On Sun, 1 Dec 1996, Terry Allen wrote:
> What I find in
>
> http://www.nlc-bnc.ca/ifla/documents/libraries/cataloging/metadata/dublin2.htm,
> "A syntax for Dublin core Metadata: Recommenda..." is an example using such
> syntax and a statement of consensus for using HTML META. The good authors
> don't even provide a recommended production to describe the parenthetic
> syntax. It's just presented as a possibility. I think it's still worth
> warning against trying to nest an admittedly "arbitrary syntax" within
> HTML, especially if it is meant to disambiguate diluted semantics.
Sorry, I got a "Not found" for that document. Could you repost the URL
please?
> | We can decide how to encode author affiliation and say so, and if
> | we say it is part of Author, it isn't tag abuse. Of course, if
> | we put it into a field called "DATE", people will look at us oddly,
> | so it does have to be at least marginally plausible.
>
> It's tag abuse because it waters down the semantics of Author
> (Author's name) and, as can be seen from the HTML Jon Knight supplied,
> it makes it more difficult to extract information from the
> (still underpowered) combination of attribute names, attribute
> values, and internal, non-HTML syntax (in syntax-free DCese,
> element name, qualifier names, and a potentially uncontrolled
> range of qualifier values).
I strongly disagree with this. It does not make it difficult to extract
information about the author at all; if anything it makes it easier
because the default is just a name (which is what, if I understand you
correctly, is what you want) but the qualifiers also allow software to
determine whether the information is other details about the author (I'd
expect those to be _much_ rarer than names but I think we've got to
provide for them). If there are unrecognised/unregistered qualifiers the
software is completely free to ignore them (and maybe the entire element
as well). I don't consider this to be underpowered at all. We're not
aiming to recreate USMARC or TEI headers after all.
> I grant that you can reserve TYPE=name for the case now covered
> by Author, but you still don't have a way to associate miscellaneous
> author information with the author's name, a problem neither you
> nor Jon answered. Yet if you add a pointer as a qualifier on
> Author, you get precisely the association you need. That shows
> that the tag abuse is deleterious.
A pointer to what? A USMARC record? After all that's the example in
<URL:ttp://www.oclc.org:5047/oclc/research/conferences/metadata/dublin_core_report.html>.
Whoops, we've just increased the difficulty of using DC to the point where
its only usable by trained cataloguers (I'd have to read the MARC manual
again to create one of the many author related fields for example). I
think I'd rather stick to the "underpowered" qualifiers, thanks.
> I Wasn't at Warwick, and have been unable to get a clear fix on what was
> *decided* there.
Well take a look at Renato Iannella's trip report up at
<URL:http://www.dstc.edu.au/RDU/reports/warwick.html>. As he says there
were a number of threads to the activity at Warwick. Some of the outcomes
were the desire for supporting the embedding of DC metadata within
resources (mainly with HTML META elements), the creation of a concrete
Dublin Core SGML DTD for long term storage of metadata independently of
the resource and the formulation of the Warwick Framework idea for tying
arbitrary metadata packages together (including, but not limited to,
Dublin Core concrete representations). On this mailing list we've been
mainly concentrating on getting Dublin Core "beefed up" and embedded in
HTML as that was viewed by many at the workshop as the most urgent matter.
> There's been a lot of discussion and further work;
> isn't that all by way of presenting proposals to modify and extend DC 0.1?
If DC 0.1 is the concrete representation-free version of Dublin Core that
existed prior to Warwick (ie from the original Dublin meeting) then, yes,
that's exactly what we've been doing.
> DC 0.1 doesn't specify a concrete representation (perhaps its highest
> virtue), but it does supply a list of elements and what we're now
> calling qualifiers, all with definitions, without a list of qualifier
> values.
The original 13 elements are what we've still currently got (modolo some
potential renaming and maybe the odd addition from the Image guys?) but I
didn't read the original DC report
(<URL:http://www.oclc.org:5047/oclc/research/conferences/metadata/dublin_core_report.html>)
as saying these were the definitive list of qualifiers (especially seeing
as people have already pointed out problems with them). However, maybe we
should name the versions as:
v0.0 : Abstract only DC from the original Dublin Meeting
v0.1 : Concrete representation stemming from Warwick, including
the use of HTML META elements to embed it in documents
v1.0 : The outcome of the current battles, erm, discussions.
> I take it that you are proposing a change to the definition of Author
> in 0.1, and that the User Guide we discussed here recently is also
> such a proposal. If not, please put me right.
I didn't think that what I've been using was a change at all but more a
formalisation of what was intended for the abstract element defined in the
original Dublin report. Maybe we'll just have to agree to disagree on
that?
Tatty bye,
Jim'll
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Jon "Jim'll" Knight, Researcher, Sysop and General Dogsbody, Dept. Computer
Studies, Loughborough University of Technology, Leics., ENGLAND. LE11 3TU.
* I've found I now dream in Perl. More worryingly, I enjoy those dreams. *
|