RDF, URIs, XML

Going back to both Tim BL's book (_Weaving the Web_) and various documents
at the W3C regarding the 'Semantic Web', I keep encountering RDF - almost
to the exclusion of XML.  Adding to that the context of this heated
discussion, especially the recent discussions about URI usage in XML,
raises some fairly significant concerns.

I'm worried that this debate - both the relative URIs for namespace IDs and
the namespaces as references to other resources - isn't really about what's
best for XML or most acceptable for maintaining backward compatibility with
a variety of interpretations.

Rather, the debate (or its resolution at any rate) hinges on what's best
for the "Semantic Web" - and reading the various documents publicly
available, it becomes quite clear that the Semantic Web is about RDF - with
XML as a syntactical afterthought.

While a lot of participants here are most interested in preserving XML
documents, interoperability among XML implementations, or the future
extensibility of XML, a key group seems to be focused on making XML safe
for RDF.

Namespaces pointing to schemas and the various other uses of URIs described
in Tim BL's clearly heartfelt analysis of URIs are techniques that are
built into RDF, or at least seen as critical to the Semantic Web that seems
planned to be built on RDF.  

On the other hand, they don't feel like 'best practices' for XML.  They're
all possible within an XML framework, and RDF is welcome to layer them _on
top of_ that framework, but they aren't necessarily part of XML, good for
XML, or in crying demand by a large portion of the XML community, so far as
I can tell.

Let RDF handle relative URIs however it sees best for the Semantic Web, but
I'm not sure that those of us working only with XML and namespaces toward
other pursuits need to have the requirements for the Semantic Web forced
into our daily work.  It isn't clear that the Semantic Web and XML usage
are the same thing, nor is it clear that they should be.  Distinction and
acknowledged disagreement doesn't have to mean schism.

I feel strongly that the current 'status quo' position I outlined earlier
can accommodate both sets of needs, while sparing the XML community another
set of complex interoperability issues.  At the same time, definition of
what namespace URIs are for - what, if anything, they reference, and how to
handle that - needs to be explored in an XML context.  Not an RDF context,
not an XML Schemas context, but an XML and namespaces context.  

I believe this can be done without disrupting the current approaches taken
by RDF and XML Schemas.

Addressing a broader set of needs, even if it costs some time, may keep us
from stepping on future possibilities before they've been considered.

(And yes, I even like RDF and appreciate what it's trying to do.)

Simon St.Laurent
XML Elements of Style / XML: A Primer, 2nd Ed.
Building XML Applications
Inside XML DTDs: Scientific and Technical
Cookies / Sharing Bandwidth
http://www.simonstl.com

Received on Monday, 22 May 2000 10:53:22 UTC