Quoting Jonathan Rochkind <rochkind_at_JHU.EDU>:
> I think "salvage operation" accurately describes much of what we need
> to do to make our metadata serve the functions that it needs to serve
> in the contemporary environment. That's a good phrase. But that's
> where we are.
>
I prefer to look for utility in the *concepts* expressed in LCSH
rather than the prefLabels that have been selected by the library
community. "France -- History" is a string that represents a concept.
That concept could be expressed as "History of France", "France,
History", "French History", or any other number of strings in a
variety of languages. What LCSH does give us is an identifier for that
concept. So, for example, BISAC has HISTORY / Europe / France, with
the BISAC code: HIS013000. This might be considered equivalent to
http://id.loc.gov/authorities/sh85051256.
Using Jonathan's examples:
?Cookery, Indic?
?Absurd (Philosophy) in literature?
?Teachers of gifted children?
All of these seem to me to be perfectly good concepts, even if not
easily "facetable". That doesn't mean that a faceted form of the same
concept couldn't be developed that is equivalent and could be linked
to the LCSH heading. The question is whether the concept itself is
useful for linking.
The purpose that I see to linking to LCSH headings is to be able to
link library bibliographic records to topics, persons, places, events,
in the linked data world. It *isn't* about using LCSH's preferred
terminology. I would think that we would want to connect LCSH to
concepts in Wikipedia and DBpedia, for example.
The potential is that we could add any number of "altLabels" to the
identified concepts, including labels in languages other than English,
and create a more usable set of identified concepts, all pointing to
the same identifier. That would, however, require that LCSH allow for
user input, etc. Since LCSH must adhere to the LC standard, and not
veer off into untrammeled semantic exploration, we might want to
consider doing so in another environment, perhaps Ross's version. As
long as we tie everything back to the LC identifier, we should be able
to provide the needed stability of reference.
kc
--
Karen Coyle
kcoyle@kcoyle.net http://kcoyle.net
ph: 1-510-540-7596
m: 1-510-435-8234
skype: kcoylenet
Received on Mon Nov 16 2009 - 12:04:22 EST