Andy, Pete, and All,
Some comments on Guidelines for implementing Dublin Core in
XML. Actually I've commented on this document before and was
following guidelines in an earlier draft, but it seems to have
changed since then so here are more comments.
I think this is a very welcome and needed document for everyone
who is producing DC in XML. Even in cases where this is within a
specific application, so that interoperability isn't an issue, there is
much to be said for following best practice. I would think that some
of the recommendations here could be taken forward to HTML
guidelines - things like element names and capitalisation.
1. On the subject of interoperability, the place my applications
need to be interoperable is when supplying Z39.50 search results
in XML/DC conforming to the Bath Profile. This specifies using the
CIMI XML DTD for simple DC, which is not the same as the
XML/DC included in these guidelines. I am not suggesting DC
follow the CIMI DTD, but maybe something should be said about it
in this document?
2. 5.2, Recommndation 6 shows
<dcterms:available>2002-06</dcterms:available>
rather than enclosing it in <dc:date>...</dc:date> tags.
This is the opposite to the recommendation in the previous version
(draft 3) which I was using. [You need to pay attention to keep up
in the DC world!] What is the reason for this change of
recommendation?
As other people have pointed out it will make it more difficult for
applications to dumb down. Though maybe an assumption that an
application using DC will understand DC, even if it hasn't read the
schema, is reasonable.
It seems that by making this recommendation you have promoted
all the element refinements to be top-level elements in their own
right. I guess my problem is that I tend to view the world in an
object-oriented and/or hierarchical fashion, which is probably why I
have difficulty in appreciating RDF (could be why many others are
going with XML rather than RDF). Elsewhere (not in this document)
I've also seen 'dctype:Collection' which is promoting values of the
DCMI Type list to elements in their own right as well. This may be
all OK given namespaces, but I wonder if there has been DCMI
consensus on this. However, I then notice something in the
Schemas Examples document called 'Unqualified DC' which
includes 'top-level' elements in the dcterms namespace, which
implies there is a hierarchy within dcterms. So I'm feeling confused!
3. Recomendation 7. An encoding schemes namespace has been
introduced called 'dcxml'. I've not seen this before. I understand
why it's needed, because the name 'scheme' needs to be pinned
down. But I'm not too sure about the name. I know it's just an
identifier but calling it 'dcxml' seems to preclude it's use elsewhere
(at least to human readers), eg in HTML. Maybe something to
imply it's defining the DC qualifiers schemes or attributes (in the
XML sense) dcattr? (Though I later noted that the dcxml.xsd
schema includes other things, but still about qualifiers/refinements.)
4. Recommendation 9.
Maybe something should be said about where the xml:lang
attribute has appeared from, even if it is standard XML. I think there
should be some indication of which language scheme is
recommended. There appears to be some inconsistency. The
examples show "en", but at the end of the bullet list of 5.1 it says
"en-GB".
5. The example schemas are very useful. Dare I bring up the
subject of DTDs?! Is there any plan to add example ones? I don't
believe DTDs are dead yet, and some applications require one.
6. I had some problems reading the schemas and examples via the
web links from the page - I had to save them locally and then read
as text files. I realise that this is a Netscape problem, but I think
that DC documents should be generally readable. Maybe an HTML-
readable version could be at the end of the hypertext links (or just
.txt would help).
7. In dc:xsd, the specification of the language attribute is 'x:lang'. I
would have expected this to be 'xml:lang' but maybe there's some
reason I don't understand.
8. The distinction between simple, unqualified and qualified is not
immediately obvious, especially as 'simple' and 'unqualified' are
sometimes used as synonymous. I can see why it's been made. It
is caused by the fact that 'audience' is in dcterms, the rest of
which is element refinements.
9. Could you think of a better name for dc11q.xsd? I suppose it's
meant to suggest DC 1.1 + qualifiers, but it looks more like '11'!
Hope this helps,
Ann
--------------------------------------------------------------------------
Mrs. Ann Apps. Senior Analyst - Research & Development, MIMAS,
University of Manchester, Oxford Road, Manchester, M13 9PL, UK
Tel: +44 (0) 161 275 6039 Fax: +44 (0) 0161 275 6040
Email: [log in to unmask] WWW: http://epub.mimas.ac.uk/ann.html
--------------------------------------------------------------------------
