W3C home > Mailing lists > Public > w3c-rdfcore-wg@w3.org > July 2001

ACTION: literals in M&S

From: dehora <bill@dehora.fsnet.co.uk>
Date: Fri, 27 Jul 2001 17:18:09 +0100
To: <w3c-rdfcore-wg@w3.org>
Message-ID: <000b01c116b7$bae23e20$bc10893e@MITCHUM>
Hi folks,

This is really a marker message to let me split up a longer unfinished
document into digestible chunks and get some stuff to the list before
people start travelling to the f2f.


In summary form, the issues I've come across with how the M&S speaks
about literals are:

-confusion between string literals and xml literals:
probably this is a result of calling a piece of XML a literal as much as
anything, but people do tend talk about RDF literals as though they were
strings. Some subediting on the M&S would help to educate its readers.
Note that section 5 quotes literals as though they were strings.

-section 2's introductory description of RDF literals is misleading:
closely related to the above, this has to do with the examples in
section 2 which treat literals as strings: I believe these are meant to
demonstrate the conceptual simplicity of the triples form, but maybe
starting out with a triple containing an literal object isn't the best
pedagogical tool.

- parseType extensibility
parseType is being used as an extensibility mechanism (shining example
is DAML). It might be useful for the M&S to discuss parseType in this
sense rather than simply say it will get treated as XML. It might also
be appropriate for the wg to discuss reservation of parseType attribute
values of the form "rdf:*".

-literals as subjects
with the broad understanding that literals are structuring things rather
than strings, it may make sense to allow such things to become subjects.
It's easy to envisage some XML having associated metadata properties and
being "embedded" in some RDF: certainly this would be an option for
modelling xml:lang; as a property of a literal rather than a part.

-the current formal model says literals are XML
this struck me at today's telcon (july-27-01). I expect that it may have
implications for the Model Theory, n-triples and any kind of future
abstract syntax or surface syntax. As things stand, XML would have to be
embedded in all current and future RDF syntax forms. It seems to suggest
that XML _is_ the abstract syntax for literals, since section 5 doesn't
elaborate on the structure of literals. 

-semantics of literal objects:
what does it mean to say that a piece of XML is the value of a
resource's property? Do we depend on the what a machine can determine
about what the semantics of the property are? I don't know whether the
Model Theory captures this, but it would great if you guys could chew on
this at the f2f...

I will try to map these into known issues as I go, the issues list is
long enough as it is. Pointers to open issues on literals:

Why isn't xml:lang information represented within the RDF data model

Consider replacing literals with resources whose URI uses the data: URI

A literal containing XML markup is not a simple string, but is an XML

Should the subjects of RDF statements be allowed to be literals

How should a parser process namespaces in a literal which is XML markup?

Received on Friday, 27 July 2001 12:18:16 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 14:53:50 UTC