Re: regarding rdfms-identity-anon-resources

> I have a slightly heretical view that RDF is also about the RDF/XML
> documents. I take this view, at least in part, because there is
> information in the RDF/XML documents that is not currently being
> captured in the RDF graph or triple representations like qname
> mappings, anon resources, lang, occurrence vs. quoting of statements,
> etc..

This sounds quite heretical, yes :)
I would not quite agree with it because :
 - XML is about syntax, RDF is not.
   Nothing prevents you of handling the XML-Infoset view of an XML/RDF
document,
   but *that* view is not relevant to RDF processors.

 - RDF could be seen, IMO, as another kind of Infoset for XML
   Since XML-Infosets are about elements (i.e. syntactical objects)
   RDF is about resources (i.e. semantical objects)

The *lost* informations you mention are not critical to RDF :
they are not needed by the (graph) model,
but can be stored in it if needed (in a more or less pleasant way ;-P)

 - the context can be captured by means of "bagification"

 - about the xml:lang attribute of a litteral :
some suggest to use an intermediate resource, with rdf:value and xml:lang
some suggest to use data: URLs instead of literals

 - qnames mapping (if needed for re-serializing) may be stored as
   attributes of each resource...

  Pierre-Antoine

Received on Monday, 19 March 2001 07:33:44 UTC