Bernhard Eversberg wrote:
>
> Collocation:
> Presently, it is difficult to get all volumes of a multipart together,
> be it monographic or serial/periodical. GBS scans individual volumes
> and records their title page titles without regard to series title in
> the metadata.
>
Ha, have you tried doing that in worldcat, or with your own catalog
records? Our standard cataloging practices do NOT make this easy. It
doesn't help that an individual record _could_ be for the multi-volume
set OR could be for just one volume in the set, depending on what the
catalog library held at the time they cataloged. (Bernhard has written
about how German cataloging handles multi-volume sets a LOT less
ambigously).
Now, granted, Google is practically the _expert_ at trying to pull
meaningful data out of soup where it's not clearly expressed, that's the
business they are in. I have no doubt that if they decided to throw
sufficient resources at the problem of collocating multi-volume sets,
they could arise at a reasonable (but not perfect) approximation.
I would also be far from shocked if Google says "Um, that's really not
worth to us how many resources it would take to figure out."
The idea of libraries collectively "demanding" that Google figure out
how to get meaning out of our data that we haven't managed to encode in
an un-ambiguous machine-readable way in the first place (not only
legacy, but we STILL don't do it right)... while at the same time
complaining that all Google ever does is take from us and we'd rather
they didn't have our data at all or had to pay a lot of money for it...
It's pretty ironic.
Jonathan
Received on Mon Sep 14 2009 - 13:41:51 EDT