W3C home > Mailing lists > Public > public-esw-thes@w3.org > April 2012

Re: A question about SKOS condition S37

From: Antoine Isaac <aisaac@few.vu.nl>
Date: Wed, 11 Apr 2012 17:42:04 +0200
Message-ID: <4F85A64C.7090001@few.vu.nl>
To: <public-esw-thes@w3.org>
Dear Sean,

This is indeed the appropriate forum,

skos:Collection is indeed a quite difficult beast. It was created to capture specific cases, like thesaurus arrays (more on that in the SKOS Primer and Definition documents). But the name was pretty generic, and the cases, sometimes fuzzy.

I'm not 100% sure, but the decision to make S37 was to avoid too much ambiguity. It is not possible to have a collection being a concept (or a concept scheme) at a same time. This has some consequences, which one may complain about sometimes--especially, collections cannot be the object of semantic relations like skos:broader.
But it has some positive aspect, such as the one of people not being tempted to use collections for cases they were not originally intended to.

As an example, your case with Modern Art and Art would clearly be handled by representing these two resources as skos:Concepts and asserting a skos:broader between them. True, this skos:broader link may have a conceptual containment flavor in your eyes, but it's really different from what was intended when Collections and collection membership links were designed.

So as you guessed it is in particular impossible to have an OWL class defined as a subclass of both
skos:Concept and skos:Collection.

And for the record limiting the DL expressivity of SKOS datasets was never a motivation for S37, as far as I can tell!

I hope this helps!

Kind regards,

Antoine


> I have some questions considering SKOS condition S37:
>
>     S37	skos:Collection is disjoint with each of skos:Concept and
> skos:ConceptScheme. [1]
>
> I don't wish to criticize the rationale that, as I understand, must be
> represented in that condition. I presume that it may have something to do
> with limiting the DL expressivity of SKOS-based ontologies, for purpose of
> efficiency in inference systems - is that even close, though?
>
> I'm not certain if that's exactly it. If it's even "close enough", I doubt
> that that could be all there is to it. I understand that there must be a
> rationale to it, though it was not denoted directly in [1].
>
> My question arises as a matter of a concern that I would like to present for
> review: (1) That there may be a class of Concept defined that would *not* be
> naturally disjoint with a class of Collection. Secondly, on that basis: (2)
> Furthermore, that a Concept Scheme could be viewed as a collection of
> concepts. (Is my rationale too naive, in that?)
>
> I would like to illustrate point #1 with a matter of classification in the
> arts - namely, that the concept of Modern Art may be defined as being
> semantically contained within the concept of Art (and that the concepts,
> Surrealism, Cubism, Dadaism, Futurism, and so on, those may be defined as
> contained within the concept of Modern Art, in turn) .
>
> I consider that the concept of Modern Art does not exist independent to the
> concept of Art, and that it is therefore appropriate to say that the concept
> of Modern Art is contained in the concept of Art. I guess that it could be
> said to represent a mode of container membership.
>
> I would like to be able to express that sense of "conceptual container" and
> semantic entailment, in an ontology. I suppose I'm simply not certain of
> whether I could represent it in SKOS, however, given that condition S37.
>
> In addition to that I would be intrigued to understand the principled basis
> of condition S37, I would like to ask a question to the details of the
> matter. Given an OWL class that would be defined as a subclass of both
> skos:Concept and skos:Container, would that class be invalid onto SKOS?
>
> (As far as my concern #2 as noted, I think it's essentially covered within
> #1 there.)
>
> This mailing list was denoted as it being available for discussion of SKOS
> design and development[2].  I presume that this may be the most appropriate
> forum for these questions, inasmuch. Thank you. Cheers.
>
> -- Sean Champ
>
>
> [1] http://www.w3.org/TR/2009/REC-skos-reference-20090818/#L3424
> [2] http://www.w3.org/2004/02/skos/mail
Received on Wednesday, 11 April 2012 18:29:08 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 11 April 2012 18:29:09 GMT