Hiya,
On 6/21/06, Bernhard Eversberg <ev_at_biblio.tu-bs.de> wrote:
> What about browsable indexes for names, titles, subjects,
> keywords, call numbers? Even a merely "good enough" system
> should have these, but this is about "next generation".
We're using a number of indexes that pulls all sorts of interesting
stuff out of the MARC records, and do analysis and clustering on top,
but we're working on trying to make the right choice of indexes to
match a "better than average" system. There might be a given set of
indexes and clusters that works better for some than for others, and a
way to customise these things would be ideal.
> And related to that: Can the syndetic structure of authority control
> data be used to, say, make D. D. Shostakovich findable no matter
> which of his 27 spellings the user tosses in? [And beyond those,
> the name index would have to come in handy, of course].
Yes. :) That's the short answer. The longer answer is that we're
looking for the right set of indexes and clusters that would suit the
normal system, and what extensions to this is needed for more advanced
stuff. Kent just added authority records, but he told me the result
was less than great, but I'll leave it for some other on that team to
tell the details of it.
We've been toying with other means of "fixing" misspellings that
briefly dips its toe into latent semantic parsing, lexical parsing and
wordnet integration and whatsnot, but we're a poor lonely library a
long way from home, meaning we'll get right on it as soon as we can.
I'll fill you in on the details on how we're to share this stuff.
Alex
--
"Ultimately, all things are known because you want to believe you know."
- Frank Herbert
__ http://shelter.nu/ __________________________________________________
Received on Wed Jun 21 2006 - 07:21:43 EDT