OnToRisk as a comprehensive, disciplined mechanism for risk identification.
... Ontology is clearly defined as a shared concept [6]. Ontology technology has presented knowledge in a structured way by classifying objects and their properties and the analytical correlations between them within a particular domain, thereby enabling information to be amalgamated, accessed and reused [7]. It can facilitate person-to-person communication. ...
Article
Full-text available
Knowledge management in construction health and safety is an intensive process involving different stakeholders. However, this domain’s information is still fragmented and stored in various disordered formats that require systematic structure for reusing and sharing. This study aims to develop a domain ontology, HSM-Onto (Health and Safety Management-Ontology), to construct health and safety knowledge and improve health and safety management decision making. The HSM-Onto could implement the organization, storage and reuse of construction health and safety knowledge. It comprises two primary domain knowledge contexts, including construction project context and risk context. Based on the conducted analyses, the findings show that the HSM-Onto’s health and safety knowledge sharing is effective and equips health and safety employees with sound recommendations for decision making.
Article
Crop Profiles and Pest Management Strategic Plans are two sources of data that describe current and historical pest management practices for settings (e.g., agricultural commodities, schools, specialty crops, etc.) in the United States and territories. The development of documents began in 1998 as a response to the Food Quality Protection Act to ensure the collection of required data for the registration of pesticides. These documents are primary sources for government agencies, growers, crop consultants, and scientific researchers to understand and communicate production practices and issues. The documents include crop settings, priorities, worker activities, production practices, locations, pollinator protection, pests, beneficials, controls (biological, cultural, physical, and chemical), efficacy, resistance management, ecotoxicity, and timelines. Stakeholders can access these documents through the National IPM Database. The database includes functionality to develop and edit documents as well as Application Programming Interfaces to add data to the Crop Profile and Pest Management Strategic Plan documents. Current Application Programming Interface partnerships are with Bugwood and the National Pesticide Information Center. The document creation and Pest Management Strategic Plan workshop process with federal and state regulators, IR-4, state Extension professionals, industry, and grower participants is described. Potential future development of the National IPM Database is to serve as a repository for grower production guides. In conclusion, the accurate and up-to-date Integrated Pest Management data are a vital input in the regulatory process for the review of existing critical pesticides and the registration of safer alternatives.
Article
There are several lessons that can already be drawn from the current research programs on strong AI and building conscious machines, even if they arguably have not produced fruits yet. The first one is that functionalist approaches to consciousness do not account for the key importance of subjective experience and can be easily confounded by the way in which algorithms work and succeed. Authenticity and emergence are key concepts that can be useful in discerning valid approaches versus invalid ones and can clarify instances where algorithms are considered conscious, such as Sophia or LaMDA. Subjectivity and embeddedness become key notions that should also lead us to re‐examine the ethics of decision delegation. In addition, the focus on subjective experience shifts what is relevant in our understanding of ourselves as human beings and as an image of God, namely, in de‐emphasizing intellectuality in favor of experience and contemplation over action.
Article
In recent years, growing use of information technology (IT) and the big data revolution in agriculture have led many farms to adopt precision agriculture methods and to accumulate large amounts of data. Exploiting these data for effective decision support requires the ability to integrate data from various sources, to conveniently analyze the data, and to infer valuable insights. This paper presents a framework for integrating and analyzing agricultural data from various sources, which leverages cloud-computing, thereby contributing to the scalability, flexibility, affordability, and maintainability of the solution compared to existing solutions. The framework defines a functional infrastructure of cloud-based services that facilitate integration, analysis, and data visualization. These services can be either end-user applications or services intended as a platform for creating new services. To design the framework’s architecture, we applied a top-down and bottom-up approach. Based on the top-down analysis of information collected through interviews, questionnaires, and literature review, we defined an initial architecture of the framework. We then used this initial architecture to develop different applications, and based on the experience and insights gained and new requirements that were faced, the architecture was revised in an iterative process. We demonstrate the application of the framework through several use cases. Each use case represents different data integration requirements and is based on different services of the proposed framework.
Book
Second Generation Expert Systems have been a very active field of research during the last years. Much work has been carried out to overcome drawbacks of first generation expert systems. This book presents an overview and new contributions from people who have played a major role in this evolution. It is divided in several sections that cover the main topics of the subject: - Combining Multiple Reasoning Paradigms - Knowledge Level Modelling - Knowledge Acquisition in Second Generation Expert Systems - Explanation of Reasoning - Architectures for Second Generation Expert Systems. This book can serve as a reference book for researchers and students and will also be an invaluable help for practitioners involved in KBS developments.
Article
This paper describes and classifies methods to transform data to information in a three-layer, mediated architecture. The layers can be characterized from the top down as information-consuming applications, mediators which perform intelligent integration of information (I3), and data, knowledge and simulation resources. The objective of modules in the I3 architecture is to provide end users' apoplications with information obtained through selection, abstraction, fusion, caching, extrapolation, and pruning of data. The data is obtained from many diverse and heterogeneous sources. The I3 objective requires the establishment of a consensual information system architecture, so that many participants and technologies can contribute. An attempt to provide such a range of services within a single, tightly integrated system is unlikely to survive technological or environmental change. This paper focuses on the computational models needed to support the mediating functions in this architecture and introduces initial applicatiions. The architecture has been motivated in [Wied:92C].
Article
To support the sharing and reuse of formally represented knowledge among AI systems, it is useful to define the common vocabulary in which shared knowledge is represented. A specification of a representational vocabulary for a shared domain of discourse—definitions of classes, relations, functions, and other objects—is called an ontology. This paper describes a mechanism for defining ontologies that are portable over representation systems. Definitions written in a standard format for predicate calculus are translated by a system called Ontolingua into specialized representations, including frame-based systems as well as relational languages. This allows researchers to share and reuse ontologies, while retaining the computational benefits of specialized implementations.We discuss how the translation approach to portability addresses several technical problems. One problem is how to accommodate the stylistic and organizational differences among representations while preserving declarative content. Another is how to translate from a very expressive language into restricted languages, remaining system-independent while preserving the computational efficiency of implemented systems. We describe how these problems are addressed by basing Ontolingua itself on an ontology of domain-independent, representational idioms.
Article
The Defense Advanced Research Projects Agency (DARPA) and US Air Force Research Laboratory Planning Initiative (ARPI) has initiated a project to draw on the range of previous work in planning and activity ontologies to create a practically useful Shared Planning and Activity Representation (SPAR) for use in technology and applications projects within their communities. This article describes the previous work which has been used to create the initial SPAR representation. Key examples of the work drawn upon are published in this issue. The paper provides a comprehensive bibliography and related world wide web resources for work in the area of plan, process and activity representation. SPAR is now being subjected to refinement during several review cycles by a number of expert and user panels.