Re: I18N actions [was: Re: WOWG: 4Sep proposed agenda]

I think we should consider a resolution supporting RDF Core on this - 
for technical reasons.  I have tried to do my due diligence on this 
issue (read the email in RDF threads, looked at I18N's objections) 
and my belief is that the i18n WG is not really right about RDF 
breaking internationalization (cf what Guus in his emails), and that 
what they want has the potential to break some of the use cases my 
group has been implementing for OWL (although I admit we've not yet 
done the multilingual version) -- in particular, I think the i18n 
group has a document-focused aspect -- they essentially assume one 
would have the language tags associated with a whole document. 
However, RDF allows us to create graphs by grabbing and merging info 
from many different sites, as we do on the owl.mindswap.org pages for 
the RSS news feeds among others.  However, if I grab a lot of stuff 
from a lot of different places (using an RDF query interface as a web 
service, for example) it is unclear what mechanism would get the 
responsibility for finding the "non-local" language tags (that is, if 
they're not in the actual RDF context or graph as found at some URI - 
but in the scoping document). If we move to an RSS feed in different 
languages, pulled as separate query results to a page, the query 
engine would have to process language tags in complex ways, instead 
of having them tied more directly to particular triples/subgraphs, 
which seems to be more what the RDF Core group has attempted to 
design.
  I realize that I'm simplifying the two proposed solutions and 
painting them without shades of gray, but the key thing to me is that 
the "lifecycle" of RDF and OWL is envisioned somewhat differently 
than the lifecycle of many XML applications, and I think the RDF Core 
solution has better "Grokked" the way RDF documents and applications 
are likely to need to process the language tags.
  Obviously there is an infinitely large design space that could be 
explored for a better solution, but I don't think the i18N folks have 
produced any real evidence that the current design is broken -- and 
thus it seems like a lot of work could be done for little reason.
  -JH


At 11:49 PM +0100 9/3/03, Guus Schreiber wrote:
>Dan Connolly wrote:
>
>[..]
>
>>  --- 5. RDF Internationalization
>>  ACTION: Jeremy to send his discussion of some of the issues re
>>  xml:lang and literals to WG.
>>                               From: Jeremy Carroll <jjc@hplb.hpl.hp.com>
>>                                 To: www-webont-wg@w3.org
>>                            Subject: On RDF/I18N issue
>>                               Date: Mon, 01 Sep 2003 16:34:22 +0100
>>  http://www.w3.org/mid/3F5366FE.20109@hplb.hpl.hp.com
>>
>>  ACTION: Guus S. will review.
>
>I have read Jeremy's note plus links in detail. Overall, RDFCore 
>seems to have made a well-motivated  rational design choice in a 
>space of conflicting requirements. The arguments against the 
>alternatives proposed by I18N are compelling. The current post-LC 
>design works for OWL and is, from a WebOnt perspective, an 
>improvement over the LC design.
>
>>  ACTION: Guus Schreiber will send some examples of use of xml:lang to
>>  webont mailing list.
>>
>
>Well, the type of examples mentioned in the RDFCore/I18N discussion 
>did not come up in our work, where we use language tags in a very 
>simple way. For example, to represent the IconClass category 2 (one 
>of the top categories; IconClass works with a compositional 
>letter/digit id system), which stands for "Nature", can be 
>represented as the following instance of a class ic:Category:
>
>   <ic:Category rdf:ID="2">
>     <rdfs:label xml:lang="en">nature</rdfs:label>
>     <rdfs:label xml:lang="de">Natur</rdfs:label>
>     <rdfs:label xml:lang="fr">nature</rdfs:label>
>     <rdfs:label xml:lang="it">natura</rdfs:label>
>     <rdfs:label xml:lang="no">naturen</rdfs:label>
>     <rdfs:label xml:lang="fi">luonto</rdfs:label>
>   </ic:Category>
>
>Guus
>
>--
>Free University Amsterdam, Computer Science
>De Boelelaan 1081a, 1081 HV Amsterdam, The Netherlands
>Tel: +31 20 444 7739/7718
>E-mail: schreiber@cs.vu.nl
>Home page: http://www.cs.vu.nl/~guus/

-- 
Professor James Hendler				  hendler@cs.umd.edu
Director, Semantic Web and Agent Technologies	  301-405-2696
Maryland Information and Network Dynamics Lab.	  301-405-6707 (Fax)
Univ of Maryland, College Park, MD 20742	  *** 240-277-3388 (Cell)
http://www.cs.umd.edu/users/hendler      *** NOTE CHANGED CELL NUMBER ***

Received on Wednesday, 17 September 2003 11:46:52 UTC