RDF DT depends on resolution of #rdf-qname-uri-mapping

I generally agree with both Patrick Stickler and Graham Klyne's most
recent comments, particularly that the QName <-> URI issue is not one
easily solved under the current RDF syntax + model. On the other hand,
this architectural issue has implication for the following open issues:

    - XML Schema datatypes are identified by QName

    - QNames are essential XML identifiers. In the absence of a
consistent mapping of QName -> URI, and lacking a representation of a
QName in the RDF MS, I wonder how one could represent an XML literal as
other than a character string?

    - this general issue is closely related to the specific issue
regarding issue that I raised. These issues could be merged, with the
specific issue serving as an example


If the WG feels that it is beyond the current charter to fix the
QName-URI problem, which may indeed be the case, then I am concerned
that the following areas will also be placed beyond the current charter:

1) a more robust discussion/solution to the RDF datatypes issue which
ought involve XML Schema datatypes 2) robust handling of literals that
are other than simple strings e.g. XML, integers in general

Arguably such changes to the RDF model do go significantly beyond RDF 1.
My personal viewpoint is that the change is worth making because the
ability to make assertions regarding data  e.g. that "010" and "10"
represent the same number, or to make inferences in general e.g.
{xxx:yyy math:equals "010"; xxx:zzz math:equals "10"} log:implies
{xxx:yyy math:equals xxx:zzz} etc.

On the other hand I agree with Graham that RDF 1, as it is currently
used, is internally consistent. The issue becomes problematic
specifically when issues regarding integration with the XML/XSD/XQuery
arise. RDF DT really ought leverage the excellent work done by other W3C
WGs in the datatyping area.


Received on Wednesday, 23 January 2002 09:23:46 UTC