Re: OWL 2?

Hi Paul,

With regard to the use of OWL 2 to express SKOS, I'm not sure if you are
aware of work done in Greece. The SWIGroup [1] have been investigating
whether OWL 2 could be a solution to express SKOS semantics rule. @
TenForce, we have been using SKOS extensively to answer the needs of our
costumers. One of the solution we have used to enforce the SKOS integrity
was to mix constraints expressed in OWL with some expressed based on SPARQL
queries and the Schemarama 2 Framework (e.g. [2]). This approach is very
useful as we can create rules specific to clients and their applications
relatively easily. In your blog [3], I see that you use a similar approach
based on the SPIN standard [4].

With regard to OWL 2 tools, I have struggled to find a tool that was good
enough to support the development of OWL 2 ontologies. Although TopBraid
Composer [5] supports OWL2, it only support some features of it. In my
opinion, this tool is more suitable to deal with RDF rather than OWL as it
RDF/XML output suggests. Protege 4 [6] has been suggested as a tool to
develop OWL 2 ontologies. However, the work on the OWL API [7] (especially
migration from OWL API 2 to OWL API 3) means that the tool is far from being
stable. It would also be nice if some work was done on the user interface,
but I don't think this is a major focus for the developer @ this time.

In my opinion, the core advantage of OWL 2 is the introduction of new
profiles (e.g. OWL 2 EL, OWL 2 RL and OWL 2 QL) [8]. These profiles allow
the an ontology engineer to use trimmed down version of OWL by trading some
expressive power for the efficiency of reasoning. For example, the EL
description logic has been used in the medical domain and provides efficient
reasoning on large ontologies (see CEL [9] and TrOWL [10] for EL+
reasoners). But its core disadvantage is the lack of proper tools (as also
pointed out by Paul).

Kind regards,

Quentin

[1] http://swig.hpclab.ceid.upatras.gr/SKOS/Skos2Owl2
[2] http://www.w3.org/2004/02/skos/schemarama/test/B1
[3] http://goo.gl/iVHX
[4] http://spinrdf.org/
[5] http://www.topquadrant.com/products/TB_Composer.html
[6] http://www.co-ode.org/downloads/protege-x/
[7] http://owlapi.sourceforge.net/
[8] http://www.w3.org/TR/owl2-profiles/
[9] http://code.google.com/p/cel/
[10] http://trowl.eu/

On 18 September 2010 09:46, Paul Hermans <paul@proxml.be> wrote:

> I want to share my limited experiences.
> I was very eager to learn and use OWL2.
>
> My use case was the SKOS vocabulary.
> SKOS has been formalized in OWL1-Full and OWL1-DL,
> but since not all of the SKOS rules and constraints could be expressed in
> OWL1
> I went to OWL2 for trying to formalize those additional semantics.
>
> To my big surprise I discovered that OWL2 was of no use in doing this.
> This is documented at http://bit.ly/aYodzu.
>
> Frustrated, but no big deal, OWL2 must have been developed with other use
> cases in mind.
>
> But my frustration became even bigger when moving to tools using the
> OWL(2)-API.
> Although one claims that SKOS(1) is built into the OWL-API I do not succeed
> in getting the SKOS-XL vocabulary to work into these tools.
> This is documented at: http://bit.ly/aO21h3
> I tried to figure out what is going on, but no one (except Timothy Redmond)
> seems to care.
>
> My conclusion:
> I went to the store; they didn't deliver the goods and I did not encounter
> any customer oriented attitude.
>
>
>
>
> Paul
>
>
>
>
>
>
>
>
>
>
>
>
> On 18 Sep 2010, at 04:02, Sandro Hawke wrote:
>
> > On Sat, 2010-09-18 at 02:15 +0100, Nathan wrote:
> >> Hi All,
> >>
> >> Apologies for this mail as it may not have a direct answer, but unsure
> >> where else to ask. I often hear a general sentiment of negativity
> >> towards OWL 2, implying that there's something wrong with it, that it's
> >> not practical or not for the masses only for the academics.
> >>
> >> Simply, I don't understand or follow this thought and am probably
> >> missing something obvious, can any one enlighten me?
> >>
> >> Hopefully this isn't a sore subject or something that shouldn't be
> >> spoken about, and obviously no offence to anybody - just if somebody
> >> could fill me in on the background behind the sentiment and why this
> >> conclusion has been (I think?) reached, could they let me know - on or
> >> off list.
> >>
> >> Best,
> >>
> >> Nathan
> >>
> >> ps: this really isn't intended to spark a bit debate or war - even that
> >> sentence may show just how little I understand any background on this
> topic.
> >
> > I'm too biased to have any useful opinion here (I was a W3C staff
> > contact for the OWL Working Group), but I'm curious whether the
> > sentiments you've heard have been against OWL in general, or against OWL
> > 2 and in favor of OWL 1?  I've heard a little bit of grumbling from OWL
> > 1 folks who didn't see the need for the OWL 2 features, but (in the
> > feedback I saw) they were dramatically outweighed by the folks who
> > wanted the new features of OWL 2.
> >
> > In terms of who supports OWL 2, I can refer you to the official
> > implementation report, as of a year ago:
> > http://www.w3.org/2007/OWL/wiki/Implementations
> >
> > If you follow the links in the first column, you'll more about who is
> > behind the implementation.   This doesn't really answer your question,
> > though, since even if these were all academic (they are not), it
> > wouldn't speak to the user base.
> >
> >   -- Sandro
> >
> >
>
>
> Kind Regards,
> Paul Hermans
>
> -------------------------
> ProXML bvba
> XML and OWL/RDF services
> (w) www.proxml.be
> (b) experiences and opinions
> (e) paul@proxml.be
> (t)  +32 15 23 00 76
> (m) +32 473 66 03 20
>
> Narcisweg 17
> 3140 Keerbergen
> Belgium
>
>
>
>
>
>
>
>
>