On Sat, 13 Oct 2007, A. Pagaltzis wrote:
* pkeane <[EMAIL PROTECTED]> [2007-10-13 06:15]:
All that said, I had (have) a hunch that deeper use of Atom and
esp. AtomPub could make the system even more open and flexible.
Well, as long as you have sane mapping between your RDF(ish)
attributes and Atom elements, that should work just fine.
And the "sanity" of those mapping will mostly depend on the end-users
(i.e., collection managers) -- this is part of our philosopy (approach?
survival-tactic?) to push as much control to the users as possible. But
what the heck -- if they create good mappings, the feeds look good out
there and if not, they don't....
If you want to upload via Atompub, it?s probably best to make the
mapping two-way, so your collection-specific metadata is
automatically created from the Atom Entry metadata when an upload
comes in that way.
I suspect that the only metadata I'll be able to grab upon upload is the
admin stuff (and I'll be the one creating those mappings). What we've
done so far is to tell folks that they get the admin metadata to start
with, and then they can add more once the items are in the system. I have
had SOME luck grabbing actual metadata from images manipulated in iPhoto
that the users have entered temselves, but that all relies on the
application using a standard (iPhoto uses IPTC embedded in the jpeg). I
believe Adobe is starting to use RDF, but it might be in a separate
'sidecar' file, which is a whole other can o' worms.
As long as the mapping is good, things should be peachy. (Ie.
make sure that the value domains of attributes mapped to each
other are exactly equal. An important reason for Atom to define
its own metadata elements rather than use Dublin Core was
precisely this ? that DC generally places fewer constraints on
values than the WG wanted in Atom for interoperability reasons.)
Yup. I like what the DC folks have attempted, but I feel like the Atom
effort really nailed things down in a much more useful way.
My ideal scenario, aside from making uploading simpler, is to create a
system that holds all kinds of digital stuff, cataloged all sorts of
different ways and if a faculty member wants to "repurpose" those assets
as a web site, or learning module or whatever, they can hire a student web
developer who simply has to set a few "data export settings" (don't want
to frighten them into thinking they are creating metadata mappings -- I've
had too many instances of faculty screaming in terror, running off into
the night never to be seen again after I gave them my "friendly
introduction to Dublin Core"), then creating a simple web-based "mash-up"
to use that data feed. Much simpler than hiring a programmer to build yet
another PHP/MySQL application that'll need to be maintained after they
leave. This whole "expose the data with Atom" is a way to enforce some
rational constraints.
I'd love to get to the point where Atom could fill in as a simpler
solution in all sorts of domains: simpler WebDAV, simpler Dublin Core,
simpler OAI-PMH, etc. Of course those other standards are trying to do
other things as well, but maybe Atom will hit an 80/20 mark that will
allow interesting work to be accomplished much more easily.
http://plasmasturm.org/log/463/
http://efoundations.typepad.com/efoundations/2007/07/app-moves-to-pr.html
-peter keane
Regards,
--
Aristotle Pagaltzis // <http://plasmasturm.org/>
