Let's get the Literals out of the RDF Graph

It seems to me that literals simply have no place in the RDF graph or
the model theory.  Literals are only an issue for RDF Graph encoding
languages like RDF/XML or RDF/N-Triples.

A parser for RDF/XML should map a string literal into an unlabeled
node about which it adds the same information as the literal text
conveyed, namely that it is a sequence of unicode characters, and
which characters are in that sequence.  A generator for RDF/XML should
look for nodes about which it has this information and output them as
string literals.  Parser and generators should behave similarly with
XML/infoset literals, whatever they turn out to be.  

We obviously have to pick the node labels which will be used to
describe these things, of course, if we want our parsers and encoding
languages to be interoperable.  I've put together a rough ontology for
this, living at its namespace: "http://www.w3.org/2001/10/05-string".

Isn't this a lot cleaner?
 
     -- sandro

Received on Friday, 5 October 2001 12:01:26 UTC