May I suggest that we move this thread to the mailing list
[log in to unmask]
which according the what we decided in Canberra could be used as a
Dublin Core Implementors Workshop.
As Dave pointed out the other day it is maintained by
[log in to unmask]
Then on to the "minimalist" examples:
On Tue, 18 Mar 1997, Andrew Daviel wrote:
> On Tue, 18 Mar 1997, Diane I. Hillmann wrote:
>
> > <meta name = "DC.subject"
> > content = "(scheme=LCSH) Cataloging">
> > <meta name = "DC.subject"
> > content = "(scheme=LCSH) Acquisitions (Libraries)">
>
>
> .. in my simplistic way, I was thinking to store metadata in an
> associative array, so if I had content like
>
> <meta name = "Subject" content="About Gerbils">
> <meta name = "DC.Subject" content="A study of Gerbils">
> <meta name = "DC.Subject" scheme=LCSH content="Gerbils, a study">
>
> I might end up with
> element value
> "subject" "About Gerbils"
> "dc.subject" "A study of Gerbils"
> "dc.subject:lcsh" "Gerbils, a study"
>
> but in Diane's example the second value would overwrite the first.
You may store it however you want, but any implementation must
acknowledge the fact that every single element in Dublin Core is
repeatable (and optional).
.....
> (I think in PICS there is a concatenated string equivalent to two separate
> PICS-Label headers, while MSIE 3.0 ignored a second separate header ....)
>
> Incidentally, was the form
> <meta name = "DC.Subject.LCSH" content="Gerbils, a study">
> considered and rejected ?
>
The usage you are suggesting was rejected at the DC4, as I understood
the discussion. Only the type qualifier were to be used in this way,
whereas "scheme" and "lang" were to be handled some other way, which we
should discuss in html-metadata.
There is another, more fundamental issue, which is related to the problem
raised by Andrew: Grouping.
There was a grouping discussion in Canberra, and the problem has been
raised here before. The SGML encoding proposed for Dublin Core allows
for grouping, and the html-syntax put forward by Misha et al. can as
well. I think, however, that we should consider a semantics for
grouping. The reason for that is that it would help keeping the
grouping mechanisms in different encodings compatible with each other.
Yours,
Sigfrid
________________
Docent Sigfrid Lundberg, Fil. Dr. [log in to unmask]
Lunds Universitets Bibliotek
Netlab, Box 3, 221 00 Lund 046-222 36 83
|