RE: #rdfms-xmllang

Graham wrote:
> At 11:40 AM 7/11/01 +0100, Brian McBride wrote:
> >FYI, the following can be found in section 6 of M&S:
> >
> >   The xml:lang attribute may be used as defined by [XML] to associate a
> >   language with the property value. There is no specific data model
> >   representation for xml:lang (i.e., it adds no triples to the data model);
> >   the language of a literal is considered by RDF to be a part of the 
> > literal.
> >   An application may ignore language tagging of a string. All RDF 
> > applications
> >   must specify whether or not language tagging in literals is significant;
> >   that is, whether or not language is considered when performing string
> >   matching or other processing.
>
> FWIW, I think this is broken and will require some kind of fix (as opposed 
> to clarification).  I'm agnostic about the nature of the fix.

This is one place where I think the implementation record shows that
the M&S 1.0 spec needs more info. I think the simplest thing to do is

1) Change n-triples to n-quadruples, where the 4'th field is the
value of the xml:lang attribute (if any).  (Warning, this is not
the only place where we are likely to want to add more columns to
the n-triples format to get to the 'real' model. Statement IDs or
'nests' or 'contexts' may be needed. We also probably need some
simple type fields, such as to indicate if the subject is a simple
subject or is an 'aboutEach', and to indicate if the object is a literal
or a URI. But I digress).

2) Postpone any more detailed work, such as identifying the source
and predicate for the statements that should be generated to represent
the xml:lang information. The 1.0 spec says that it does not add any
triples to the model. The 2.0 spec may say something different, such as
"Literals are assigned URIs according to the procedure X" and
"Literals may be assigned data types, locales, etc. by using that
URI as the subject of additional statements. Those additional statements
MUST be flagged in such a way that they can be suppressed if 1.0
conformant serializations of a graph are to be produced."


Ron

Received on Wednesday, 11 July 2001 23:50:54 UTC