W3C home > Mailing lists > Public > public-esw-thes@w3.org > May 2009

Re: FAQ:

From: Alan Ruttenberg <alanruttenberg@gmail.com>
Date: Fri, 22 May 2009 03:09:52 -0400
Message-ID: <29af5e2d0905220009i1b3c4d92ubdf5b5a4903fb2fc@mail.gmail.com>
To: "John F. Madden" <john.madden@me.com>
Cc: Kevin Doyle <kdoyle@teranode.com>, public-esw-thes@w3.org
Please consider coordinating with the OWL WG if you plan to pose and answer
this question in a FAQ. I don't think the below answer is particularly
accurate or helpful. For one thing, RDF and OWL have model theoretic
semantics, but this does imply the semantics are extensional. For example,
we can have have c1 equivalentClass c2 but c1 differentIndividual c2. I have
further criticisms, but would rather discuss them in the context of putting
together an answer that does justice to the two efforts and their aims.

Also, there are some actual differences in the semantics that should be
mentioned, as SKOS has some semantic conditions that can not be expressed in
OWL or RDF, such as S14: A resource has no more than one value of
skos:prefLabel per language tag, which can't be expressed in OWL 1 or OWL 2,
 or S12: The rdfs:range of each of skos:prefLabel, skos:altLabel and
skos:hiddenLabel is the class of RDF plain literals, which can't be
expressed in OWL 1, but probably will be able to be expressed in OWL 2.

On Wed, May 20, 2009 at 4:47 PM, John F. Madden <john.madden@me.com> wrote:
> Two possible reasons, by my lights:
> (1) If the goal is mapping legacy terminologies into an RDF dialect:
> RDF/OWL has a formal, model theoretic, extensional semantics. Most legacy
> terminologies have an informal semantics that is entirely intensional.
> Putting them directly into RDF/OWL imputes to them an ability to support a
> kind of inferencing that they were never intended to support. The result
> liable to be unexpected or undesirable inferences; or in other cases,
> failure to entail inferences that one would imagine to be natural
> consequences.
> SKOS, on the other hand, doesn't treat the legacy terms as classes with
> extensions (as RDF/OWL does). Yet SKOS still gives one a vocabulary to
> about the most important intensional meanings that such terminologies tend
> to use (notions like narrower, broader, related, etc.).
> (All this is much better said than I ever could in the SKOS Primer.)
> (2) If the goal is to model a set of terms de novo:
> For the same reasons as stated in (1), SKOS provides a way to depict
> informal, notional relations among ideas without having to buy into the
> rigorous RDF/OWL semantic model, which may be too constraining for certain
> kinds of modeling (for example, the more freewheeling kinds exemplified by
> "concept-mapping", "mind-mapping"; or the type of intuitive models that
> tends to get from domain experts).
> John
> On May 20, 2009, at 3:57 PM, Kevin Doyle wrote:
> Hi,
> I have a question I would like to put on the SKOS FAQ, because I don't
> the answer.  Also, this is the first place that I looked for the answer.
>  Why SKOS and not OWL?  Or maybe to put the question another way, what are
> the advantages of using SKOS over OWL?
> Kevin S. Doyle
> Client Solution Manager
> Teranode Corp.
> www.teranode.com
> Tel: +1-617-710-5155
Received on Friday, 22 May 2009 07:11:01 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 2 March 2016 13:32:11 UTC