RE: Close to final draft of "classes as values" note

 
> OWL is built upon RDFS, so it is already in there. The issue is more 
> terminological difference I think: RDF people say 'vocabulary' and 
> OWL people say 'ontology'. Perhaps if we wrote 'RDF/OWL' more often, 
> the commonality might be made more widely appreciated?

Thanks Dan, and I'm indebted to an offlist discussion with my colleague Ian
Dickinson which has prompted the comment I'm about to make.  This does not
mean that Ian agrees with me and I hope he'll feel free to contribute his
views.

I suggest it is important to bear in mind the decentralised nature of the
web.  I suggest that a central goal of the semantic web is reuse of
published information.  Whilst I may publish data or an ontology with a
particular purpose in mind, and whilst I may know say, that  an Owl Full
reasoner will be used to achieve that purpose, I cannot know what reasoners
will suit other purposes for which this information may be reused.  That is
the nature of the web.

With that in mind, what advice would we give to Joesephine User, new to the
semantic web and ontologies, about how to represent information which might
naturally be represented using classes as values.  What should she do to
gain maximum reusability?

In such circumstances we might have hoped to appeal to the principal of
minimum requirements as promoting maximal opportunity for reuse.
Unfortunately however we have a double bottomed (with difficulty I refrain
from use of the vernacular) stack.  Is RDFS or Owl Lite the minimum
requirement? 

I am suggesting that we frame the purpose of the note on which Natasha has
done such excellent work in the context of the semantic web as a whole
rather than in how to solve some problem in OwlDL.  What advice do we give
her?  Stick to the common subset of RDFS and OwlLite?

Brian

Received on Monday, 17 May 2004 06:52:35 UTC