An Extensible Approach for Modeling Ontologies in RDF(S)

About This Presentation
Title:

An Extensible Approach for Modeling Ontologies in RDF(S)

Description:

Ideal Modeling (WYMIWYG) Modeling not constrained by any language ... For Semantic Web and DAML more than light-weight is needed! Axioms ... –

Number of Views:32
Avg rating:3.0/5.0
Slides: 19
Provided by: michael2507
Category:

less

Transcript and Presenter's Notes

Title: An Extensible Approach for Modeling Ontologies in RDF(S)


1
An Extensible Approach for Modeling Ontologies in
RDF(S)
  • Steffen Staab, Michael Erdmann, Alexander
    Mädche, Stefan DeckerResearch Group Knowledge
    Management
  • Institute AIFB, University of Karlsruhe,
  • DB Group, Stanford UniversityLisbon,
    September 21, 2000

2
What is an Ontology?
  • Light-weight Ontology
  • concepts, atomic types
  • is-a hierarchy among concepts
  • associations between concepts
  • Heavy-weight Ontology
  • cardinality constraints
  • taxonomy of relations
  • reified statements
  • Axioms / semantic entailments of various tastes
  • expressiveness (DL, propositional, horn, or first
    order logic, higher order)
  • inferences

3
Tools for Ontologies
  • Light-weight
  • uncontroversial
  • all Tools support light-weight
  • Protege, Stanford
  • OntoEdit, Karlsruhe
  • UML-Tools, several
  • Heavy-weight
  • no consensus yet
  • layering seems appropriate/necessary

4
Concepts, Relations, ....
5
Modeling
  • Ideal Modeling (WYMIWYG)
  • Modeling not constrained by any language
  • All appropriate epistemological primitives and
    modeling styles should be usable
  • Real Modeling
  • A particular language always restricts allowed
    primitives (modeling language)
  • A particular language is needed in
    applications(application language)
  • gt distinguish modeling language from final
    application language

translate automatically
6
Axioms
  • For Semantic Web and DAML more than light-weight
    is needed!
  • Axioms
  • Framework for conceptual modeling of axioms
  • Ontology of axiom patterns
  • language specific axiom-schemata can work with
    that knowledge
  • Interoperability is an issue
  • RDF / RDFS seem appropriate

next slide
next but one slide
7
Axiom Patterns
  • 1. Axioms for a relational algebra
  • (a) Reflexivity of relations
  • (b) Symmetry of relations
  • (c) Asymmetry of relations
  • (d) Transitivity of relations
  • (e) Inverse relations
  • (f) Irreflexivity of relations
  • (g) Antisymmetry of relations
  • 2. Composition of relations
  • 3. (Exhaustive) Partitions of Concepts

8
Axiom Patterns
  • 4. Axioms for subrelation relationships
  • 5. Axioms for part-whole reasoning Winston 87
    Chaffin 92
  • PhysicalPartOf
  • MemberOf
  • PortionOf
  • PhaseOf
  • FeatureOf
  • SubRegionOf
  • 6. Nonmonotonicity
  • 7. Axioms for temporal and modal contexts
  • 8. (General axioms (application specific, ad hoc))

9
(No Transcript)
10
Example of Composition of Relations
ltoComposition rdfID"FatherInLawComp"gt
ltocomposee rdfResource"fatherInLawOf"/gt
ltofirstComponent rdfResource"fatherOf"/gt
ltosecondComponent rdfResource"marriedWith"/gt lt/
oCompositiongt
rdfsProperty
rdfClass
oComposition
oRelation
Composition(fatherInLawOf, fatherOf, marriedWith).
oIrreflexiveRel
forall R,Q,S,X,Y,Z XR -gtgt Z lt- Composition(R,
Q, S) and XQ -gtgt Y and YS -gtgt Z.
osecondComponent
ocomposee
applmarriedWith
ofirstComponent
forall X,Y,Z XfatherInLawOf -gtgt Z
lt- XfatherOf -gtgt Y and YmarriedWith -gtgt Z.
applfatherOf
applfatherInLawOf
11
OntoEdit supports Axiom Classification
fatherInLawOf
fatherOf
marriedWith
12
Ontology Engineering using OntoEdit
  • Interaction with the user on a conceptual level
  • Multiple views for concepts, relations and axioms
  • Multilinguality
  • Linkable to NLP domain lexicon
  • Exports ontology (incl. axioms) into several
    formats
  • F-Logic (main language)
  • RDF/RDFS
  • DTDs (as far as possible)
  • ORDB-Schema (as far as possible)
  • OIL (partially and in RDF)
  • UML/XMI (planned)
  • the DAML language (when specified -)

13
Ontoedit
14
Frame-Logic Inference Engine Access
Automatically Derived from Axiom views
Pure F-Logic
Debugging Instances Rule Debugging
15
FaCT DL Engine Interface
  1. Generate FaCT LISP KB (future OIL)
  2. Call FaCT Client, transform ontology on FaCT
    server
  3. Ask server

16
Flexible Epistemological Level
XRDF
OIL
DAML
17
Conclusion
  • No Method fits all
  • Different applications need
  • different representation languages
  • with their underlying reasoning services
  • Ontology development must be aware of this
  • conceptual modeling
  • mechanisms to access/integrate several ontologies
  • distributed on the web
  • identified by (XML-) namespaces

18
Thank You!
Write a Comment
User Comments (0)
About PowerShow.com