I know I'm coming late to this discussion, but here are a few thoughts:
I also would LOVE to have what Emily asks for: given an item, show me
co-located items ... virtually. Improve on the physical world
limitations, if we can. Present the information well: familiar /
easy to learn / usefully. I'm not in love with call numbers, per
se ... but the "colocation by subject" is useful. Bookstores don't
use call numbers, but they still shelve books by subject, and then by
author within subject, generally.
I did some thinking about this 5 years ago: both the meaning of call
numbers and how to present things visually. I don't pretend to have
any answers, but I think I might have touched on a few of the key
questions:
http://infoviz.comm.nsdl.org/cgi-bin/wiki.pl?VizChallenges
Why Bibliographic Metadata poses Visualization Challenges
* there are large quantities of textual information
* it is hard to represent all the information in a full bookcase
onto a single monitor screen.
o bookspine info
o organizational info (binned and linear sorts from
classifications ...)
* single items can be associated with multiple parts of a
hierarchy.
* tree hierarchies can have vastly uneven leaf levels
* trees can be sparsely populated
o there may be incomplete or missing information to
associate metadata records with hierarchies
* controlled vocabulary problems
o hierarchies associated with bibliographic metadata values
more or less implicity refer to controlled vocabularies.
o Controlled vocabularies are often too limited or too
broad for the task at hand
1. The user is not broken. Our faculty are very vocal in desiring a
"virtual shelf list" that will allow them to, given a specific item,
look for "closely located" items. Call numbers have facilitated co-
location of (some) related physical materials, which facilitates a
browsing experience that users enjoy. Maybe it's nostalgia, maybe
it's something else ... but they enjoy it and find it useful. They are
used to call numbers, and by god, they want call numbers. Who are we
to naysay?
This has been implemented already in some systems. My local public
library catalog has this, but Stanford does not:
go to: http://plsiii.plsinfo.org
do a search
select a record
click on the call number link
It's not beautiful, but it's there ... more than I can say about our
ILS or about our OSS discovery interface.
2. Call numbers must be unique for circulation purposes. For a
virtual shelf list, we do not need to display multiple copies on the
shelf. http://infoviz.comm.nsdl.org/cgi-bin/wiki.pl?CallNumbers
Call numbers
3. Giving the user a new paradigm is a good thing, but it may not
alleviate the need for the familiar. Plus, the new paradigm has to be
sufficiently useful / familiar / easy to learn. Have you seen
Aquabrowser? The graph thingy on the left is neat looking ... but is
it used? Perhaps it is - I don't know.
I like the concept of a coverflow ... but how much can you see at a
glance? Do users want to see more at a glance, or want to see
covers? I have no idea - let's get a usability expert out there with
some prototypes (thanks for the prototypes everyone!)
Let's not throw the baby out with the bath water: information on book
spines has had decades to evolve into what is considered useful. In a
bookstore. In a home library. Do we "know better"? Let's
experiment knowledgeably, get user feedback and iterate!
4. call numbers use classifications that define a (subject) hierarchy
that is not a strict tree structure. There has always been a tension
between the linear shelving and the hierarchy of classification
schemes. http://infoviz.comm.nsdl.org/cgi-bin/wiki.pl?ClassVizReq
Can we resolve this successfully with a virtual browse?
5. items often fit more than one classification (more than one
subject). There is no reason why we can't have MULTIPLE linear
orderings, based on various subjects / fictitious call numbers. I'd
love to get away from linear, but I haven't yet found a replacement
that is familiar enough and easy enough to learn. See http://infoviz.comm.nsdl.org/cgi-bin/wiki.pl?VizChallenges
.
6. can we combine subject searching (text searches, even) with these
virtual shelves? Imagine a treemap of resources, and as you type your
search strings, matching areas of the tree map are highlighted. As it
gets down to a single area, the UI zooms in ...
For the record, we have at least THREE distinct call number systems
here: LC, Dewey, and SUDOC. It's ugly ... and it's our reality.
What should be combined? How should the software facilitate this?
Count me in for working on a solution.
- Naomi
On Oct 1, 2008, at 9:16 AM, Stephens, Owen wrote:
I agree with this in general - and this was my point about the
'Coverflow' in iTunes, that it allows a variety of sorting methods -
although it is still limited.
I think there are perhaps some other factors as well. Shelf-browsing
allows users to wander into 'their' part of the library and look at
stuff - but I don't think most OPACs have the equivalent. With a
bookstore (physically and virtually) we might see genre sections we
can browse. This might also work for public libraries? In research
libraries we tend to just present the classification without further
glossing I think - perhaps this is something we ought to consider
online?
The other thing that occurs to me about browsing by class mark is
that it presents a 'spectrum' view of a kind. This could be easily
lost in the type of 'search and sort' system you suggest (although I
still think this is a good idea btw). At the same time I'm a bit
reluctant to stop at providing a classification browse, as it seems
inherently limited.
I agree with the point about browsing the shelves and exploring the
material in more depth are related - which suggests integration with
other content-rich services are needed (Google Books, e-books, other
providers)
Owen Stephens
Assistant Director: eStrategy and Information Resources
Central Library
Imperial College London
South Kensington Campus
London
SW7 2AZ
t: +44 (0)20 7594 8829
e: [EMAIL PROTECTED]
-----Original Message-----
From: Code for Libraries [mailto:[EMAIL PROTECTED] On
Behalf Of
Keith Jenkins
Sent: 01 October 2008 13:22
To: CODE4LIB@LISTSERV.ND.EDU
Subject: Re: [CODE4LIB] creating call number browse
I think that one advantage of browsing a physical shelf is that the
shelf is linear, so it's very easy to methodically browse from the
left end of the shelf to the right, and have a sense that you haven't
accidentally missed anything. (Ignore, for the moment, all the books
that happen to be checked out and not on the shelf...)
Online, linearity is no longer a constraint, which is a very good
thing, but it does have some drawbacks as well. There is usually no
clear way to follow a series of "more like this" links and get a
sense
that you have seen all the books that the library has on a given
subject. Yes, you might get lucky and discover some great things,
but
it usually involves a lot of aimless wandering, coming back to the
same highly-related items again and again, while missing some
slightly-more-distantly-related items.
Ideally, the user should be able to run a query, retrieve a set of
items, sort them however he wants (by author, date, call number, some
kind of dynamic clustering algorithm, whatever), and be able to
methodically browse from one end of that sort order to the other
without any fear of missing something.
Keith
On Tue, Sep 30, 2008 at 6:08 PM, Stephens, Owen
<[EMAIL PROTECTED]> wrote:
I think we need to understand the
way people use browse to navigate resources if we are to
successfully
bring
the concept of collection browsing to our navigation tools. David
suggests
that we should think of a shelf browse as a type of 'show me more
like this'
which is definitely one reason to browse - but is it the only
reason?
Naomi Dushay
[EMAIL PROTECTED]