Practical difficulties in
the construction of
ontologies and
different kinds of
knowledge
representation in the
Semantic Web
[email protected]
What is a knowledge representation in
the digital world?
•Itentity
is a surrogate, a substitute for the thing itself, used to enable an
to determine consequences by thinking rather than acting
•should
It is a set of ontological commitments, to answer in what terms
we think about a domain
•Itthree
is a fragmentary theory of intelligent reasoning, expressed in terms of
components: a. the representation's fundamental conception of
intelligent reasoning; b. the set of inferences the representation
sanctions; c. the set of inferences it recommends
•about
It is a medium of human expression, a language in which we say things
the world
•It is a medium for pragmatically efficient computation
Ref: Davis, R., Shrobe, H., and Szolovits, P. What is a Knowledge
Representation? AI Magazine, 14(1):17-33, 1993.
http://groups.csail.mit.edu/medg/ftp/psz/k-rep.html
What is a knowledge representation in
the Semantic Web ?
•
•
•
Description of content and formal aspects of web resources
This description is expressed by a metadata structure in a markup
language: RDF, Resource Description Format
In a way understandable by machines
There is not only one structure of knowledge representation.
There are different structures and elements that belong to
different languages
So Semantic Web knowledge representation formalisms
consist of different kinds of logics in different kinds of
representation formats
Semantic web: logics, tools and levels of Knowledge
Representation
Le ve ls
1. Resource
description:
data and
metadata
2. Form and
Content:
human
classification
and
catalogation
Application tools
HTML, XML. RDF,
DC
CDU, ISBD,
Dew ey,
Thesauri
Conce ptual tools
We b re s ource s
m ark up le nguage s
Controlle d
language s and
catalogation
s tandards
Artificial language s
OWL(Ontologie s )
3. computers
interactivity:
middleware
Logic tools
De s cription
logic
Inde xing
logic
Obje ct
orie nte d
logic
RDF - SKOS
Domain experts: knowhow in the field to be
ontologically modeled
© Mela Bosch2006
D
e
c
l
a
r
a
t
i
v
e
K
n
o
w
l
e
d
g
e
r
e
p
r
e
s
e
n
t
a
t
i
o
n
First level of knowledge representation
D
e
c
l
a
r
a
t
i
v
e
Se m antic w e b: logics , tools and le ve ls of Know le dge Re p re s e ntation
Le ve l 1
Le ve ls
1. Resource
description:
data and
metadata
Application tools
HTML, XML. RDF,
DC
© Mela Bosch2006
Conce ptual tools
We b r e s our ce s
m ark up le ng uag e s
Logic tools
De s cr iption
lo gic
K
n
o
w
l
e
d
g
e
r
e
p
r
e
s
e
n
t
a
t
i
o
n
Methodologies for intelligent
systems
Procedural  knowledge integrated in the program.
Advantages : a great specificity: algorithms for each case.
Disadvantages : lack of versatility, difficulty to modify.
Declarative  knowledge representation is independent of
the computational processes.
Advantages: flexible and with hard logical base.
Disadvantages: great level of abstraction, difficulty to maintain a
consistent logic.
Semantic web: declarative knowledge
representation using metadata
The metadata is expressed in
Markup language:
<rdf:RDF
xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
</rdf:RDF>
Description logic also known as terminological logic
•controlled language in which the syntax is independent of
•
system procedures and the terminology is adequate to
application domain
well defined semantics, very expressive
Markup language: procedimental
specifity and declarative abstraction
Se m antic w e b: logics , tools and le ve ls of Know le dge Re pre s e ntation
Le ve ls
1. Resource
description:
data and
metadata
2. Form and
Content:
human
classification
and
catalogation
Application tools
HTML, XML. RDF,
DC
CDU, ISBD,
Dew ey,
Thesauri
Conce ptual tools
We b re s ource s
m ark up le nguage s
Controlle d
language s and
catalogation
s tandards
Logic tools
De s cription
logic
Inde xing
logic
Second level of knowledge
representation: well known for
library science professionals
© Mela Bosch2006
D
e
c
l
a
r
a
t
i
v
e
K
n
o
w
l
e
d
g
e
r
e
p
r
e
s
e
n
t
a
t
i
o
n
Second level of knowledge
representation:
another type of logic
The logic of indexation for cataloguing and classifying
But it is not the same to index
an object and to index something which is
the reference to that object
Semantic web indexes resources web
That is to say, objects, not
material objects, but digital
objects.
Objects are described through an ontology
Reference items of the digital objects: authoring, date, etc.
Similar description to library science classification
systems
Others aspects of the digital objects such as:
attributes, behavior, relationships and cardinality are
expressed by another logic
Semantic web: logics, tools and levels of Knowledge
Representation
Le ve ls
1. Resource
description:
data and
metadata
2. Form and
Content:
human
classification
and
catalogation
Application tools
HTML, XML. RDF,
DC
CDU, ISBD,
Dew ey,
Thesauri
Conce ptual tools
We b re s ource s
m ark up le nguage s
Controlle d
language s and
catalogation
s tandards
Artificial language s
OWL(Ontologie s )
3. computers
interactivity:
middleware
Logic tools
De s cription
logic
Inde xing
logic
Obje ct
orie nte d
logic
RDF - SKOS
© Mela Bosch2006
has become the mainstream
technique in the software industry
D
e
c
l
a
r
a
t
i
v
e
K
n
o
w
l
e
d
g
e
r
e
p
r
e
s
e
n
t
a
t
i
o
n
Third level of knowledge representation:
Object oriented modeling paradigm
Object-oriented point of view:
Computer program: a set of
interacting individual units, or
objects,
that manage their own state
and operations
Opposed to a traditional view in
which a program may be seen
as a collection of functions, or
a list of instructions given to
the computer.
Third level of knowledge representation:
Object oriented logic to:
•
•
•
make domain assumptions explicit
separate domain knowledge from the operational
knowledge
using terms for representing concepts that are an
abstraction of objects’ main properties
Advantages
Disadvantages
Intuitive: direct mapping from the real-world
Difficult to build large coherent and complete
representation
Hand crafting object hierarchy
Object oriented logic: Fundamental
concepts
Class — the unit of definition of data and behavior for some
kind-of-thing.
Object — an instance of a class, an object
Encapsulation — a type of privacy applied to the data,
ensures that an object can be changed only through
established channels
Inheritance — provides a way to define a (sub)class as a
specialization or subtype or extension of a more general class
Abstraction — the ability to ignore the details of an object's
(sub)class and work at a more generic level when appropriate
Polymorphism — polymorphism is behavior that varies
depending on the class in which the behavior is invoked, that
is, two or more classes can react differently to the same
message.
Ref: wikipedia.org
Object-Oriented Modeling and Ontology Engineering
There are steps in
common:
•an iterative process
•Defining concepts in
the domain (classes)
•Arranging the
concepts in a hierarchy
(subclass-superclass
hierarchy)
Defining which
attributes and
properties classes can
have and constraints on
their values
Defining individuals
and filling in slot values
•
Concept
“Jaguar“
Abstract Class: anim al
Abstract Class: Car
Class: fe line
•
subclass:
Jaguar, spots
Subclass:
Jaguar, luxe
Different: attributes,
properties and values
Object-Oriented Modeling and Ontology
Engineering
The same logic but:
An ontology
reflects the structure of the world
is often about structure of concepts
actual physical representation is not an issue
•
•
•
An Object Oriented class structure
reflects the structure of the data
is usually about behavior
describes the physical representation
of data (long int, char, etc.)
•
•
•
(Ref: http://protege.stanford.edu)
Object Oriented logic for construction of ontologies
and Object Oriented software: the same methodological
premises as document classification systems
.
Top-down: define the most general concepts first and then
specialize them
Bottom-up: define the most specific concepts and then organize
them in more general classes
Combination: define the more salient concepts first and then
generalize and specialize them
Document classification systems
Concepts
•
•parts of the concepts
•How they relate to other
Ref: http://www.ncess.ac.uk/insight/tutorials/datagrids/data_sh/ontologies/
concepts
Object Oriented logic and document classification
• concepts
• parts of the concepts
• How
they relate to other
concepts
• properties
(attributes): contain
primitive values (strings,
•
numbers)
complex properties: contain (or
point to) other objects
Hybrid knowledge
representation :
supports a rich knowledge
model with different kinds of
representation at the same
time
Ref: http://www.ncess.ac.uk/insight/tutorials/datagrids/data_sh/ontologies/
Hierarchy: French wines
Examples from: Protégé: a graphical ontology- development
tool,
open-source and freely available: (http://protege.stanford.edu)
.
Class structure usually constitute a
taxonomic hierarchy:
Class
Subclass
:
Hierarchy: Pizza
Slots in a class definition describe
attributes of instances of the class
and relations to other instances:
Each wine will have color, sugar
content, producer, etc.
Types of properties
“intrinsic” properties: flavor and color of wine
•
•“extrinsic” properties: name and price of wine
•parts: ingredients in a dish
•relations to other objects: producer of wine
(winery)
Terminology
Class=concept
Instance= object
Slot= property
Facet=values
Slot cardinality = the number of values a
slot has (common facet)
Slot value type = the type of values a slot
has (common facet): string, num, boolean
Minimum and maximum value = a range of
values for a numeric slot (common facet)
Default value = the value a slot has unless
explicitly specified otherwise(common facet)
Superclass
Wine
Subclass
French wine
Class instance
creation
A subclass inherits all the slots
from the superclass, but with a
list of own allowed values
Examples from:
Ref: http://www.ncess.ac.uk/insight/tutorials/datagrids/data_sh/ontologies/
http://protege.stanford.edu/publications/ontology_development/ontology101.html
Common problems: Is a Margherita Pizza a
Vegetarian Pizza?
Errors in understanding common logical constructs
A class can
have more
than one
superclass
A subclass
inherits slots
and facet
restrictions
from all the
parents
Different
systems
resolve
conflicts
differently
Ref: http://www.co-ode.org
Common problems: Is a Margherita Pizza a Vegetarian Pizza?
Different subclasses
Closure Axiom: only
Correct hierarchy
Ref: http://www.co-ode.org
•Semantic Web knowledge representation formalisms
Conclusion
consist of different kinds of logics in different kinds of
representation formats
there are many aspects involved such as description
logic, classification systems, object oriented data
structure and markup languages
•
“Every ontology is a treaty
Gruber says: – a social agreement – among
people with some common
motive in sharing”
Ref: http://www.sigsemis.org/newsletter/october2004/tom_gruber_interview_sigsemis
Collaborative approach to construction of ontologies
•Communication between collaborators from
different disciplines is difficult
•Ontology Building, maintenance and reuse:
time consuming activities, cost analysis is
complex
Web developers:
markup language:
RDF, OWL, SKOS
Software Engineering:
Object Oriented Design
Collaborative
approach to
construction of
ontologies
Library Science and
Terminology:
controlled languages
and classification
schemes
Domain experts: knowhow in the field to be
ontologically modeled
Conclusion
•Methodologies for collaboratively creating and managing
shared information: Modeling semantically heterogeneous
data sources and services, Representing and reasoning with
ontologies and mappings between ontologies
Semantic community support systems and collaboration
applications: Groupware tools for supporting collaborative
ontology design, Semantic Wikis, semantic blogging,
Case studies and experience reports on semantics-aware
collaborative applications
Cost Estimation Models for Ontology Engineering
•
•
•
References
Examples of ontologies:
ProtegeOntologiesLibrary
http://protege.cim3.net/cgibin/wiki.pl?ProtegeOntologiesLibrary
•Hodgson,
Ralph; Keller, Paul. Collaborative Ontology-Based Systems. Innovator Perspectives and Demonstrations of New
Open Standards and Technologies in Support of Ontology Engineered Solutions. TopQuadrant and NASA Ames. Collaborative
Expedition Workshop #38. February 22, 2005 at NSF Semantic Conflict, Mapping, and Enablement: Making Commitments Together.
http://www.topquadrant.com/documents/talks/TQ%20OntologyBased%20Collaborative%20Environments%20(v4).pdfBased%20Collaborative%20Environments%20(v4).pdf
• Díaz, Alicia, Baldo, G. CO-Protégé: A Groupware Tool for Supporting Collaborative Ontology Design with Divergence. Lifia,
Fac. Informática- UNLP, La Plata, Argentina- Loria, Campus Scientifique, Vandœuvre-lès-Nancy cedex, France.
http://protege.stanford.edu/conference/2005/slides/6.2_A.Diaz_Co-Protege_slices_and_flyer.pdf;
http://protege.stanford.edu/conference/2005/submissions/abstracts/accepted-abstract-diaz.pdf
•Gamper, Johann, Nejdl, Wolfgang; Wolpers, Martin. Combining Ontologies and Terminologies in Information Systems.
European Academy Bolzano/Bozen, Scientific Area ``Language and Law''Bozen, Italy - Institut für Rechnergestützte
Wissensverarbeitung University of Hannover, Germany. http://www.kbs.uni-hannover.de/Arbeiten/Publikationen/1999/tke99/
•W3C Working Draft 07 March 2002, Requirements for a Web Ontology Language. Latest version:http://www.w3.org/TR/webontreq/
•Institut für Informatik, Freie Universität Berlin. Ontology Engineering Cost Estimation with ONTOCOM. http://ontocom.agnbi.de//index.html
[email protected]
Descargar

Practical difficulties in the construction of ontologies