[dlpdev] fine-grained dlp subject search: possible direction?

[Douglas Carnall]

At Thu, 8 Apr 2004 13:14:23 +0000 evol@c6.org wrote:

> i would be glad 2 read yr thoughts however draft-ish

OK, well this is as far as I got:

At Sun, 28 Mar 2004 17:40:21 +0100 deadpoets_uk@yahoo.com wrote:

> i'm asim and i help at the antisystemic library at
>LARC, where we don't use subject categories. we log
>all our books as OTHER then we make reading lists/
>booklists of publishers. see 
>http://dlp.theps.net/userinfo.php?user=LARC

I found the add new booklists feature but it seems to require reentering
book data which seems a chore without reward. Indeed I got an error when
I experimentally tried to construct a short list for Unix beginners and
gave up. Would some sort of checklisty feature for making sublists of
books once they are on the system help? A "makeBookList" function
available from search results pages? Each booklist could have an
associated author and textual 

>we hope to develop the dlp so that it is founded on a
>semantic mapping of the libraries,

I am not quite sure what you mean by this. The libraries are currently
represented in the dlp as the collections of individuals or groups. The
dlp software records the stated interests of the person/people running
the library. Is it these interests that are/will be semantically mapped?
As they are entered in free text presumably any dlp node maintainer
(=librarian?) can state anything as an interest. Is the notion that
search would then find an individual with that interest, then browse
their collection to find a particular book on a subject? This has the
problem that different nodes could classify their interests differently
(I think I am interested in "software" for example but don't find node
owners who are interested in "computers" or "programming" unless I also
search for these key words). It would be cool if entering a book on the
system automatically updated the owner's interests.

It seems to me that classifying libraries by owners' interests could have
problems of scale. An example might help: every NHS GP practice that
trains GPs must have a practice library. Each library could be a node in
the dlp. There are probably 30 or so such practices in East London. All
of the libraries' owners will be interested in general practice, and
likely will hold certain books and journals in common. But as a searcher
I am probably more interested in the differences between the libraries:
i.e. the books and journals that they have that I don't. It would not be
workable to wade through 30 individual collections to determine this; I'd
rather use a fine-grained subject index and find the 2 or 3 libraries
that share that particular sub-interest and hold books about it.

Could say more, but when? ;-)

D.


--
Douglas Carnall
+44 (0) 7900 212 881
http://navarino.net:8080/blog
there are DNS problems with carnall.org
please see http://navarino.net for further details