W3C home > Mailing lists > Public > public-swd-wg@w3.org > December 2007

Re: TR : [ISSUE-36 related] Linking a ConceptScheme and a Concept subclass using OWL

From: Antoine Isaac <aisaac@few.vu.nl>
Date: Wed, 26 Dec 2007 18:48:17 +0100
Message-ID: <477293E1.5060104@few.vu.nl>
To: SWD WG <public-swd-wg@w3.org>, SKOS <public-esw-thes@w3.org>, Bernard Vatant <bernard.vatant@mondeca.com>

Hi Bernard,

It might be useless to answer so late, but I really had no time to write 
a line on this when I received the mail :-(

> De: public-esw-thes-request@w3.org de la part de Bernard Vatant
> Date: ven. 30/11/2007 14:26
> : SKOS
> Cc: Thomas Francart; Benoit Carcenac
> Objet : [ISSUE-36 related] Linking a ConceptScheme and a Concept 
> subclass  using OWL
> Hello all
> We're having again at home (Mondeca) this permathread about Concept
> Scheme vs subclass of Concept.
> We had this discussion six months ago and the final (?) conclusion at
> the time was by Antoine
> http://lists.w3.org/Archives/Public/public-esw-thes/2007Jun/0046.html
> To sum up Antoine's point, Concept Schemes and subclasses of Concept
> have in general different purposes, and different semantics. Point well
> made, and I agree with the conclusions.
> Nevertheless, in quite a number of cases, we need to assert the
> equivalence between attachment to a Concept Scheme and belonging to a
> Concept subclass, if only when we want to specify that some property
> range has values from a given Concept Scheme. For example in a Tourism
> ontology you have a Class "Accomodation", with a property
> "environmentalDescription" for which you want to use values from GEMET,
> or maybe even a specific "Theme" or "Group" from GEMET, e.g.,
> "ENVIRONMENT". ... assuming a GEMET Group or Theme is a ConceptScheme,
> (which seems to be declared nowhere).
> gemet:group10111  a  skos:ConceptScheme
> ex:environmentDescription  rdfs:range   ex:_b
> How do you express that instances of ex:_b  are concepts having
> skos:inScheme value gemet:group10111 ?
> You can't do that in RDFS AFAIK, you can do it in OWL.
> ex:_b      a   owl:Restriction
> ex:_b     owl:onProperty      skos:inScheme
> ex:_b     owl:allValuesFrom    gemet:group10111
> This way, using SKOS in a OWL environment, one can express the
> equivalence between a Concept Scheme an a Concept subclass.

Though this use of subclass-ing skos:Concept is indeed not the first one 
I would think of, there is nothing I think which should prohibit it, if 
you really need it.
Additionally, the solution you propose seems quite sound...

> A side order of such declarations is the possibility to indirectly
> infer, and use, a hierarchy of Concept Schemes from the corresponding
> classes hierarchy, since there is no way in SKOS itself to assert a
> hierarchical relationship (partOf, whatever) between Concept Schemes (or
> is there?)

True, there is nothing there. We just have the possibility to "include" 
concept schemes at the level of individual concepts, by multiple use of 
skos:inScheme on a same Concept.
Actually the only way for the moment to perform "general concept scheme 
inclusion" would be by using your pattern :-|

I guess there are two simple arguments explaining why there will not be 
much more than that in the foreseeable future:
- if we design such a thing at the vocabulary level it will raise a lot 
of other problems, like concept scheme version management, which I think 
we don't have time to handle for now.
- (quite surprisingly, to me) this requirement has not appeared clearly 
from the use cases.
I'm not saying that the problem will never occur and should not be dealt 
with, just trying to find some justification for not putting right now 
our (quite limited) workforce on it.

> BTW Is it in the scope of the recommandation, or some informative annex,
> to show such examples of SKOS integration with OWL?.

Both Reference and Primer should have something. Maybe not in a complete 
way in the first versions that will be anounced around, but in later 
ones, yes!

Received on Wednesday, 26 December 2007 17:48:32 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:31:47 UTC