Search Mailing List Archives


Limit search to: Subject & Body Subject Author
Sort by: Reverse Sort
Limit to: All This Week Last Week This Month Last Month
Select Date Range     through    

[p4-feedback] untyped rdf:Property vanishes

Timothy Redmond tredmond at stanford.edu
Wed Oct 14 13:49:14 PDT 2009


You are not going to be successful with the Protege 4 (or 4.1) owl
editor and this ontology style.  The problem is that Protege 4 uses the
Manchester OWL api which at a deep level is based on the OWL 2 language
defined by [1].  In that language there simply is no syntactic construct
that can be used to represent a property that is not known to be an
annotation, object or datatype property.

There is of course the alternative version of the OWL 2 language (OWL 2
Full) defined in [2].  This language can describe what you want.  To
properly support this language one would almost certainly need an
rdf-based tools.  Other approaches (e.g. Protege 3.4) quickly become
quite complex.  I think that I can say with some certainty that the
Manchester OWL api will never support OWL 2 Full (in those cases where
it is considerably different than OWL 2 DL).

We are thinking of making a Protege 4 rdf editor at some point and we
could make support of OWL 2 full constructs a requirement.  It looks
like Protege 3 will work with your ontologies and there are rdf-based
editors that you could try (e.g. topbraid).  Alternative you could
conceivably use some technique such as reification to make ontologies
that are compliant with [1].

Now for your specific questions.  The following snippet doesn't produce
anything.

> > <?xml version="1.0" encoding="UTF-8"?>
> > <rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" >
> >   <rdf:Description rdf:about="foo">
> >     <rdf:type rdf:resource="http://www.w3.org/1999/02/22-rdf-syntax-ns#Property"/>
> >   </rdf:Description>
> > </rdf:RDF>

I loaded this into the owl api and printed all the axioms and there was
nothing.  The owlapi could not find a triple that it could represent in
OWL 2 [1]. 


> rather than saying it is not OWL.

It would be nice to at least get an error message.  I will see if I can
add a tracker item  for this.

> > <?xml version="1.0" encoding="UTF-8"?>
> > <rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
> >  xmlns:rdfs="http://www.w3.org/2000/01/rdf-schema#">
> >   <rdf:Description rdf:about="foo">
> >     <rdf:type rdf:resource="http://www.w3.org/1999/02/22-rdf-syntax-ns#Property"/>
> >     <rdfs:isDefinedBy rdf:resource="bar"/>
> >   </rdf:Description>
> > </rdf:RDF>

Here the information is also insufficient to be sure how it should be
parsed into OWL 2.  There is one triple that is potentially meaningful
(the annotation of some resource).  The OWL API makes a guess and
assumes that this resource is an OWL Individual.  This is a guess and
there will be cases where  this does not meet the desires of the person
who wrote the ontology.

-Timothy


[1] http://www.w3.org/TR/2009/PR-owl2-syntax-20090922/
[2] http://www.w3.org/TR/2009/PR-owl2-rdf-based-semantics-20090922/


On Tue, 2009-10-13 at 20:49 -0400, Bob Morris wrote:
> Why does P4 decline to render the resource foo in the RDF below?
> 
> <?xml version="1.0" encoding="UTF-8"?>
> <rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" >
>   <rdf:Description rdf:about="foo">
>     <rdf:type rdf:resource="http://www.w3.org/1999/02/22-rdf-syntax-ns#Property"/>
>   </rdf:Description>
> </rdf:RDF>
> 
> rather than saying it is not OWL.
> 
> But
> 
> <?xml version="1.0" encoding="UTF-8"?>
> <rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
>  xmlns:rdfs="http://www.w3.org/2000/01/rdf-schema#">
>   <rdf:Description rdf:about="foo">
>     <rdf:type rdf:resource="http://www.w3.org/1999/02/22-rdf-syntax-ns#Property"/>
>     <rdfs:isDefinedBy rdf:resource="bar"/>
>   </rdf:Description>
> </rdf:RDF>
> 
> 
> results in the only(?) rendering possible (namely as an individual).
> 
> WonderWeb takes the same approach. Is there something in the OWL
> formal semantics that would acount for this?
> 
> This arises in http://rs.tdwg.org/dwc/rdf/dwcterms.rdf  where the
> modeler doesn't want mandatory datatyping, but wants to classify
> properties with classes that are domain specific, so introduces his
> own object property organizedInClass and adds assertions on the
> untyped properties, e.g.
>    scientificName organizedInClass Taxon
> That model uses rdfs:isDefinedBy everythwhere(?)
> 
> Thanks
> Bob Morris
> 
> 




More information about the p4-feedback mailing list