Re: Cheers and Concerns

Hello Autumn,

Thank you for your feedback. Please see my responses to your points
inline below...

Dan Connolly wrote:
> 
> [comments on the WebOnt specs should be
> directed to public-webont-comments@w3.org;
> I'm forwarding your message there.
> www-webont-wg is for WG members;
> if you're interested in joining the WG,
> let us know.]
> 
> On Wed, 2002-04-17 at 23:34, Autumn Cuellar wrote:
> > Hello,
> >
> > I was very happy to learn of your group's massive undertaking, as I've
> > tried to use RDF Schema and DAML+OIL in defining an XML ontology for an
> > XML-based biological modelling language we're developing, and both have
> > fallen short in areas.  So you know where I'm coming from: we have
> > already done the work (released a specification) for a language that
> > describes cellular models.  We're interested in using ontologies to not
> > only describe the language that we've created, but to continue to add to
> > that language.  For instance, we have the ability to describe reactions,
> > now we want to add the ability to import a commonly used reaction.
> > First we have to be able to define what those commonly used reactions
> > are.  Anyways, enough about my work and on to yours....
> >
> > I wanted to bring up a couple of problems I've had in the past, and
> > maybe others have had these problems and they might be worth discussing:
> >
> > 1)  I've found the common way of referencing attributes and elements in
> > a namespace inconvenient (url#element_name, or
> > http://www.example.org/attribute_name).  Perhaps rather short-sightedly,
> > we've given elements and attributes the same name, so you can see the
> > problem that arises when you're trying to describe two different
> > concepts (an element with the name of "units" and an attribute with the
> > name of "units") with an rdf:about="http://www.example.org#units".
> > According to the XML namespace spec
> > (http://www.w3.org/TR/1999/REC-xml-names-19990114/), these two "units"
> > could be given their expanded names which would establish that they are
> > not the same, but how do you reference the expanded names in an
> > rdf:about attribute?  This is possibly an issue that I should raise
> > elsewhere (suggestions on where?) and is probably out of the scope of
> > your intentions, but it is an issue that could affect any ontologies
> > built with your language.

This issue really is out of the scope of WebOnt. In RDF, which WebOnt
builds on, there is no distinction between attributes and elements,
there are only resources. You may want to address this particular
concern to the XML Schema working group.

> > 2)  Scientific groups using XML, of which I am sure there are more than
> > a few, might find it helpful to be able to define things in an ontology
> > mathematically, for instance, one thing is a ratio of two other things.
> > Have you considered allowing the use of MathML in an ontology language?

Because of the nature of ontology languages simple merging with other
XML DTDs/Schemas is not as easy as it might sound. For something like
this, it would be more appropriate to create a Math ontology in the
WebOnt language. You could then build software that translate between
MathML and this WebOnt Math ontology.

> > 3)  I'm supportive of your use of DAML+OIL, but the one complaint I have
> > about DAML+OIL is their method of using user-defined datatypes
> > (referencing an XML Schema file).  They suggest referencing a class that
> > hasn't been specifically defined as such.  Additionally, I would prefer
> > to define the datatypes in the same file.

WebOnt and RDF assume that the world consists of classes, properties,
and resources. It is possible to view a datatype as simply a set of
values, and from this viewpoint a datatype is a class (although it may
have additional information, such as ordering, equality, etc.). As for
defining datatypes in the same file, that would put extra burden on
WebOnt parsers. They would have to be XML Schema parsers as well. By
keeping them in separate files, the parsing of datatypes can be handed
off to a separate program and common datatypes can be handled with
special purpose code.

> > 4)  Will import capabilities (of other ontologies) be dealt with at some
> > point?  Can't imagine you'd create a web ontology without import
> > capabilities, but I want to make sure it's somewhere on your list.

Yes imports is a key feature. DAML+OIL had an imports property and the
WebOnt requirements document has "Explicit ontology extension" listed as
a requirement (this is basically imports under another name).

> > Cheers for your efforts so far.  I like what I'm seeing and will coninue
> > to watch your progress closely.
> >
> > Best wishes,
> > Autumn
> >
> > --
> > Autumn A. Cuellar
> > Bioengineering Institute
> > The University of Auckland
> > New Zealand
> > www.cellml.org
> --
> Dan Connolly, W3C http://www.w3.org/People/Connolly/

Received on Thursday, 9 May 2002 11:54:48 UTC