Reasoning LEVEL RDFS

1 view
Skip to first unread message

Maatary Okouya

unread,
Dec 18, 2014, 11:22:06 AM12/18/14
to sta...@clarkparsia.com
Hi, 


just for clarification.  In the doc it is written:


RDFS

For the OWL 2 axioms allowed in RDF schema (mainly subclasses, subproperties, domain, and ranges).

This is to be interpreted as a specific Stardog profile of OWL2, limited to RDFS Axioms. 

ONE should not expect a pure RDFS REASONING but  OWL2 reasoning at the RDFS LEVEL. 


For instance, in RDFS, rdfs:label is a rdf:property and that is it. While in OWL2, it also a owl:annotationProperty. 

Meaning while a pure RDFS reasoner would reason over rdfs:label and its sub-property, an OWL2 reasoner would not. 




I would like to confirm that because this would explain the problem that i have with SKOS and STARDOG. Simply having the statement  A skos:prefLabel  "B" do not entail  A rdfs:label "B". 


This is of a particular importance, because if one expect to specialize things like alternativeLabel such as in:

 ex:code rdfs:subpropertyOf skos:AltLabel

Then later on querying all the alternative label won't include the one coming from ex:code


Best,

Daniel


Héctor Pérez-Urbina

unread,
Dec 18, 2014, 12:12:34 PM12/18/14
to stardog
On Thu, Dec 18, 2014 at 11:22 AM, Maatary Okouya <maatar...@gmail.com> wrote:
Hi, 


just for clarification.  In the doc it is written:


RDFS

For the OWL 2 axioms allowed in RDF schema (mainly subclasses, subproperties, domain, and ranges).

This is to be interpreted as a specific Stardog profile of OWL2, limited to RDFS Axioms. 

ONE should not expect a pure RDFS REASONING but  OWL2 reasoning at the RDFS LEVEL. 

Correct. RDFS does not refer to the RDFS entailment regime.
 


For instance, in RDFS, rdfs:label is a rdf:property and that is it. While in OWL2, it also a owl:annotationProperty. 

Meaning while a pure RDFS reasoner would reason over rdfs:label and its sub-property, an OWL2 reasoner would not. 




I would like to confirm that because this would explain the problem that i have with SKOS and STARDOG. Simply having the statement  A skos:prefLabel  "B" do not entail  A rdfs:label "B". 


This is of a particular importance, because if one expect to specialize things like alternativeLabel such as in:

 ex:code rdfs:subpropertyOf skos:AltLabel

Then later on querying all the alternative label won't include the one coming from ex:code

If such an axiom were part of your TBox, asking for skos:AltLabel with RDFS should include the assertions made with ex:code. Is this not the case?


 


Best,

Daniel


--
-- --
You received this message because you are subscribed to the C&P "Stardog" group.
To post to this group, send email to sta...@clarkparsia.com
To unsubscribe from this group, send email to
stardog+u...@clarkparsia.com
For more options, visit this group at
http://groups.google.com/a/clarkparsia.com/group/stardog?hl=en



--
Best,
Héctor

Maatary Okouya

unread,
Dec 18, 2014, 2:26:16 PM12/18/14
to sta...@clarkparsia.com, hec...@clarkparsia.com
You are right it worked. 

Still i'm puzzled. 


So let me try to figure out something with you. I have simplified my previous example


Here is my very very simple SKOS dataset:

<http://iadb/skosTraning2>

  rdf:type owl:Ontology ;

  owl:imports <http://www.w3.org/2004/02/skos/core> ;

.

:DC

  rdf:type skos:Concept ;

  skos:prefLabel "DC"@en ;

.

:USA

  rdf:type skos:Concept ;

  :relatedCity :DC ;

  skos:prefLabel "USA"@en ;

.

:relatedCity

  rdf:type owl:ObjectProperty ;

  rdfs:subPropertyOf skos:related ;

  skos:prefLabel "related city"^^xsd:string ;

.


I query the dataset as such:

select ?concept (rdfs:label AS ?prop) ?label
where {
?concept a skos:Concept .
  
  ?concept rdfs:Label ?label
}


I get nothing as an answer. I find it very strange given that we have the same situation as before. The only difference is that, i explicitly define:

ex:code rdfs:subpropertyOf skos:AltLabel

while in this case 

skos:prefLabel rdfs:subpropertyOf  rdfs:label

is defined in SKOS which is supposed. I suppose that the version of skos of the archetype contain that axiom or does it not ?

In any case, why would it be different ? Why do i get nothing as a result ?

Maatary Okouya

unread,
Dec 18, 2014, 2:30:06 PM12/18/14
to sta...@clarkparsia.com, hec...@clarkparsia.com
What is the subsequent steps that one shall take when using the SKOS archetype. Do I need to load SKOS ontology anyway ? simply activating the archytype does not means that the skos ontology is part of the TBOX in an implicit fashion ?

Maatary Okouya

unread,
Dec 18, 2014, 3:44:25 PM12/18/14
to sta...@clarkparsia.com, hec...@clarkparsia.com
To complement my example. Still with the same dataset.


<http://iadb/skosTraning2>
  rdf:type owl:Ontology ;
  owl:imports <http://www.w3.org/2004/02/skos/core> ;
.
:DC
  rdf:type skos:Concept ;
  skos:prefLabel "DC"@en ;
.
:USA
  rdf:type skos:Concept ;
  :relatedCity :DC ;
  skos:prefLabel "USA"^^xsd:string ;

.
:relatedCity
  rdf:type owl:ObjectProperty ;
  rdfs:subPropertyOf skos:related ;
  skos:prefLabel "related city"^^xsd:string ;


When i run the following query:

select ?concept ?prop ?value 

where {
 ?concept a skos:Concept .
  

where {
 ?concept a skos:Concept .
  
   ?prop rdfs:subPropertyOf skos:semanticRelation .
  
   ?concept ?prop ?value
}


I get:



Which is the correct answer in this case, but then it is just strange what is happening with skos:related. It is like semanticRelation is there but not output ? Why ? It should not be there indeed !!!! What means those empty fields?


I have attached the file for your convenience. 


Finally if i do 

select ?concept ?prop ?value

where {

?concept a skos:Concept .
  
  ?prop a owl:ObjectProperty .
  
  ?concept ?prop ?value
}


I get 



Again i'm really puzzled. Either the SKOS Archetype is not fully ready or i am doing something extremely wrong that i can't see. SKOS is fundamental to my project. Hence either way, I need a way forward. Hence i hope you guys can guide me here

On Thursday, December 18, 2014 12:12:34 PM UTC-5, Héctor Pérez-Urbina wrote:
export (1).xml

Héctor Pérez-Urbina

unread,
Dec 22, 2014, 10:05:32 AM12/22/14
to Maatary Okouya, stardog
Dear Maatary,

I just wanted to let you know we have found and fixed a bug regarding reasoning with annotation properties. The fix will be included in the next release. We will continue analyzing the information you sent just in case there are other issues.
--
Best,
Héctor
Reply all
Reply to author
Forward
0 new messages