> If that policy were reversed is the architecture group
> prepared to recommend an approach for encoding the
> non-English schemas? The language translations the registry
> imports are not (IMO) schemas, but rather just additional
> assertions about previously identified resources.
Hmmmm.... I'll sidestep whether they are "schemas"! Increasingly it all
just seems to me like sets of assertions about "our stuff"...!
But yes, there might be an argument that these "additional data sources"
should contain _only_ the supplementary data that is "supplementary" to
what you get currently i.e. only the assertions that involve the
non-en-US literal values:
e.g.
dc:title rdfs:label "Titre"@fr-FR .
dc:subject rdfs:label "Sujet"@fr-FR .
and these sources should _not_ repeat the language-independent
assertions which are available in the existing data, like
dcterms:alternative rdfs:subPropertyOf dc:title .
That depends on how we see the data being accessed/used. This approach
might be appropriate if they are always used in _addition_ to the
existing data. And that's how I read what was suggested here
http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2003Oct/0206.html
Which is interesting as that doesn't seem to me to fit so well with the
content negotiation approach.
With the content negotatition approach it seems to me the different
language data sources would be accessed/used independently of each
other, and used primarily as an _alternative_ to the en-US data, and so
it would seem more appropriate to repeat the language-independent
assertions too.
Pete
|