W3C home > Mailing lists > Public > w3c-rdfcore-wg@w3.org > January 2002

Re: Datatypes and xml:lang

From: Patrick Stickler <patrick.stickler@nokia.com>
Date: Thu, 31 Jan 2002 12:42:21 +0200
To: ext Dave Beckett <dave.beckett@bristol.ac.uk>
CC: RDF Core <w3c-rdfcore-wg@w3.org>
Message-ID: <B87EEC2D.CCF6%patrick.stickler@nokia.com>
On 2002-01-31 12:18, "ext Dave Beckett" <dave.beckett@bristol.ac.uk> wrote:


>> Firstly, was it not decided that xml:lang was to be removed
>> or at least ignored?
> 
> Not at present:
> http://www.w3.org/2000/03/rdf-tracking/#rdfms-xmllang
> 
> Summary: "This is a mess - it is in the syntax and not in the
> model. Should have used an RDF vocabulary for language. It should
> be removed from the syntax."
> 
> ... 
> Currently: for discussion
> 
> Which is really, waiting for the datatypes/what is a literal?
> discussion to complete.

I stand corrected. Though I still don't see this as a datatyping
issue. A literal value with xml:lang qualification is not a
"kind of" xml:lang. The xml:lang attribute does not assign a
type to the literal.
 
>> Secondly, this is a matter of value qualification (or statement
>> qualification, depending on your particular bent)
>> 
>> At present, it seems the preferred way of handling this is
>> by using a blank node that has the literal value and
>> the qualifiers hanging off it. E.g.
>> 
>>    X ex:title _:1 .
>>    _:1 rdf:value "foo" .
>>    _:1 xml:lang "en" .
>> 
>> [or in RDF/XML:
>> 
>>    <rdf:Description rdf:ID="X">
>>      <ex:title xml:lang="en" rdf:value="foo"/>
>>    </rdf:Description>
> 
> That's OK, if you model like that, see below.  I don't think using
> xml:lang as a property is widespread.

What else could it be?
  
>> with the datatyping presumption
>> 
>>    xml:lang rdfs:range xsd:lang .
> 
> Doesn't that require schema support in order to use it?  xml:lang in
> M&S doesn't need that

Only if you want the datatyping to be explicit in the
RDF graph rather than left to the application layer.
I.e., I was presuming a datatyping interpretation of
xml:lang to be xsd:lang, not requiring the range
constraint to be made explicitly.

>> or an alternative is reification
>> 
>>    X ex:title "foo" .
>> 
>>    _:s rdf:type rdf:Statement .
>>    _:s rdf:subject X .
>>    _:s rdf:predicate ex:title .
>>    _:s rdf:object "foo" .
>>    _:s xml:lang "en" .
> 
> That attaches it to the statement, not the unicode string, see below

Nothing attaches to the unicode string. Literals do
not act as subjects.

You must use some form of idiom to infer properties of
literals (or rather of the values they denote).

It is not the literal that is "English" but rather the
value denoted by the literal.

After all, what does "pan" mean. Is it English? Spanish?
The literal itself means nothing. It's just a sequence of
characters. You must have context to determine meaning.
The xml:lang property provides context for the interpretation
of natural language strings.
 
This problem is disjunct from the problem of interpretation
of lexical forms per datatype.

>> So, this really isn't a datatyping issue at all. It's
>> a qualification/scoping issue. The literal "foo" does
>> not have a datatype of (xml:lang,"en") but rather that
>> is a property of the value (not the literal).
>> 
>> Right?
> 
> I don't know about prefered way;

Sorry, I meant a preferred way to express value or statement
qualification, not for using xml:lang. Apologies for being
unclear.

> I think that since we use XML for the syntax, and M&S mentions
> xml:lang, we must continue to support it.

In principle, I agree, though it is a difficult issue. We would want
support of xml:lang, as with e.g. XML Schema datatypes, to be provided,
but by generic mechanisms and not with special interpretation in
either the syntax or model.

> The in-scope xml:lang
> attributes affect all inner literal element content so that I want to
> know about these statements:
> 
>   <rdf:Description rdf:ID="X">
>     <ex:title xml:lang="en">foo</ex:title>
>     <ex:title>foo</ex:title>
>   </rdf:Description>
> 
> and how you think they should be modelled?

I would treat this as two statements where the first is qualified
for language but the second is not. Whether a given application
wishes to consider these statements synonymous in a give context
is undefined.
 
> Do literal-labelled nodes become (unicode string, optional xml:lang)
> pair as the quoted paragraph above I think indicates - "part of the
> literal"?

I hope not.

Patrick

--
               
Patrick Stickler              Phone: +358 50 483 9453
Senior Research Scientist     Fax:   +358 7180 35409
Nokia Research Center         Email: patrick.stickler@nokia.com
Received on Thursday, 31 January 2002 05:42:52 EST

This archive was generated by hypermail pre-2.1.9 : Wednesday, 3 September 2003 09:44:04 EDT