Re: Issue: versionIRI for SKOS-DL ontology? - PROPOSED RESOLUTION

PROPOSED RESOLUTION

That http://www.w3.org/2009/08/skos-reference/skos-owl1-dl.rdf
be edited to add the triple:

    <http://www.w3.org/2004/02/skos/core>
        <http://www.w3.org/2002/07/owl#versionIRI>
            <http://www.w3.org/TR/skos-reference/skos-owl1-dl.rdf> .

NOTE

-- Section C.3 of the SKOS Reference [1] says that the 
   informative (non-normative) OWL 1 DL Sub-set should be
   cited with [2], which redirects to [3].

-- If we can agree on a resolution, a versionIRI triple
   could be added, with the decision recorded at [4].

[1] http://www.w3.org/TR/skos-reference/#namespace-documents
[2] http://www.w3.org/TR/skos-reference/skos-owl1-dl.rdf
[3] http://www.w3.org/2009/08/skos-reference/skos-owl1-dl.rdf
[4] http://www.w3.org/2006/07/SWD/SKOS/reference/20090811-errata

Tom



On Tue, Nov 23, 2010 at 01:05:10PM -0500, Thomas Baker wrote:
> The following issue regarding the formal representation of
> the "DL prune" of SKOS has been raised by Frederick Giasson
> <fred@fgiasson.com>:
> 
>     The main problem is that the skos-core-dl version is using the 
>     same ontology IRI as skos-code. Theoretically, there is no issue 
>     in itself, but practically, in a OWL 2 environment, there are a 
>     few issues related to this situation. As noted here [1]:
> 
>     "If an ontology has an ontology IRI but no version IRI, then a 
>     different ontology with the same ontology IRI but no version IRI 
>     /SHOULD NOT/ exist."
> 
>     Obviously, the current skos-dl ontology is breaking this rule. 
>     While this new ontology versioning mechanism is part of OWL 2 and 
>     was introduced after the last SKOS recommendation, I think that 
>     this rule should be applied to OWL 1.1 ontologies as well, 
>     without being explicit in the specification.
> 
>     The problem is that breaking this rule does affect some current 
>     implementations of OWL 2 in different libraries/software. One 
>     good example of this is with the OWLAPI and Protege 4.1 (at 
>     least) as explained here [2] and check Timothy's answer [3]. I 
>     don't know what the exact problem is, but I guess that it is due 
>     to the fact that the OWL API does have a few built-in 
>     understandings of the skos-core ontology IRI which clash with the 
>     imported skos-dl version of the ontology. Since the ontology IRI 
>     is the same, then it uses the skos-core ontology instead of the 
>     skos-dl. I can also imagine that other frameworks/systems could 
>     behave unpredictibly because of this broken assumption.
> 
>     So, I don't know what can be done, if anything. But certainly 
>     that changing the ontology IRI for the skos-dl ontology could 
>     help a lot. I am not sure you would be willing to add any OWL 2 
>     constructs into this version of the ontology, but I would suggest 
>     you consider using the owl:versionIRI predicate to specify the 
>     specific version of that ontology.
> 
>     May I suggest something like:
> 
>           owl:versionIRI<http://www.w3.org/2004/02/skos/core/dl#>.
> 
>     [1] http://www.w3.org/TR/2009/REC-owl2-syntax-20091027/#Ontology_IRI_and_Version_IRI
>     [2] http://protege-ontology-editor-knowledge-acquisition-system.136.n4.nabble.com/SKOS-labels-annotation-vs-datatype-properties-tt1840259.html#a1840259
>     [3] http://protege-ontology-editor-knowledge-acquisition-system.136.n4.nabble.com/SKOS-labels-annotation-vs-datatype-properties-tt1840259.html#a1840495
> 
> -- 
> Tom Baker <tbaker@tbaker.de>

-- 
Tom Baker <tbaker@tbaker.de>

Received on Tuesday, 23 November 2010 18:17:17 UTC