lang tag stuff ISSUE-71

I have reviewed the Carroll/Phillips paper Multilingual RDF and OWL.


The mechanism proposed there would probably not sit well with this group.

Here is a brief description of the core mechanisms which are expressed 
in an OWL Full sort of way:

Add a new property in a semantic extension rdflg:lang that relates a 
plain literal to its language tag, or an XML Literal to any language tag 
that occurs in it.

Add a new property rdflg:mainLang that relates an XML Literal to the 
outermost xml:lang value if there is a unique such value.

Add a new class rdflg:PlainLiteral for literals that are not typed literals.

Add a new class rdflg:XHTMLLiteral as a subclass of rdf:XMLLiteral for 
XHTML support.

====
The rest of the mechanisms are built up on top of these.

In particular a goal is to reflect the ontology from RFC 4646 and RFC 
4647 into OWL.

====

A smaller step around which I believe it would be easier to find 
consensus would be to create a new datatype constructor

LiteralsWithLang(langtag)

e.g.

LiteralWithLang("en-US")

that given a langtag is the (conceptually infinite) set of plain 
literals with that language tag (case insensitive comparison).

A triples syntax for that may be:

_:a rdf:type owl:PlainLiterals .
_:a owl:withLanguageTag "en-US"^^xsd:language .


I believe that in OWL Full this would be equivalent to the 
non-XMLLiteral part of the Carroll/Philipps paper, and that this is more 
likely to be acceptable to the OWL DL community.

This does not address XML and XHTML integration which was one of the 
goals in the paper.

It also leaves the non-trivial task of converting the ontology in RFC 
4646/4647 and the various ISO standards into OWL as an exercise for the 
reader.

I am hoping for some feedback before making a proposal to close.

Jeremy

Received on Tuesday, 22 April 2008 15:44:07 UTC