towards an ontology for XML-related resources (was: [namespaceDocument-8] RDF and RDDL)


----- Original Message -----
From: "Jonathan Borden" <>
To: "TAG" <>
Sent: Thursday, April 11, 2002 9:27 AM
Subject: Re: [namespaceDocument-8] RDF and RDDL

> I've
> started to convert the RDDL spec itself into a blend of XHTML + RDF
> -- you'll need to "view source" and look
> for the RDF statements intermixed with the XHTML. This RDF is intended to
> have exactly the same semantics as the RDDL XLink is intended to convey.

Tim's and Jonathan's recent posts have shown that RDDL and XPackage both
agree with the fundamental model of describing namespaces in RDF: the
namespace is the subject (the thing being described), any RDDL "nature" is a
property of the related resource, and that the RDDL "purpose" is the actual
RDF property (or predicate) that relates the namespace and another resource.

What Jonathan's RDDL+RDF model (which is similar to the model XPackage
proposes) makes clear is that RDDL+RDF is really just a specification of a
metadata ontology for XML documents. It describes things like namespaces and
DTDs and transformations and stylesheets for use in describing XML
namespaces, but they could be used to describe documents as well.

XPackage's primary concern is describing an ontology for packaging, but *out
of necessity* it also describes an ontology for XML-related documents
because there was no such ontology existing. I would rather modularize
XPackage so that it doesn't carry a definition for an XML-related ontology.

I propose, then, that we remove the XML ontology section from XPackage and
together work to make RDDL a complete ontology for XML-related resources.
The XML ontology described by XPackage is incomplete in some areas, as is
the ontology described by RDDL.

This unified XML-related ontology would describe concepts such as the

* rddl:namespace

* rddl:nature
* rddl:style
* rddl:transformation
* rddl:validation

This would allow us to remove the XML-related properties from XPackage and
would allow XPackage to only address what it's supposed to address:
packaging. This would also allow RDDL to become the standardized metadata
ontology for discussing aspects of XML, which would allow documents like
those currently envisioned by RDDL (associating DTDs and stylesheets to
namespaces), but would allow that ontology to be leveraged in *other* RDF
applications such as XPackage, to provide a stylesheet association mechanism
for normal XML documents, for example. It could even be mixed with *other*
ontologies, allowing such properties as mime:mediaType or unicode:script to
express the media type or Unicode script of a related resource.


Garret Wilson

P.S. See the RDDL example at to get
an idea of how such a modularization would work. Run the example through the
W3C RDF validator at .

P.P.S. Jonathan, there are some errors that don't allow to parse correctly in the W3C RDF

Received on Saturday, 13 April 2002 13:08:47 UTC