[Sigia-l] Y!Q - Just-in-time search

[James Aylett]

sigia-l-bounces at asis.org wrote:

>> What I'd prefer to do is to segment the result set from Google
>> clearly, so that users could decide how to further refine their
>> query once they'd got some sample results.
>
> Google gives you 'Search within results' already (not great UI). But
> how would you segment the result set? Y!Q shows the search parameters
> clearly, with checkboxes for each. If you had, say, six such search words,
> it'd be very confusing to have all the permutations as segmented results.
I'm
> assuming all these efforts are for people who don't grok Boolean
> search to begin with. Could this ever be anything more than an advanced
users'
> tool? I'd love to see Yahoo and Google test results on this.

I guess I'm really thinking of this as totally separate to the idea of
adding/removing terms from the query (what Xapian, the IR system I'm
familiar with, calls 'top terms' for historical reasons :-). I was thinking
that some kind of topic map categorisation could go on, with results in the
top ten being grouped by their topics, so you can then say "Yes, show me
more results for my search 'current' like <this one>, which is about
electricity", ignoring the result about current affairs.

In theory it's possible to do this automatically and on the fly, although
it's computationally expensive, and obviously not as good as using metadata
(providing the metadata is fairly accurate). I have a hunch, but absolutely
no evidence to support it, that automatic on the fly categorisation will
turn out to be better than automatic pre-classification (because it will
implicitly take into account the search context). It could involve running
the search again once for each result in the top ten, though.

Cheers,
James

--
James Aylett
  Chief Technical Architect, Tangozebra
  t 020 7535 9850 f 020 7535 9900
  w http://tangozebra.com/


This e-mail message, including any attachments, is intended only for
the person or entity to which it is addressed, and may contain
confidential information. If you are not the intended recipient, any
review, retransmission, disclosure, copying, modification or other use
of this e-mail message or attachments is strictly forbidden.

Copyright Tangozebra 2005. All Rights Reserved.


---
Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.854 / Virus Database: 582 - Release Date: 03/02/2005


________________________________________________________________________
This e-mail has been scanned for all viruses by Star. The
service is powered by MessageLabs. For more information on a proactive
anti-virus service working around the clock, around the globe, visit:
http://www.star.net.uk
________________________________________________________________________