170 likes | 385 Views
July 2002. CSM2002, AB HS. 2. The Message. An ontology is a body of structured knowledge, shared by all stakeholders for (re-)use by man and computerTo define knowledge on modeling:Ontology of the modeling process (modeling tasks)To define knowledge on the problemOntology of the object sys
E N D
1. July 2002 CSM2002, AB + HS
1 Ontologies to structure models and modeling tasks Adrie J.M. Beulens and Huub Scholten
Wageningen Universiteit, Toegepaste Informatiekunde
2. July 2002 CSM2002, AB + HS 2 The Message An ontology is a body of structured knowledge, shared by all stakeholders for (re-)use by man and computer
To define knowledge on modeling:
Ontology of the modeling process (modeling tasks)
To define knowledge on the problem
Ontology of the object system
To develop and exchange models : need for a common model representation format --> develop first:
Ontology of (quantitative, algebraic) models
3. July 2002 CSM2002, AB + HS 3 Topics Introduction
Why ontologies?
What are ontologies?
Why ontologies to solve modeling problems
An overall picture
Discussion
4. July 2002 CSM2002, AB + HS 4 Introduction 20 years ago:
Ontology: esoteric part of philosophy, about being, about what can be mentioned (Gruninger & Lee, 2002)
Aristotele distinguished in his system theoretical philosophy (physics, ontology, logics) and practical philosophy (ethics, politics, poetics).
The ontology studies ‘what is’ as such , their nature, characteristics and mutual relations
Now (May 2002):
Google finds 310.000 websites on ‘ontology’
Term borrowed from knowledge engineering
examples
5. July 2002 CSM2002, AB + HS 5 Example sites on ‘Ontology’ Gene ontology consortium
What is an ontology?
Ontology.Org - enabling virtual business
Ontology - descriptive and formal
The ontology page
Buffalo ontology site
W3C web ontology (webont) working group
KR/DB conferences and journal cfps
Enterprise project: The enterprise ontology
Kbs/ontology projects worldwide
6. July 2002 CSM2002, AB + HS 6 Why ontologies? (1) A help in structuring complex definitions and arrive at shared understanding
Here:
what elements are needed/allowed in a model definition
what knowledge of an object system is required/essential/relevant to solve a problem using models
what knowledge/expertise on modeling is required for a Good Modeling Practice
Used for:
defining knowledge in general: Internet/WWW
defining protocols in medics and other guidelines
hierarchical knowledge (ecosystems, car repair, ….)
7. July 2002 CSM2002, AB + HS 7 Why ontologies? (2) Communication:
Between people
Between computers/systems
Between man and computer
Re-use of knowledge
Make assumptions Explicit
Ordering and structuring of knowledge
Analysing of knowledge
8. July 2002 CSM2002, AB + HS 8 What are ontologies? Two definitions:
Gruber 1994 (for AI systems):
A formal specification of a shared conceptualisation
(Concept = what “exists” is that which can be represented)
Borst, 1997:
An ontology is a formal specification of a shared conceptualisation
Ontologies consists of
Concepts (things you can discuss)
Relations between concepts (consists of, must be preceded by, etc.)
Functions (relations with 1 result)
Instants (specific concept, not generic)
Axiomata (knowledge on concepts/relations that can be checked on its logics)
9. July 2002 CSM2002, AB + HS 9 Representations of ontologies Languages:
Old: ontolingua , KIF, OKBC, etc.
New (xml-based): XOL, RDF, OIL, DAML+OIL
Tools:
Old: server in Stanford University Knowledge Systems Laboratory
New:
Protegé2000
OilEd
OntoEdit
Examples ‘formats’: internet
10. July 2002 CSM2002, AB + HS 10 The role of ontologies in MBDS MB-DS = Model Based Decision Support
Discussed here ontologies on
Problem/object-system ontology
Model ontology
Modeling ontology
How are these related
11. Example (greenhouse): diagram
12. Example (greenhouse): ontology in XML/RDF
13. July 2002 CSM2002, AB + HS 13 Problem/object system ontology Making explicit:
What we are interested in in the OS + reason (problem)
Application domain (which knowledge / science / theories etc.)
Problem owner
Problem:
type of application
planning
design
research
operational management
description
Existing models
14. July 2002 CSM2002, AB + HS 14 Model ontology What is a model ontology?
ontology of concepts and mutual relations describing the structure of models (within a modeling paradigm?)
What part of a OS ontology is reflected in a model (paradigm) ontology and how is it reflected?
Examples of model ontologies:
Jan Top (ATO): physical models
to be done (Huub Scholten)
ontology on eco-physiological processes of bivalves (+instances)
RWS/RIKZ:
generic model for estuarine ecosystems
15. July 2002 CSM2002, AB + HS 15 Modeling ontology Examples of preliminary steps towards a modeling ontology:
GMP handbook
NEN-norms for modeling water management in the Netherlands
Process oriented: tasks and mutual dependencies
name of task
what? definition
who? modeller, client stakeholder
how? (advice on) methods
which problems to be expected? pitfalls & sensitivities
16. MB-DS-ontology
17. July 2002 CSM2002, AB + HS 17 Projects related to MB-DS-ontology Dutch GMP handbook
Norms for model use for Dutch water management
HarmoniQuA
ATO
AMEPS
18. July 2002 CSM2002, AB + HS 18 Discussion How consistent is the presented approach?
Object system ontology
Model ontology
Modeling ontology
How to deal with domain specific aspects?
Representation formats important?