Weinheimer Jim wrote:
> In Google Books, the American Classical School at Athens has made lots of books available for free. These are especially useful to my students and faculty. I do not have time to catalog them all, and in any case, they seem to still be adding them. I made the following record: http://www.galileo.aur.it/cgi-bin/koha/opac-detail.pl?bib=22971 where the 856 field uses a Google query. So, I cataloged it as a type of collection. Quick and dirty, I admit, but otherwise, the materials in Google can't be found.
This is clever. A next step occurs to me. Instead of simply making a
collection record, an automated process could be created that uses that
same query to actually download metadata from Google Book Search, using
the Google Book Search Data API. It won't give you marc (if you were
doing similar things with Open Library/Internet Archive, it might, but
not Google Data API) , but it'll give you simple fields like author,
title, date, publisher. An automated process can transform this into
MARC (not necessarily valid AACR2 marc you'd want to share with OCLC,
but something your catalog is capable of indexing and transforming), and
you could actually include the individual items in your local catalog.
And have them updated with new items (and/or corrections to data) on a
regular basis, re-running your automated process.
[I wonder how likely users are to stumble upon your collection record at
the moment? Seem like they'd be more likely to stumble upon the
individual items, even if they are 'bad' limited records, possibly
without even LCSH. Or you could automatically assign some generic
collection-level LCSH to them all, although that's not the way LCSH is
intended to be used, it might be effective anyway.]
This is actually totally do-able in the present day, I believe. (Since
Google provides the Data API. If they take it away, it becomes no longer
do-able with Google Books, but still do-able with Open Library/Internet
Archive instead).
I would foresee us starting to do more and more of this kind of thing,
and standardizing methods to do it so a new tool doesn't neccesarily
need to be written for every third party source you want to harvest
records from.
Jonathan
Received on Wed Oct 28 2009 - 12:14:04 EDT