LCSH subdivision semantics (was Re: Tim Berners-Lee on the Semantic Web)

From: Simon Spero <ses_at_nyob>
Date: Wed, 4 Nov 2009 19:54:05 -0500
To: NGC4LIB_at_LISTSERV.ND.EDU
 Tue, Nov 3, 2009 at 1:24 PM, Karen Coyle <lists_at_kcoyle.net> wrote:

>
> Order is complicated in LCSH (as is everything else ;-)). In a given
> heading, the order that the subfields are written in the MARC record is the
> order that must be retained. In some cases, changing the order actually
> changes the meaning of the heading. HOWEVER, the order does not give you the
> facet type. So in some faceted systems, you have a fixed number of facets
> and they are always in the same order:
>

[I had hoped to discuss this at last week's VoCamp at the Library of
Congress; unfortunately there wasn't much of a library presence]

It all depends on what you mean by "mean" :-)

As far as I know, the only  large study that covers the effect of
subdivision order on human understanding (covering real people, children,
and both kinds of librarians) is the University of Michigan project
documented in Drabenstott (1998).  The headings chosen for the study  were
chosen from subjects whose meaning considered especially likely to change in
meaning based on the order of subdivisions.

The "correct" meaning was determined by an expert (Dede) with over 25 years
of experience in LCSH cataloging and who was familiar with the study
protocol.  The "native speaker's" intuition was validated by comparison
against a second expert, with reported full inter-rater consistency.

Several headings were assigned multiple meanings by Dede.  If a user
supplied any of these correct  meanings listed it was scored as correct.

In some cases the expert gave different meanings for a heading based on
whether it was presented in isolation, in an alphabetical list, or in the
context of a bibliographic record.  This makes it hard to determine what
 the  "correct" meaning or meanings is or are.

The relevant part of the study are covered  in Drabenstott et al. (2000),
and contain a mixture of good and bad news.

The good news is that changes in subdivision order did not significantly
affect user understanding of the heading.

The bad news is that even librarians gave incorrect readings for   nearly
 half of the headings  on the surveys the completed.  This was true for both
technical service and reference  librarians.  If headings are misinterpreted
by the people assigning them, it doesn't matter what the "correct" meaning
is, since documents will be assigned the "wrong" heading, or the "correct"
heading will be assigned to the wrong documents.  If the documents to which
it is assigned provide the extensional semantics of a heading, it is this
"meaning" that the end user must discover in order to find the documents
they are looking for.

It is possible that automated natural language generation techniques may
outperform naive or even professional users.  This would help with display
related issues, but does not deal with existing records bearing incorrect
headings.

As Drabenstott (1998, pp. 175-176) notes:

Some readers might be tempted to review the findings about subject
heading understanding
in this report and conclude that the entire LCSH system ought to be thrown
out because library patrons understand less than half the subdivided subject
headings they encounter. To be honest, we researchers do not know how to
interpret this finding because there are no other studies of subject heading
understanding with which to compare.


In conclusion, *FRIN*.

Simon
-------
Drabenstott, Karen Markey (1998). Understanding subject headings in library
catalogs. Tech. rep. University of Michigan.
   URL: http://deepblue.lib.umich.edu/handle/2027.42/57992

Drabenstott, Karen Markey, Bonnie A. Roeber Dede, and Melanie Leavitt
(2000). “The Changes of Meaning in Subdivided Subject Headings”. In:
Cataloging & Classification Quarterly 28.3. P. 19. URL:
http://www.informaworld.com/10.1300/J104v28n03_03
Received on Wed Nov 04 2009 - 19:55:38 EST