> My (erroneous) assumption was that if a record did not have a broader
term (i.e. a 550 $wg value) then it would sit at the top of the subject
tree, and that they would be the very general subjects headings. As I
found this obviously not the case.
-- You're corrrect - LCSH doesn't work like this. It's not a thesaurus.
> Looking at our authority records it appears that it is not possible
to build a thesaurus like structure which shows the relationship between
terms.
-- Is that because your authority records do not have 5XX $wg's in them?
Or because it is not a thesaurus?
All said, take a look at http://id.loc.gov/
ID has all of LCSH available as a bulk download in a couple of flavors
of RDF. In the data, the broader and narrower relationships are
expressed. You could grab one of those and work with the data locally.
Alternatively, you could use a combination of the label service [1] and
content-negotiation [2] to query for entries and then extract
broader/narrower relationships in real-time. For example, this label
search will match the lexical word Hurricanes to its resource at ID.
http://id.loc.gov/authorities/subjects/label/Hurricanes
Following it, you'll get the HTML, but add a little content-negotation
and you can get something more computer friendly:
curl -L -H "Accept: application/rdf+xml"
http://id.loc.gov/authorities/subjects/label/Hurricanes
( There is a little short cut to the same - append ".rdf" to the end of
the label request, like so
"http://id.loc.gov/authorities/subjects/label/Hurricanes.rdf";, -
allowing you to sidestep the content-negotiation altogether. Also works
for ".json" and ".nt")
In any event, you'll see the broader/narrowers in the output.
Hope this helps,
Kevin
[1] http://id.loc.gov/techcenter/searching.html#knownLabel
[2] http://id.loc.gov/techcenter/
On 10/29/2012 10:04 AM, Stephen Graham wrote:
Hi All - this is a new area for me, so it might be that I'm barking up the
wrong tree or missing something fundamental. We are trying to build a browsable
subject list in our discovery service (built on the Endeca technology). The
idea would be to expand terms and reveal the facets below which in turn are
expandable - a subject tree which our users can browse the narrowr/broader
terms.
We use LCSH and my initial thought was to use the authority records from our
catalogue (Symphony) to determine the relationship between terms. Dumping the
authority records out I thought that I could use the 550 field - in particular
the $w subfield and /0 (special relationship ). My (erroneous) assumption was
that if a record did not have a broader term (i.e. a 550 $wg value) then it
would sit at the top of the subject tree, and that they would be the very
general subjects headings. As I found this obviously not the case. Looking at
our authority records it appears that it is not possible to build a thesaurus
like structure which shows the relationship between terms.
I know The Library of Congress offer "Classification Web"
(http://www.loc.gov/cds/classweb/). This might be overkill for what we need. Is there an
open API which may help achieve what we want?
Stephen
