Re: [UNITS, OEP] FAQ : Constraints on data values range

Jeremy, Guus, Bernard, All

I'd like to come in just to say that in a recent set of travels and discussions with potential users of OWL, the inability to express numeric constraints involving inequalities for things such as "Big Wheel" etc. is a big problem.  Virtual all my biomedical examples require normal values, value ranges, etc. which need this kind of expressiveness .

To be blunt, there is no way that many potential users in these fields will adopt OWL if it can't manage these 'simple' constraints.  As a temporising measure, we may have to build some hybrid tools, but this is not really satisfactory.

Jeff Pan and co are indeed working on formal solutions.  I think the role of the best practices group should be to be sure that they have a full set of requirements from the various applications against which to test those solutions.  I've been having a considerable discussion with them which I think has finally got past my misconceptions.

If we can gather together a good set of use cases and examples into a summary paper, I think they would find that valuable - as I hope would the group.

This is a place where the existing standard is inadequate to express what users need.  I hope we can help get to the point to move it forward with a clear set of user requirements - plus workarounds and help to users not to 'bang their heads against the wall'.  Sometimes knowing that something can't be done is almost as useful as knowing how to do it.  Most users I have encountered have assumed it was their misunderstanding rather than an intrinsic limitation in the language.

Regards

Alan




Jeremy Carroll wrote:

> > formal model. I recommend a paper "Web Ontology Reasoning with
> > Datatype Groups"
> > form J. Pan and I. Horrocks.
> > I think this is a problem that has to be explain in a FAQ.
> >
>
> I think that we currently have, in OWL, particularly OWL DL, and
> unnecessarily restrictive langauge for this problem. However, I am not clear
> that enough industrial class apps are up with what datatype support there is
> already in OWL, so I am in no hurry to move the goal posts towards
> expressivity.
>
> Some of Pan and Horrocks' work goes well beyond OWL expressivity and looks
> really useful for the units stuff amongst others.
>
> I remain doubtful as to the scientific validity of the current exclusion of
> inverse functional datatype properties (IFDP) in OWL DL, but didn't make a
> song and dance about it at the appropriate time, and won't now either. As
> far as I remember it is a complexity argument rather than a decidability
> argument, and it seems to me that OWL DL + IFDP is of the same complexity as
> OWL DL (maybe I have misunderstood; I've read the papers but couldn't
> reconstruct them).
>
> Jeremy

--
Alan L Rector
Professor of Medical Informatics
Department of Computer Science
University of Manchester
Manchester M13 9PL, UK
TEL: +44-161-275-6188/6149/7183
FAX: +44-161-275-6236/6204
Room: 2.88a, Kilburn Building
email: rector@cs.man.ac.uk
web: www.cs.man.ac.uk/mig
        www.opengalen.org
        www.clinical-escience.org

Received on Friday, 30 April 2004 17:39:18 UTC