RDFX: My "vision" for an RDF/XML serialization, for both assertions and queries (RDF 2.0)?

Given all the recent discussion about qnames vs. URIs, namespaces,
and other issues with the present RDF/XML serialization model,
I thought to post something that I've been working on for a while,
which is an XML serialization for RDF which is compatable with the
revised RDF/XML serialization in expressive power, but which has the
following properties:

1. No qnames are used to denote resources, only URIs.
2. Provenance and statement attribution is central to the model.
3. Literals *could* be allowed to be subjects, should that later
   be permitted, and that is anticipated by the serialization.
4. It provides for the expression of queries and scope/context
   in RDF using the same serialization model.
5. It provides for the expression of implications similar to CWM's
   log:implies.
6. Facility for inclusion of other schemas for modular design.

This is a work in progress (though I consider it to be mostly done)
so comments are both welcome and hoped for.

Hopefully the examples will sufficiently illustrate how it works.

In short, attributes either identify resources or qualify
statements. Qualifying attributes (scope, source, authority)
apply to all statements/assertions within their XML scope.

One comment about qnames. Qnames are used only for elements
and attributes of the serialization, not for resources themselves;
with on exception. Qualification attribute qnames are mapped to
property URIs in the RDF graph -- however, no mapping or round
tripping issues arise as this is a fixed set with the explicit
mapping defined by the serialization/schema rather than a generic
mapping from any arbitrary qname to URI.

I expect to write some XSLT to map this serialization into
the revised RDF/XML syntax for testing. There will be an option
to either generate both triples and reifications, or just
reifications distinguished by RDF class (statement vs. assertion)
for query engines which would directly support that (I'm working
on one ;-) and which is a much more concise representation of
the RDF knowledge -- and also analogous to (compatable with)
the present graph model.

Cheers,

Patrick

--
               
Patrick Stickler              Phone: +358 50 483 9453
Senior Research Scientist     Fax:   +358 7180 35409
Nokia Research Center         Email: patrick.stickler@nokia.com