I am also stumbling in my UML class diagram over the wording of
section 2.3 of the Primer:
http://www.openarchives.org/ore/0.9/primer#Named_Graphs
"A Named Graph is NOT the RDF Graph itself."
Why not just accept that an RDF Graph can be a resource in the HTTP
sense with a URI assigned to it? In principle, resolving this URI
could produce a variety of content-negotiated representations
including application/rdf+xml, application/atom+xml, image/*, etc. If
Named Graphs were modeled this way, you could presumably do away with
the "extends RDF" comment and the "RDF reification" allusion.
As far as I can tell, there is a 1-to-1 relationship between an RDF
Graph and a Named Graph in ORE. Is there a use case where the RDF
Graph needs to have a different URI from the Named Graph? My vague
impression is that a Name Graph is only allowed to have one
representation. Maybe that's the difference?
I'll take my RDF pain pills for now and push on, but I'm skeptical
that this difference is real.
Jeff