Natural RDF

Hi list,
Just for fun I decided to put together a proof-of-concept doc to see what a
purely syntactic definition of RDF might look like. The basic idea is
pretty simple; just define a meta syntax, some relations on graphs (shape
isomorphism, congruence), and a bunch of "reduction" rules (a/k/a
inferencing rules, transformation rules). The rules are modeled on the
introduction/elimination rules of natural deduction.
It's not yet complete but is sufficiently detailed to establish the
plausibility of the approach. I think. The result looks pretty promising
to me; I suspect a rigorous, simple, and clear definition of RDF without the
model theory stuff could probably be done in under five pages. Comments?
RDIL: RDF with a Human
Face<http://blog.mobileink.com/2011/03/rdil-rdf-with-human-face.html>
Thanks,
Gregg