Ontology & Semantic Web – A Dummy's Overview of Modern ...

Ontology & Semantic Web
–

A
Dummy’s Overview of Modern
Technologies for Sharing Knowledge

Mitsunori Ogihara

Center for Computational Science

What Is an Ontology?

•
Merriam
-
Webster: “The branch of
metaphysics dealing with the nature of being”

–
What does it mean to exist?

–
What exists?

•
In the field of computer science an ontology is
“a specification of a conceptualization”
–

Tom
Gruber

World, Specification,
Conceptualization

•
Human observes
the world and
conceptualizes it

•
That human
conceptualizatio
n is put into a
specification

•
The world
matches the
specification

What an Ontology Can Conceptualize

•
Things to exist

–
Individuals, not necessarily physical existence

–
Classes of individuals

•
Relations among things

–
Is a part of

–
Is not equal to

•
Properties about things

–
Has a value of


Problem

•
Conceptualization is ambiguous and
inaccurate

–
How a person A sees the world is not necessarily
equal to how a person B sees the world

•
Specification is difficult

–
Formal specification is tiresome


•
How efficiently can one develop an ontology?

•
How efficiently can one compare
ontologies
?

Why Was the Idea of Ontology
Created?

•
Artificial Intelligence … a branch of computer
science that studies computational methods
of mimicking human intelligence

•
Intelligence includes ability to

–
Understand data obtained through senses

–
Acquire knowledge

–
Apply knowledge to solve problems

–
Understand emotion

Knowledge Representation

•
An area that studies how to formally think

–
[Davis,
Shrobe
, and Solovitz’93] Knowledge
Representation is

•
A surrogate

•
A set of ontological commitments

•
A fragmentary theory of intelligent reasoning

•
A medium for efficient computation

•
A medium of human expression

–
Commitments are filters through which the world
is observed

Semantic Web

•
The first generation of Web is HTML
(Hypertext Markup Language)

–
This is designed so as to present texts in a format
specification that can be easily understood and
rendered

–
Search engines can find documents that may
contain certain information by using keyword
matches, but can’t find an answer to a question

Semantic Web

•
A new generation of web should provide not texts but
structured information, a part of which may be texts

–
Resource Description Framework (where the resources
are)

–
XML (Extensive Markup Language)

•
A user
-
definable format

•
Documents conforming to the format

•
Idea:

–
Decide on what information can a web page might contain

–
Decide on how to describe such information

–
Annotate the web page with such information in a
predetermined format

Ontology Development Tools

•
OWL (Web Ontology Language)

–
Currently the most popular ontology description
language

–
http://www.w3.org/TR/owl
-
features/

–
OWL DL (Description Logic, standard version)

–
OWL
Lite

(restricted version) … basic constructs
exist to logically express constructs of DL

–
OWL Full (for RDF)

–
http://www.cs.manchester.ac.uk/~horrocks/ISWC
2003/Tutorial/examples.pdf

A History of Ontology Description
Languages

•
KIF (1992) … Stanford, first
-
order logic

•
Loom (1992) … USC, first
-
order logic, for KR
nor necessarily for ontologies

•
FLogic (1995) … Karlsruhe, combination of
first
-
order logic and frames

•
OKBC (1997) … DARPA

•
XOL (1999) … SRI, an XML version of OKBC

•
OWL (2001) … W3C

Ontology Development Tools

•
Created along with development of
description languages

Popular Free Tools

•
Protégé
-
2000

•
Swoop … an open source project, hosted at
Google

Ontology Building Process

•
Vocabulary

–
Need to settle on a set of words to be used to describe the
domain knowledge (or the domain of the web contents)

–
Where to start? Thousands of words?

•
Knowledge Base Building

–
Express domain experts’ knowledge in terms of ontology

–
Who will translate knowledge into logical forms?
Ambiguity issues?

•
Inference

–
Make new discovery

–
Identify classes and properties of an individual

–
Inference engines, compute
-
intensive

Exporting Ontologies

•
Protégé and Swoop (and others) have the
ability to export/import data in various
formats

–
Enables information exchange between ontologies


Finding a Nice Mapping

•
A mapping
f

of an ontology O to an ontology O’ is one that
maps each class of O to a class of O’ and each property of O
to another property of O’. We want:

–
For all classes
c

and
d

of O,
c

is a subclass of
d

if and only if
f(c
) is

a subclass of
f(d
) in O’

–
For all class
c

and property
p

of O,
c

has property
p

if and only if
f(c
) has property
f(p
) in O’

•
Finding a perfect mapping is hard, and practically such a
perfect mapping rarely exists

•
Finding a mapping that maximizes a certain quantity is also
difficult, and is NP
-
hard

–
Heuristic methods are usually used (based on graph properties)

References

•
T.R.Gruber

(1993), A Translation Approach to Portable Ontology
Specifications, Knowledge Acquisition

•
V.Devedzik
(2002), Understanding ontological engineering,
Communications of the ACM

•
J.Gennari
,
M.Musen
,
R.Fergerson

(2003), The evolution of Protégé:
an environment for knowledge
-
based systems development,
International Journal of Human
-
Computer Studies

•
A.Kalyanpur
,
B.Parsia
,
E.Sirin
,
B.Grau

(2006), Swoop: A web
ontology editing browser, Web Semantics: Science

•
O.Corcho

et al. (2003), Methodologies, tools and languages for
building
ontologies
. Where is their meeting point?
Data&Knowledge

Engineering

•
L.Lacy

(2005), OWL: Representing information using the web
ontology language

•
J.Euzenat
, P.
Shvaiko

(2007), Ontology Matching, Springer