- From: Jan Wielemaker <J.Wielemaker@vu.nl>
- Date: Tue, 27 Sep 2011 11:19:30 +0200
- To: Andy Seaborne <andy.seaborne@epimorphics.com>
- CC: <public-rdf-wg@w3.org>
On 09/27/2011 09:55 AM, Andy Seaborne wrote: > See Jeremy's message: > > http://lists.w3.org/Archives/Public/public-rdf-wg/2011May/0425.html Interesting reading. My quick reading interprets this as ``we can say sensible things about language identifiers in OWL, but it is not easy to get complete and right''. Would that be a reasonable interpretation of the paper? I think my ideal would be that I can use a language tag as an identifier and I can apply XML language matching logic to them. Would that be possible? Maybe by adding the original tag as an attribute to the language URI. So, "foo"@en is "foo"^^lang:en en { lang:en something "en" }? If something in the spirit of 3 doesn't work, I do not see much profit in any of the alternatives and would simply stay at option 1 (and make "foo" equivalent to "foo"^^xsd:string). --- Jan > The details of language tags don't map onto datatypes very well (e.g. > scripts which are different lexical forms so don't work with > sub-datatypes). It would need a significant amount of time (WG time) > to attempt an option 3 approach. > > Personally, I don't think there is a solution that respects the work > people have put into language identification in ISO 639 etc. But if > there is, option 2* does not completely preclude option 3* as later > work as it only adds the datatype at the root of the subtree. > > And what about RDF/XML (rdf:datatype, xml:lang) for option 3? > > For me, rdf:LangString proposal (any option 2) at least gives all > literals a datatype which is a (small) step forward, and we're > changing plain literals anyway.
Received on Tuesday, 27 September 2011 09:20:40 UTC