Jimmie,
I believe I'm one of those programmers...unfortunately the sad fact is
that there is little or no funding or support for innovative library
catalogue development.
For example, I still can not get a library to give me data to develop my
application mostly because the local subject authority files are
commercially protected by LC (and other sources)...meanwhile Amazon
offers a free Web service to their whole collection organized by
subject...what's a pragmatic developer to do?
As well, my committee was much more enthusiastic when I suggested I
could develop an innovative search interface for Web based information
(i.e., Amazon). The current state of affairs is that funding agencies
see future in the Web (semantic Web, Web 2.0, 3.0, etc) and library
applications, as a representative of pre-Web information storage and
retrieval, receive little attention.
Trust me I wanted to work on library data...I know subject headings are
underutilized and not understood by end users...but there is just no
incentive to do so...except being a contrarian.
I sure hope there's a way out of this impasse for libraries...it just
hasn't presented itself to me.
Charles-Antoine Julien
Ph.D Candidate
School of Information Studies
McGill University
-----Original Message-----
From: Next generation catalogs for libraries
[mailto:NGC4LIB_at_listserv.nd.edu] On Behalf Of Jimmie Lundgren
Sent: February 11, 2008 6:51 AM
To: NGC4LIB_at_listserv.nd.edu
Subject: Re: [NGC4LIB] Browsing percentages / analytics
Good Morning,
In reference to the statement below: " Thanks to
the screwball rules for constructing precoordinated headings, you can
expect around half (literally) to be unique even in very large
catalogs which limits their usefulness for collocation purposes," I
would like to point out that in the superior browse retrieval that was
available in the old NOTIS catalog results of subject searches were not
hidden behind their headings but able to be seen along with them. Thus
even though a particular string might be used only once the record could
be seen together with those records before and after it with similar but
not exactly the same strings.
For example, "Corn--Diseases and pests--Algeria" might have only one
record associated with it in my catalog. In the NOTIS catalog one was
able to see multiple headings and records simultaneously to easily also
see those titles that might have the subject "Corn--Diseases and
pests--Albania" and consider if they might also be useful. More "modern"
OPACS force one to separately open each heading to view the lists of
records associated with them.
This is also a hindrance in considering authors who may have some works
listed with their full names and others only with their middle initials.
The superior catalog would permit viewing of multiple author headings
together with the titles of their works and cross-references to
preferred forms. This will become especially critical if catalogs
increasingly blend article records that lack authority control into the
mix of records.
My ideal next generation catalog would not only offer users the fabulous
faceting such as my catalog now has with Endeca, but also a transparent
and user-friendly browse of headings with associated records and
cross-references. I've just got to believe there is a programmer out
there smart enough to understand the value of what I'm saying and make
that happen.
Thanks and best regards,
Jimmie
-----Original Message-----
From: Next generation catalogs for libraries
[mailto:NGC4LIB_at_LISTSERV.ND.EDU] On Behalf Of Kyle Banerjee
Sent: Saturday, February 09, 2008 10:40 AM
To: NGC4LIB_at_LISTSERV.ND.EDU
Subject: Re: [NGC4LIB] Browsing percentages / analytics
On Feb 7, 2008 3:55 PM, Bennett Ponsford <BPonsfor_at_lib-gw.tamu.edu>
wrote:
> One of the (many) things that surprised me about our results was how
seldom our library faculty/staff used the author headings browse (2.70%)
or the subject headings browse (1.05%). Particularly given their
preference for a nice, traditional, left-anchored title search (35.04%)
over title keyword (6.06%).
>
The title is usually the name a work is known by, and calling for
things by name rather than associated attributes (such as author)
makes perfect sense. Subject headings browse is for geeks. Thanks to
the screwball rules for constructing precoordinated headings, you can
expect around half (literally) to be unique even in very large
catalogs which limits their usefulness for collocation purposes.
Subject heading searches in most catalogs remind me of constant
examples in the computer world where we try to mold human behavior
around the needs of our systems rather than trying to figure out how
to work with what people do intuitively (e.g. requiring dates
expressed YYYYMMDD, prohibiting certain common characters for many
operations, Byzantine password regulations, etc)
On Feb 8, 2008 8:32 AM, Bennett Ponsford <BPonsfor_at_lib-gw.tamu.edu>
wrote:
> Thanks for the info, Karen. One of the things I also want to look at
is response to no hits messages at small libraries. My experience -
entirely anecdotal - at my previous place of employment is that a lot of
the students, especially grad students and faculty, didn't expect us to
own much and so accepted a no hits message at face value.
>
The problem isn't just with no hits. It's also about what people do
when they do get results.
Even in our own literature, it is often obvious which 2 or 3 keyword
searches where used to locate resources for a "literature review."
Figuring out the best way to let users know of related works is not a
straightforward problem. Logfile analysis can be helpful as can
actually observing users, but if we don't have the motivation of the
search to give us context, there is an awful lot of wiggle room for
interpretation.
kyle
Received on Mon Feb 11 2008 - 13:19:17 EST