Paul Prescod:
> Well, the devil is the details. I've studied the standard web services
> architecture and found it wanting. I don't understand why you WANT to
> replace the RDF Model (not its syntax...I've already said you
> could use
> the model without the syntax). Can you convince anyone that there's
> something seriously wrong with the RDF model and that you know how to
> fix it?
That's a very good question. If by the RDF model you mean the ability to
attach semantic information to a schema, then I think that this is a very
good and useful thing. If you mean that a model separate from the document
schema must be created and maintained, then I do think that something is
seriously wrong with it (and I also think I know how to fix it). Whether I
can convince anyone of this is what I'm trying to find out.
What bothers me about RDF is the notion that a new schema needs to be
created that has significant overlap with the document schema and a pretty
hairy syntax (not that XSD doesn't, but that's a whole other story). At the
end of the day, I can deduce the RDF model from the annotations in the
example I presented and process it just like a model expressed in normal RDF
syntax. This is very similar to what Didier did by creating a URI scheme
called "soap", thus enabling access to SOAP resources via URIs.
What led me to bring this is up in the first place is that the syntax *IS*
important. We seem to agree that this is the case with XML and with web
services. If we want to create a "semantic web" with as much useful semantic
information as possible, then it makes sense first of all to leverage the
semantics inherent in document schemas, since huge effort is being poured
into creating these (witness UBL, CIDX, XBRL, etc., etc.). Much of the work
of designing a document schema involves expressing useful semantics such as
cardinality, data ranges, parent element types, etc.
It seems crazy to throw all of this out and force people to create an
entirely new schema in order to do semantic querying. Much more sensible is
to augment the information in the schema with further semantics, as needed.
The end effect is to expose a model exactly like RDF (or even the RDF model
itself, why not?), but to do it in a way that is far more harmonious with
the underlying principles of XML.
Anyone convinced?
Matt
