Re: Dublin Core and multiple element concern

>>   2 Ntriples+ (with lang-strings):
>>     <thing> dc:title "title of thing"(en) .
>>
>
>As pointed out to Pat's message, this doesn't work
>because the literal itself is not English. It may
>have an English interpretation in a particular context
>but the literal itself, theoretically, could have
>meaning in multiple languages.
>
>C.f. http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Feb/0171.html
>
>>The DC idea being referred to here was historically labelled "dumbing
>>down".  This was for DC applications just looking for literals about
>>a resource (node) for some purpose, they follow the rdf:value arcs
>>until they get a string, and use that.  It is a flattening algorithm
>>for all the properties of a resource.
>>
>>See 3.2 of http://dublincore.org/documents/2001/11/30/dcq-rdf-xml/
>>
>
>I have the sinking suspicion that both the DC "poor man's"
>approach and the dumbing down principle, may need to be
>rethought and/or refined in terms of typed data literals
>rather than the present basis of strings that are presumed
>to have globally consistent meaning (but, based on all
>that has come out of these datatyping discussions,
>actually don't).

Im beginning to wonder quite what the point of the lang tag is 
supposed to be. Up to now Ive been thinking of it as a kind of 
datatype, but is that really appropriate? The basic reason for having 
datatyping is to allow a single literal to have many meanings. But 
that really isnt an issue with lang, right? I mean, nobody is going 
to think that a German title is an English title, and even in Canada 
when they write 'sauce soy sauce' on the bottles, its clear which 
part is English and which part is French. So I guess the point of 
lang is really to *record* the fact of the matter about which languae 
the literal is written in; its really a fact about the literal 
(string) rather than anything like a datatyping mapping applied to 
the literal. (I mean, what would the value space be for, say, lang:en 
? It would be the identity on English strings and nothing on others, 
persumably. But if we are dealing with titles, almost anything 
*could* be an English string, eg "10", or an Italian string 
("8&1/2"), or whatever. )

All of which leads me to think that the proper way to handle lang 
would be as a property of a literal, if only we had properties of 
literals....

Pat


-- 
---------------------------------------------------------------------
IHMC					(850)434 8903   home
40 South Alcaniz St.			(850)202 4416   office
Pensacola,  FL 32501			(850)202 4440   fax
phayes@ai.uwf.edu 
http://www.coginst.uwf.edu/~phayes

Received on Monday, 11 February 2002 11:10:44 UTC