ons 2006-05-24 klockan 15:17 +1200 skrev Douglas Campbell:
> Mikael,
>
> Thanx, I look forward to reading your next draft.
It's there now:
http://kmr.nada.kth.se/~mini/dc-rdf/2006-05-23/
>
> I don't have any use cases but I feel there may be and I just haven't
> found them yet!??
Yes, of course. I just wanted to know what you had on your mind...
>
> As part of the DC-Date work I've been thinking there may be a case for
> a new date encoding scheme that allows embedding of textual
> descriptions (eg. for indicating a period like "Baroque" alongside a
> numeric representation), which of course means both syntax and
> language may need to be specified. But that's not very firm yet, eg.
> I'm thinking ideally it would allow multiple labels to allow the date
> value itself to include multiple languages.
Note that you can also do this by specifying multiple value strings, one
with a numeric representation, and the others being plain text,
language-tagged. This does not solve the "single string encoding" that
you proposed, but maybe that's the wrong approach? Or maybe in that case
you should even embed the language tag inside the string?
> I can certainly see a case for embedding HTML formatting in a rich
> representation (eg. a description with paragraphs), but I guess you
> could do that in XHTML. The issue is that if you can't specify the
> syntax encoding you don't know whether the XML is XHTML, MathML, or
> whatever....? I also want to be able to store multiple rich
> representations in multiple languages (eg. paragraph/HTML marked up
> description in different languages).
Right, the only was to know the kind of XML is by looking at the XML
namespaces used within the XML literal. And as Pete said, language
tagging is explicitly meant to be done inside the literal.
The problem with introducing simultaneous data typing and language
tagging is that the RDF constructs become much more complex... and I
think we're already pushing the limit of what complexity people will
endure for DC metadata...
>
> I'll need to keep thinking about the other RDF structural stuff -
> you've given some pointers for me to think about. I guess there's a
> real reliance on looking up RDF schemas to know that dcrdf:valueString
> is a sub-property of rdf:value, etc. Most of the apps I build don't
> have that lookup ability so rely on "basic" RDF constructions.
Yes, I agree. I think I'm personally still keeping the option open to
fall back on rdf:value and ignore the fact that it's slightly
underspecified, for that very reason.
/Mikael
>
> Thanx,
> Douglas
>
|