SPARQL Inference with SKOS

427 views Asked by At

We are trying to showcase inference with linked-data.

The simple graph looks like the following in turtle-format:

PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
PREFIX ex: <http://schema.example.com/>
PREFIX skos: <http://www.w3.org/2004/02/skos/core#>

ex:Places rdf:type skos:ConceptScheme .

ex:Localities rdf:type skos:Concept .
ex:Localities skos:prefLabel "Localities" .
ex:Localities skos:inScheme ex:Places.

ex:Countries rdf:type skos:Concept .
ex:Countries skos:prefLabel "Countries" .
ex:Countries skos:inScheme ex:Places.

ex:Continents rdf:type skos:Concept .
ex:Continents skos:prefLabel "Continents" .
ex:Continents skos:inScheme ex:Places.

ex:Persons rdf:type skos:Concept .
ex:Persons skos:prefLabel "Persons" .

ex:livesIn a rdf:Property .

ex:isPartOf a rdf:Property .

ex:Localities skos:broader ex:Countries .
ex:Countries skos:broader ex:Continents .

ex:Europe a ex:Continents .

ex:Switzerland a ex:Countries .
ex:Switzerland ex:isPartOf ex:Europe.

ex:France a ex:Countries .
ex:France ex:isPartOf ex:Europe.

ex:Bern a ex:Localities .
ex:Bern skos:prefLabel "Bern".
ex:Bern ex:isPartOf ex:Switzerland.
 
ex:Thun a ex:Localities .
ex:Thun skos:prefLabel "Thun".
ex:Thun ex:isPartOf ex:Switzerland.

ex:Paris a ex:Localities .
ex:Paris skos:prefLabel "Paris".
ex:Paris ex:isPartOf ex:France.

ex:Hans a ex:Persons.
ex:Hans skos:prefLabel "Hans".
ex:Hans ex:livesIn ex:Bern.

ex:Fritz a ex:Persons.
ex:Frits skos:prefLabel "Fritz".
ex:Fritz ex:livesIn ex:Thun.

ex:Jaques a ex:Persons.
ex:Jaques skos:prefLabel "Jaques".
ex:Jaques ex:livesIn ex:Paris.

The Idea would be to do the following query in SPARQL:

PREFIX ex: <http://schema.example.com/>
ASK where { ex:Hans ex:livesIn ex:Switzerland }

It should return YES, but it returns NO.

Is there a possibility to model the data, that this ASK-Statement can get answered with YES?

In terms of inference it should be True, since ex:Bern is a ex:Localities and this is a skos:broader of ex:Countries, which is ex:Switzerland.

I haven't found any good examples for skos-Modelling and Inference yet. We would like to work with skos:concepts and rather not with rdfs-subclassing. Because it should showcase the inference over concepts in situations, where you are not able to subclass everything.

I am using:

  • GraphDB 9.5.0 EE
  • OWL2-RL - Ruleset
1

There are 1 answers

0
Damyan Ognyanov On BEST ANSWER

To complete the question, I'm posting my comment above as an answer...

To make it work, You need to define some meaning to your properties ex:isPartOf and ex:livesIn. Suggest to make ex:isPartOf transitive and then to define ex:livesIn as a property chain over ex:isPartOf, e.g.:

ex:isPartOf a owl:TransitiveProperty . 
ex:livesIn rdf:type owl:ObjectProperty;              
    owl:propertyChainAxiom( ex:livesIn ex:isPartOf) .