Re: Linking RDF

From: "Lee Jonas" <lee.jonas@cakehouse.co.uk>

> There are three problems this 'interpretation' of namespaces gives us:
open
> grammar, which is harder to validate simply (and nigh on impossible to do
> properly with DTDs); weird, unwieldy namespaces with different semantics
to
> other XML namespaces; resolution of RDF schemas clashes with resolution of
> XML schemas.

I think what we really need is that XML validation of OPEN  RDF stick to
syntactic issues, and leave the vocabulary issues totally alone.  There is a
well defined line between the syntax of RDF and it's vocabulary ... when you
get into the semantics that line disappears ... it's a slippery slope.   The
only validation of  OPEN RDF that we can practically do at parse time is to
check that elements nest correctly, and that pure syntactic terms like
rdf:description, rdf:about, rdf:resource (which don't end up in the triples)
are used properly.  Checking  vocabulary and semantics beyond that should be
left to whatever process is charged with calculating an appropriate
response.  Moreover, in any case,  semantic validity is better employed when
*writing* rather than *reading*.

> Discussions on these issues seem to have died down, yet the issues have
not
> been resolved and the new RDFCore working group are not even going to
> address them.  It seems that these aspects will not get reviewed for some
> time to come.  A pity.

A crime!

Seth

Received on Friday, 20 April 2001 12:45:05 UTC