- From: Boris Motik <boris.motik@comlab.ox.ac.uk>
- Date: Thu, 26 Mar 2009 14:47:03 -0000
- To: "'W3C OWL Working Group'" <public-owl-wg@w3.org>, "'Axel Polleres'" <axel.polleres@deri.org>, "'Jie Bao'" <baojie@cs.rpi.edu>
Hello, I've just done some work on the rdf:text specification. (BTW, it would be good to add the URL http://www.w3.org/2007/OWL/wiki/InternationalizedStringSpec to the sidebar of the Wiki; otherwise, one always has to hunt for it around.) Here is the summary of the changes: - I've changed the definition of characters to be exactly as in XML Schema, which is exactly as in XML 1.1. Indeed, Unicode provides for 1,114,112 different code points. Furthermore, XML 1.1 disallows the use of some of them, which leaves us with exactly 1,112,061 code points. I've updated the example in the document to this extent. I was not aware of this -- sorry for the confusion! - I've changed slightly the definition of language tags. In the previous version of the spec, it was unclear whether the language tags need to be validated against the IANA registry. The current solution is to make our definitions independent of the IANA registry. (Otherwise, the semantics of OWL 2 ontologies and the consequences one can draw from them might change each time the IANA registry is updated -- clearly a bad situation.) - In response to numerous comments (e.g., by Addison Phillips), I've changed the definition of the facet for language matching. We are currently using RFC 4647 instead of arbitrary regular expressions. In order to reflect this change, I've renamed the facet to rdf:langRange. This is because we now use the so-called language ranges from RFC 4647, and I thought it would be good to make a clear distinction with the xs:pattern facet (which is a regular expression). We are using now the so-called basic language range matching. RFC 4647 defines also a more expressive extended language range matching; however, I am really not sure how to implement this in OWL 2. (Note that we don't just need to match a language range against a language tag; instead, we need to be able to test emptiness and/or finiteness of such ranges, and we need to be able to intersect the language ranges with regular languages; for extended language ranges this seems quite complex.) - In response to a comment by Jos de Bruijn, I've promoted two subsections to sections. - I've done some general editorial work on the spec. I've moved the discussion of compatibility with RDF into the introduction, and I've tried to keep the definitions crisp and clear. There are still two editorial comments left in the document, but these are in the functions part; I believe that these need to be resolved by RIF. Please let me know should you have any comments. Regards, Boris
Received on Thursday, 26 March 2009 14:48:14 UTC