Yes, it's exploring what advantage there may be of using overlapping search
terms to help bridge the differences between LCSH and "common usage", or
"what the searcher is thinking of" that motivated this subject display.
For example, the person-in-the-street would reasonably think that when
searching a library catalogue where people have gone to the bother of
subject-classifying, then the results on searching "bay of pigs" would
return everything relevant, even if that string didn't appear in the title,
even if the full text wasn't being searched.
*http://ll01.nla.gov.au/search.jsp?searchTerm=bay+of+pigs
<http://ll01.nla.gov.au/search.jsp?searchTerm=bay+of+pigs>*
But LCSH organises that content under "Cuba -- History -- Invasion, 1961".
There's a high correlation/overlap between "bay of pigs" results and this
subject, which is why this subject string is highlighted in the prototype
results. But a search on that subject, removing "bay of pigs" as a search
term, returns, for example, from 1962 "The Cuban Invasion : the chronicle
of a disaster / by Karl E. Meyer and Tad Szulc ".
http://ll01.nla.gov.au/show.jsp?rid=000000643156
This isn't returned on a search for "bay of pigs" (on the prototype, or on
Trove). Maybe "bay of pigs" wasn't even "a thing" when this book was
catalogued, or, if it was, it was thought to be an ephemeral description.
On Trove (and I guess most library catalogues), by paying carefully
attention and tallying the subjects assigned to a "bay of pigs" results,
you may eventually realise a good search may be:
"Cuba Invasion 1961" OR "bay of pigs"
because there are LOTS of resources I'd want to know about if I were
searching for "bay of pigs" that don't have an assigned subject string
"Cuba -- History -- Invasion, 1961" for various reasons, such as a tendency
not to assign more subjects that you want cards in the catalogue! (eg
http://trove.nla.gov.au/work/13249747 )
I guess a better approach is to do this automatically for the searcher: to
note "bay of pigs" results have a high but not total correlation with
results assigned LCSH's "Cuba -- History -- Invasion, 1961", and I guess
that's one of the attractions of searching on Google: that we take this
type of "magic" for granted.
Kent Fitch
On Thu, Apr 14, 2016 at 11:19 AM, Harper, Cynthia <char...@vts.edu> wrote:
From a librarian’s perspective, we know searching is messy – a researcher
can’t hope to find the perfect subject heading that will reveal all their
related content in one term. Searching is exploring through overlapping
terms, and compiling a bibliography from the pearls found in the process.
This interface makes clearer what the related terms may be, given a borad
term like practical theology. And it’s so nice that it combines the
classification structure with the subject headings.
Cindy Harper
@vts.edu<http://vts.edu>
-----Original Message-----
From: Code for Libraries [mailto:CODE4LIB@LISTSERV.ND.EDU<mailto:
CODE4LIB@LISTSERV.ND.EDU>] On Behalf Of Kent Fitch
Sent: Wednesday, April 13, 2016 8:17 PM
To: CODE4LIB@LISTSERV.ND.EDU<mailto:CODE4LIB@LISTSERV.ND.EDU>
Subject: Re: [CODE4LIB] LCSH, Bisac, facets, hierarchy?
About ten years ago, I was wondering how to make the structure in LCSH,
or
at least how it was encoded in MARC subject tags more useful, so when
implementing a prototype for a new library catalogue at the National
Library of Australia, I tried using the subject tag contents to
represent a
hierarchy, then counted the number of hits against parts of that
hierarchy
for a given search and then represented the subject tags in a hierarchy
with hit counts. One of the motivations was to help expose to the
searcher how works relevant to their search may have been
LCSH-subject-catalogued.
I'm a programmer, not a UI person, so the formatting of theresults were
fairly primitive, but that prototype from ten years ago ("Library Labs")
is
still running.
For example, search results for /ancient egypt/
http://ll01.nla.gov.au/search.jsp?searchTerm=ancient+egypt&keywords=0.5&keywordWildcard=0.05&titlePhrase=12.0&authorPhrase=9.0&subjectPhrase=9.0&genrePhrase=9.0&titleWords=4.0&authorWords=3.0&subjectWords=3.0&genreWords=3.0&titleExact=18.0&authorExact=15.0
/computer art/
http://ll01.nla.gov.au/search.jsp?searchTerm=computer+art&keywords=0.5&keywordWildcard=0.05&titlePhrase=12.0&authorPhrase=9.0&subjectPhrase=9.0&genrePhrase=9.0&titleWords=4.0&authorWords=3.0&subjectWords=3.0&genreWords=3.0&titleExact=18.0&authorExact=15.0
/history of utah/
http://ll01.nla.gov.au/search.jsp?searchTerm=history+of+utah&keywords=0.5&keywordWildcard=0.05&titlePhrase=12.0&authorPhrase=9.0&subjectPhrase=9.0&genrePhrase=9.0&titleWords=4.0&authorWords=3.0&subjectWords=3.0&genreWords=3.0&titleExact=18.0&authorExact=15.0
This prototype also explored a subject hierarchy which had been of
interest to the NLA's Assistant Director-General, Dr Warwick Cathro, over
many years, the RLG "Conspectus" hierarchy, which I guess was not unlike
BISAC in its aims. It is shown further down the right-hand column.
Both the subject hierarchy and Conspectus were interesting, but neither
made it into the eventual production search system, Trove, implemented at
the NLA, in which subject faceting or hierarchy is absent from results
display:
http://trove.nla.gov.au/book/result?q=ancient+egypt
http://trove.nla.gov.au/book/result?q=computer+art
http://trove.nla.gov.au/book/result?q=history+of+utah
The "Library Labs" prototype is running on a small VM, so searching may
be
slow, and it hasnt been updated with any content since 2006.. But maybe
the way it attempted to provide subject grouping and encourage narrowing
of
search by LCSH or exploring using LCSH rather than the provided search
terms may trigger some more experiments.
Kent Fitch
On Thu, Apr 14, 2016 at 3:11 AM, Mark Watkins <m...@thehawaiiproject.com
<mailto:m...@thehawaiiproject.com>>
wrote:
<head starting to swim> :)
sounds like there is a lot of useful metadata but somewhat scattered
amongst various fields, depending on when the item was cataloged or
tagged.
Which seems to correspond to anecdotal surfing of the Harvard data.
I guess my new task is to build something that aggregates and
reconciles portions of LCSH, LCFGT, and GSAFD :).
Thanks for the additional perspective!
