There are a number of questions that come up in the context of Subject
Access in a Catalog, so I'll attempt to break up my arguments into
(hopefully) digestible chunks. Take it all with a grain of salt -- I'm
not a cataloger, and there's a lot of relevant literature that hasn't
even made it into the piles of paper on my desk, much less been read by
me.
*********************
One of the most frequently heard comments about using LCSH (or any other
controlled ontology) is "Why bother? Nobody searches with subject
headings anyway"
And this is actually pretty much this case. Aside from professional
librarians, hardly anybody knows what those big red books are for.
Subject tracings in OPAC record displays aren't followed that often, and
absolutely *no* end users have yet figured how to do a real subject
search in our current systems. However, I would argue that all this is
due less to any inherent deficiencies of LCSH than that of our current
crop of ILSs. Indeed, the OPAC doesn't "suck"; that is much too gentle
a term. The OPAC is mind-bendingly awful -- crippled in ways that we're
just beginning to discover, especially with regards to non-known-item
searches.
Consider PubMed, an elegant bibliographic database (paid for with our
tax dollars -- anybody know anything about the development team?) that
leverages an ontology (MeSH) which to the naked eye is just about as
hairy and non-user-friendly as LCSH.
Type the term "Heart Attack" into the (single) PubMed search box, and
you get back 116,507 results. Now, there's actually only 1,910 items in
the database that actually have the phrase "heart attack" in the record
(it's not really a clinical term). What PubMed does is quietly map that
to ("myocardial infarction"[TIAB] NOT Medline[SB]) OR "myocardial
infarction"[MeSH Terms] OR heart attack[Text Word]. This happens in the
background -- the default record display doesn't even *show* you the
MeSH headings. All quite helpful, and all quite impossible without a
bunch of people at the National Library of Medicine reading articles and
assigning them subject headings.
In the future, I would like to type "spanish recipes" into a Next
Generation Catalog and have my query quietly mapped in the background to
"Cookery, Spain". Obviously somebody has to build (or computationally
derive) those crosswalks, but there's no fundamental reason it can't be
done.
There's still a lot of low-hanging fruit to be found in terms of using
subject terms to guide retrieval. It would be a pathetic legacy indeed
if at the very historical moment we were starting to build tools that
actually *used* LCSH for something (Endeca implementations, experiments
with Solr, etc.) that we discarded it entirely.
Ed Sperr
Digital Services Consultant
NELINET, Inc.
153 Cordaville Rd. Suite 200 Southborough, MA
(508) 597-1931 | (800) 635-4638 x1931
Received on Wed May 23 2007 - 13:43:42 EDT