- From: Jeremy Carroll <jjc@hplb.hpl.hp.com>
- Date: Mon, 8 Oct 2001 14:42:55 +0100
- To: "Aaron Swartz" <aswartz@upclink.com>
- Cc: <w3c-rdfcore-wg@w3.org>
Aaron: > Oh dear, I sure hope my RDF parser doesn't start returning > _that_ any time soon. We've built something of a layered > architecture in RDF, and I want to keep it that way. I see no > reason an RDF parser should have to believe in the RDF Model > Theory. Instead, it should just spit out triples for a > higher-level application to deal with. It should definitely not > be allowed to start adding or deleting triples (unless > specifically asked to by the application. > I find this paragraph more convincing now than I did last week. It suggests that rather than the conventional concrete versus abstract syntax distinction, we have a rather richer set of layers: Semantics: Model Theory Second Level Abstraction: Graph First Level Abstraction: Triples (Multiset) Concrete Syntax: RDF/XML & N-Triple I guess last week I was assuming that the first level abstraction was not useful. Jeremy
Received on Monday, 8 October 2001 09:43:15 UTC