Re: Analyzing OWL problems in some projects

On Mon, 2003-10-20 at 16:31, Jonathan Borden wrote:
> Jim Hendler wrote:
> > ...We have found a number of problems of various kinds, but one that 
> > seems to come up a lot is people using owl:subclassOf, owl:range, and 
> > owl:domain -- I'm not sure if we have a test that actually exposes 
> > these -- might be worth creating some to make it more obvious 
> These sorts of typos are best caught by syntactic validators i.e. XML 
> schemas --- the problem is that there isn't one for OWL. Would it be 
> helpful to extend the RELAXNG schema for RDF/XML to prevent "extending 
> the OWL namespace"?

I think so, yes.

I just did some OWL editing with nxml-mode*
(which comes with a relax-NG schema for RDF) and it ROCKS.
It finds "striping" errors such as missing rdf:subClassOf
between a class element and a restriction element.

OWL/RDF/XML syntax is pretty tedious and nxml-mode
is a *huge* help. The sort of thing you're talking
about seems like it will help even more.

* xmlhack: James Clark unveils a new XML mode for GNU Emacs

>  or else are we happy to prevent a small subset of 
> potentially infinite class of errors of this type. e.g. owl:rng 
> owl:range owl:renge ...
> It is relatively easy specify *all* the illegal names at the XML 
> validation level by the pattern:
> owl:* - (owl:Class|owl:Thing|owl:Restriction ...)
> Jonathan
Dan Connolly, W3C

Received on Monday, 20 October 2003 17:17:48 UTC