W3C home > Mailing lists > Public > public-owl-wg@w3.org > February 2008

Re: A thought on fragments and rec-track

From: Carsten Lutz <clu@tcs.inf.tu-dresden.de>
Date: Fri, 15 Feb 2008 09:04:27 +0100 (CET)
To: Vojtech Svatek <Svatek@vse.cz>
Cc: public-owl-wg@w3.org
Message-id: <Pine.LNX.4.64.0802150849240.8936@frege.inf.tu-dresden.de>


On Fri, 15 Feb 2008, Vojtech Svatek wrote:
> Correct me if I am wrong, but my intuitive understanding of the main
> aspects of a web ontology language has always been:
> 1) MODELLING: the language serves to express real ontologies with some
> content and structure
> 2) WEB ENGINEERING: the language has to be shared over the web environment
> 3) LOGICS: the language should allow for formal inferencing.
> 99% of the disucssions I so far saw on the mailing list, apart from WG
> management issues, dealt with aspects 2), 3), and their possible clashes.
> I am actually pretty sure that even many people who primarily act on behalf
> 2) and 3) have a solid ontology engineering background and experience from
> large applicative projects. Is it intentional that the aspect 1) is not
> openly discussed?

You are right that Point 1) is very important, but does not make up a
large part of the discussion so far. So if you or anybody else has
relevant input, let the WG know.

On the other hand, to be fair I should say the following. It is my
understanding that most people have 1) in the back of their heads
while discussing 2) and 3), but maybe fail to make it sufficiently
explicit. And it should also be said that a lot of input from 1)
went into the original proposal of OWL 1.1 that the WG started off

What I also agree with is that, in the current discussion, there is
too much focus on web and databases. These are doubtlessly important
things, but OWL does not only contain a W, it also contains an O (even
in front of the W though the spelling out of "OWL" suggests the
opposite :). In fact, most of the *existing* (as opposed to envisioned) 
success of OWL is due to its use as an ontology language. Disclaimer:
I do not aim at undermining the proposed peaceful co-existence between 
the RDF and DL worlds with this comment, only to emphasize to "co".

Finally, let me take this opportunity to say that your Point 1) is
one of the main reasons why I advocate such much EL++ as a fragment
of OWL *DL*: there is quite a number of very serious and commercial
ontologies which use this fragment. To name only a few: SNOMED, the NCI
(national cancer institute's) thesaurus, and the Gene ontology. Other
large ontologies such as Galen consist of EL++ statements to more than
90%. In fact, I do not know any large-scale ontology that does not 
satisfy these conditions, but I am happy to stand corrected. This can 
hardly be said of any other fragment of OWL DL, and I note again that 
I speak about what already *exists*. On top of this, EL++ also pushes 
the limits of what *could be* because we did a very careful job on 
pushing the expressivity border as far as possible in the direction of
practically useful expressivity while retaining polytime inferencing.
This was done in close collaboration with ontology designers such as
Kent Spackman, the chief ontology designer of SNOMED, and we are 
getting very positive feedback on our work.


*      Carsten Lutz, Institut f"ur Theoretische Informatik, TU Dresden       *
*     Office phone:++49 351 46339171   mailto:lutz@tcs.inf.tu-dresden.de     *
Received on Friday, 15 February 2008 08:04:47 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:42:02 UTC