>>>Roland Schwaenzl said: > Dear Dave, dear Patrick, > > > think it is one of the major problems > with the current RDF drafts, that they > say a lot of nice words about datatypes, > but in fact do not allow to "define" > any. RDF datatypes don't require 'defining', although you can do that somewhat. They requirements are: 1) identification of the datatype 2) giving its lexical form. You can 'define' if you like them with: ex:myDatatype rdf:type rdfs:Datatype but that is implied by: .. ex:prop "foo"^^ex:myDatatype If you want to do such defining, then you can give more information about the ex:myDatatype. (see below) RDF datatypes are extensively explained in the RDF Primer Typed Literals http://www.w3.org/TR/2002/WD-rdf-primer-20021111/#typedliterals and RDF Concepts and Abstract Syntax Datatypes http://www.w3.org/TR/2002/WD-rdf-concepts-20021108/#section-Datatypes which tell you what they are and how to use them. However, there is more work being done on this which you can read in RDF Primer section about defining vocabularies: [[ 5. Defining RDF Vocabularies: RDF Schema @@This section currently does not include a discussion of rdfs:Datatype, and the declaration of specific datatypes in schemas, and requires further synchronization in a number of other areas with the RDF Vocabulary Description Language specification.@@ ]] -- http://www.w3.org/TR/2002/WD-rdf-primer-20021111/#rdfschema I think this has been done in the next draft, not out yet. > I can't see how to get a literal2value > map in RDF - there doesn't even seem to be > a standardized way to point to a resource, > which would provide such a map somehow. We leave that to existing datatype systems such as W3C XML Schema (WXS) datatypes, for example. There is extensive documentation and implementation of that. If your application handles them, by using those datatype URIs, you get the mappings. > This might be one of the reasons rdf-semantics > is that weak on datatype entailments. It was a non-goal to duplicate all the datatype work done elsewhere, such as in WXS datatypes. RDF datatyping just needed identification of the datatype along with the lexical form (string). The mapping is up to the datatype designer. > Think there is really missing some piece of > vocabulary in the rdfs draft as compared > to the rdf-concepts draft. What piece? You aren't saying exactly what is missing. > I don't think DCMI can issue any recommendation > based on drafts, which may change till they > reach W3C recommendation - or at least a level, > where W3C gives an official "call for implementation" > level to the papers, which are now just drafts. These documents may change at any time but we are preparing them for what the W3C calls Last Call Working Draft, so the WG is getting ready to consider them for the "call for implementation" which is the stage following that, W3C Candidate Recommendation. This is already begining to be implemented, there is at least one implementation of RDF datatypes with XSD using the Euler system. Dave