Upload
cherie
View
18
Download
1
Embed Size (px)
DESCRIPTION
OWL Instance Data Evaluation. Li Ding, Jiao Tao, and Deborah L. McGuinness Tetherless World Constellation Computer Science Department. Motivation. Focus of this talk OWL instance data Why is it interesting? Instance data dwarfs class data on the web - PowerPoint PPT Presentation
Citation preview
1
OWL Instance Data Evaluation
Li Ding, Jiao Tao, and Deborah L. McGuinness
Tetherless World ConstellationComputer Science Department
2
Motivation
• Focus of this talk– OWL instance data
• Why is it interesting?– Instance data dwarfs class data on the web– Instance data may be published independently from
class and property descriptions and may cause unexpected issues
– Instance data is often used by many applications, some of which may not be aware of the class and property descriptions
3
A new ontology A new instance data
Wine ontology
Semantic Inconsistencyin Distributed Publishing
wine:EarlyHarvest wine:LateHarvest
wine:Wine
rdfs:subClassOf rdfs:subClassOf
owl:disjointWith
wine:BadWineDefinition
rdfs:subClassOf rdfs:subClassOf
wi:BadWineInstance
rdf:type rdf:type
Case1: semantic inconsistencycaused by a new class.
Case2: semantic inconsistencycaused by a new instance.
4
Limited Scope of Data
Instance data A Instance Data B
Wine ontology
wine:Zinfandel
wine:Wine
rdfs:subClassOf
wi:W1wi:W1
rdf:typerdf:type
• Consider Instance data A, application may need additional instance data to meet cardinality requirement
Surfing via namespace of wine ontology
wine:hasMaker
owl:Restriction
"1"^xsd:nonNegativeInteger
rdfs:subClassOfowl:onProperty
owl:cardinality
wi:WineMaker1
wine:hasMaker
wine:Red
wine:hasColor
5
Issues with OWL instance data• General Issues
– RDF syntax– OWL DL semantic consistency
• Issues related to an individual’s type description: – Unexpected individual type: (rdfs:domain, rdfs:range,
owl:allValuesFrom)– Redundant individual type– Non-specific individual type
• Issues related to property restrictions: – Unexpected individual type: (owl:hasValue)– Missing property value (owl:minCardinality, owl:cardinality)– Excessive property value (owl:maxCardinality, owl:cardinality)
• (Customizable and Extensible) Issues related to style: – Naming conventions– Missing annotations,…
6
Implementation
• Load instance data - (D)– Is D syntactically correct?
• Load the corresponding ontologies - (O={O1,O2,…})– Is any Oi inaccessible?
• Derive the inferred model, - Inf(D+O)– Is D semantically consistent with O?
• Use Pellet OWL DL reasoner
• Check style issues– Is more explicit description about the instance needed?
• Run Sparql on (D) and Inf( D+O )– query D to confirm the issue is caused by D– query Inf(D+O) to check presence of issue
– Is there any redundancy or ambiguity in D?
7
Wine Ontology (Partial)Class(a:Wine partial restriction(a:hasSugar cardinality(1)) restriction(a:madeFromGrape minCardinality(1)) restriction(a:hasBody cardinality(1)) restriction(a:hasColor cardinality(1)) restriction(a:hasFlavor cardinality(1)) restriction(a:hasMaker cardinality(1)) restriction(a:locatedIn someValuesFrom(a:Region)) restriction(a:hasMaker allValuesFrom(a:Winery)))
Class(a:Zinfandel complete intersectionOf(restriction(a:madeFromGrape value (a:ZinfandelGrape)) a:Wine restriction(a:madeFromGrape maxCardinality(1))))
Class(a:Zinfandel partial restriction(a:hasSugar value (a:Dry)) restriction(a:hasFlavor allValuesFrom(oneOf(a:Moderate a:Strong))) restriction(a:hasBody allValuesFrom(oneOf(a:Medium a:Full))) restriction(a:hasColor value (a:Red)))
8
An Example SPARQL Query forMissing Property Value
1. PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>2. PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>3. PREFIX owl: <http://www.w3.org/2002/07/owl#>4. SELECT ?i ?p ?c5. FROM NAMED <http://tw.rpi.edu/2008/03/wine-instance.owl>6. WHERE { { ?c rdfs:subClassOf ?r.7. ?r rdf:type owl:Restriction .8. ?r owl:onProperty ?p .9. ?r owl:cardinality ?card .10. FILTER( ?card = 1 ) 11. OPTIONAL { ?i ?p ?o . }12. "FILTER( !BOUND(?o) ) }13. GRAPH <http://tw.rpi.edu/2008/03/wine-instance.owl>14. { ?i rdf:type ?c . } }
9
Conclusion
• Current work– identified OWL instance data evaluation
problem– Listed several common issues in OWL
instance data– used SPARQL + Inference to clarify issue
definition and simplify issue detection
• Future work– Use SPARQL-DL API
10
Questions?or
More Slides?
The 7th International Semantic Web ConferenceISWC 2008 • Karlsruhe DE • 26-30 October 2008
http://iswc2008.semanticweb.org/
11
A Questionnaire for OWL User Experiences
• OWL constructs– What have been used?– What are still missing?
• OWL inference– What inference has been used to solve problem? – What other inference is used together with OWL
inference, e.g. sparql, swrl?• OWL user experience
– How hard is it to build/reuse OWL ontology– How hard is it to build/reuse OWL instance data– How OWL help web users?
12
OWL Instance Data Evaluation Architecture
semantic web data evaluation services
RDF parsing and validationRDF parsing and validation
referenced ontology resolutionreferenced ontology resolution
OWL species classificationOWL species classification
RDFS/OWL semantics validationRDFS/OWL semantics validation
general style evaluationgeneral style evaluation
domain specific style evaluationdomain specific style evaluation
instance datainstance data
referenced ontologiesreferenced ontologies
URI or text of RDF document evaluation report
optionalontologiesoptional
ontologies