W3C home > Mailing lists > Public > public-owl-wg@w3.org > November 2007

Re: ISSUE-78 (OWL LitEL++): OWL-Lite as EL++

From: Bernardo Cuenca Grau <bcg@cs.man.ac.uk>
Date: Wed, 28 Nov 2007 19:48:50 +0000
Message-ID: <474DC622.1030507@cs.man.ac.uk>
To: OWL Working Group WG <public-owl-wg@w3.org>

When writing the Tractable Fragments document, the intention was to 
offer to the users and tool builders a set of sensible fragments of OWL 
1.1 that they could use and report about the advantages of each of them. 
In principle, I tried to give each of them equal status. Now, we are 
facing an standardization process and I see the appeal in choosing one 
particular fragment and give it a name such as ``OWL Light''. I would be 
open to this, but this raises various issues for discussion that may be 

- If we are to choose one fragment, which one.Carsten advocates for EL++ 
and gives good reasons for it. Other people may advocate for other 
fragments. I think the users and tool builders will have something to 
say here.
- If different people are advocating for different fragments to be the 
``chose one'', there is little hope for a ``compromise'', meaning that 
there is little hope for designing a new fragment that includes features 
from the different candidate fragments and still enjoys nice 
computational properties.
- If we are to choose one fragment, what would happen with the other ones?
       - Removed from the document, or
       - Included in an informative note, but not in the normative document?
    - Other options?

In summary, this is something we should all discuss and reach consensus on.


OWL Working Group Issue Tracker wrote:
> ISSUE-78 (OWL LitEL++): OWL-Lite as EL++
> http://www.w3.org/2007/OWL/tracker/issues/
> Raised by: Bijan Parsia
> On product: 
> (On behalf of Carsten Lutz.)
> In my opinion, the fragment EL++ sticks out from the other ones for
> an important reason: it is (truely tractable and) used for a lot of
> practically very relevant ontologies. Examples:
> - SNOMED, a commercial medical ontology underlying the standardization
>   of medical terminology in the health systems of US, UK, etc.
> - NCI, the national cancer institutes medical ontology
> - Gene Ontology
> In OWL 1.0, there was an "OWL Light" version of OWL. The idea was to
> provide a lightweight version of OWL for which reasoning is simpler,
> but then it was ill-designed (reasoning actually wasn't simpler), and
> deprecated in OWL 1.1.
> I would like to advocate having a new OWL Light, which is EL++. My
> main reason for proposing this is that having an official name with
> "OWL" in it is likely to increase the visibility of this fragment *a
> lot*. This is good for two reasons:
> 1. We open up the OWL world for ontology developers that want to work
>    with a tractable languages. There are quite some developers who
>    insist on tractability (to name only one example, the SNOMED people).
> 2. Just by choosing the proper name, we can make stronger claims about
>    the relationship between OWL and a number of important ontologies.
>    For example, when we choose EL++ as OWL Light, we can then claim
>    that SNOMED is written in OWL Light, which may again draw attention
>    to OWL Light and OWL in general.
> In summary, I feel that we have a real chance here to truely extend
> the scope and visibility of OWL, simply by choosing a name. Note that
> in contrast "being one of the 27 tractable fragments of OWL" sounds
> much less convincing. Or in yet other words: the current "tractable
> fragments" document does not standardize anything, it rather has an
> informative character. To standardize something, you cannot list all
> options, but you have to make a *decision*. This is what I advocate.
> Note that I do not insist on the name "OWL Light". Other options
> such as "OWL Poly" may be fine as well, but it should have OWL in
> it, and there shouldn't be 5 other fragments that also have OWL
> names (for otherwise the effect described above vanishes again).
> Disclaimer: With EL++, I am advocating my own work here. I believe
> that my arguments are objective, but still you should know this.
Received on Wednesday, 28 November 2007 19:53:46 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:42:00 UTC