On Tue, Aug 26, 2008 at 6:52 AM, Oliver Flimm <flimm_at_ub.uni-koeln.de> wrote:
> another step in linking library records would be to supply a mapping
> between individual records and the corresponding subject or
> classification scheme, e.g. with ISBN or something like Bibkey
> (http://www.gbv.de/wikis/cls/Bibliographic_Hash_Key).
Yes, these are definitely interesting. At a meeting at Open Library
earlier this year I listened in on a conversation between some folks
including Karen Coyle and Rob Styles (Talis) about this subject. It
definitely seems like there are lots of ways of doing the matching
[1]. At the end of the day, after you've done the painful part of
figuring out the linkages, you need a way of expressing them. URLs for
the resources being described, linked together w/ owl:sameAs or what
have you seems like a good, explicit way to start. And there's a
growing community of people doing the same thing in other pockets of
the web. [2]
> One possible solution would be to implement corresponding WebServices
> that get a ISBN or Bibkey and then deliver the appropriate LCSH or any
> other kind of related meta data. In my opinion a much better solution
> would be to offer the mapping data as a feed of flat files like
> LibraryThing (http://www.librarything.com/feeds/), e.g.
>
> ...
> <mapping>
> <isbn>123<isbn>
> <lcsh>aaa</lcsh>
> <lcsh>bbb</lcsh>
> </mapping>
> <mapping>
> <isbn>456<isbn>
> <isbn>678<isbn>
> <bibkey>ea24ad</bibkey>
> <lcsh>ccc</lcsh>
> <lcsh>bbb</lcsh>
> <ddc>345.223.444</ddc>
> <bk>12.34</bk>
> <osc>xyz</osc>
> <rvk>yyy</rvk>
> </mapping>
Sure, this is a service-oriented approach--personally I am more
interested in resource-oriented approaches. It's early days still,
but imagine if bibliographic resources were identified with URLs, and
could be resolved, and human/machine readable representations be
retrieved? This is basically the vision of linked-data.
Apart from the lcsh.info experiment, the LCCN service [3] is a step in
this direction. The simple thing that this service does, and which
other next-generation catalogs are doing, is assigning clean,
hopefully application independent URLs to their bibliographic
resources. The only thing they have to do is respond meaningfully with
machine readable data at those same URIs and we've got the beginnings
of linked data. This is why I found Martin's experiment so
interesting.
> Is LoC perhaps already thinking in this direction?
I'm not, but LoC is a big place, so maybe?
//Ed
[1] http://www.kcoyle.net/temp/merge.html
[2] http://esw.w3.org/topic/SweoIG/TaskForces/CommunityProjects/LinkingOpenData
[3] http://lccn.loc.gov
Received on Mon Sep 01 2008 - 13:37:58 EDT