Regarding multiple language capabilities and a Web Ontology Langu age.

It seems to me that there are two goals.  WOL should be language
neutral and multi-language capable.  These are generic W3C goals that
we support.  Do we need to restate them?  If so, that is fine, but we
need to be clear that whatever we provide here is a convenience, like
'rdfs:comment'.  We can make no commitment to providing any sort of
semantic interpretation of these tags.  They are just labels.
 
Does the W3C XML Schema syntax provide a multi-lingual capability?
Kind of.  We are encouraged to provide 'xml:lang' attributes for
schemas as a whole and for annotations.  But the primitives are all
English and there is no mechanism for tag synonymy across languages.
 
Fundamentally, a good multi-language system will depend on tool
support.  When reading an ontology most users will not want to look an
element labeled:
 
<rdf:Description rdf:about="dog">
 <rdfs:label xml:lang="en">dog</daml:label>
 <rdfs:label xml:lang="de">Hund</daml:label>
 <rdfs:label xml:lang="it">cane</daml:label>
 <rdfs:label xml:lang="fr">chien</daml:label>
 <rdfs:comment>Insert the rest of the EU languages here.</rdfs:comment>
 ... 
</rdf:Description>
 
Obviously, this will be permitted by whatever we come up with.
Various projects might even want to enforce this type of description
so that their multi-lingual teams can work together effectively.  But
this is not a problem to be solved by WOL.
 
- Mike

Michael K. Smith 
EDS Austin Innovation Lab 
98 San Jacinto, Suite 500 
Austin, TX 78701 
Work: 512 404-6683 
Cell: 512 789-4477 
Fax : 512 404-6655

-----Original Message-----
From: Raphael Volz [mailto:volz@fzi.de]
Sent: Thursday, December 13, 2001 12:44 PM
To: www-webont-wg@w3.org
Subject: Multilinguality



Hi all !

The topic of Mulitlinguality just came up at the telcon...

The RDFS Schema definition already contains xml:lang markup for "rdf:label"
properties.
xml:lang is intended for filtering of tags from an xml perspective,
if a user opens a document with a given locale, only those tags
that have been marked with his locale are retrieved (and those not marked
with the xml:lang attribute).

This encoding is also part of rdf using one of the abbreviated ways of
encoding
rdf in XML. Unfortunally this is not sufficient.
The value of the label is a literal, and can only therebe referenced by one
edge. Usually the same label can reference different concepts (i.e. each
meaning of a word: concept "Bank_1" being river bank, "Bank_2" being the
financial institution".

KAON (kaon.semanticweb.org) extends the RDFS metalevel with addtional
primitives for representing a (multi-lingual) lexical layer...
See http://kaon.semanticweb.org/2001/11/kaon-lexical.rdf
<http://kaon.semanticweb.org/2001/11/kaon-lexical.rdf>  The syntax is
pretty adhoc, but it could be used as a discussion basis.

The above n:m relation between concepts and there lexical representation
are actually standards in computational linguistics. Eventually our ontology
applications in text mining and document management
management scenarios always rely on such a mapping.

Regards,


Raphael