> > From your todo list I really like 3 - list all words in an index
>> Would it be possible to postprocess an index file to get a
>> crossreference kind of list with "What's related?".
>Can you give me an output example?
>Some one asked for a -k option like:
>swish-e -k g -f indexfile
>This will output all words starting with 'g'
>But I think that you would like more info...
Some thoughts on "what's related":
I expect users not being knowledgable about the "problem domain" they
might know a keyword to start with. With dictionary they are shown
lists of words in the index. The really cool function I'm thinking
about, but don't know how to implement yet is that given a word with
a click you get other words that are correlated with that word. (now
they just search from dictionary)
Based on preliminary tests i can say that
1: very common words are useless (cliche)
2: very rare words are of less value (too rare)
3: the middle (frequent) group is interesting
we would need a file with all words cross-tabulated against eachother
with a value in the cell with some measure of correlation. (Building
this file could be a lengthy process I presume.)
then something like this could hopefully be shown:
access: location, mod_access, mod_info, auth_dbm, password
Swish-E in principle indexes possible inputs for this process so only
the .swish file needs to be read.
Perhaps it's best to have the crosstab file in a file with a
mechanism for editting manually like /etc/aliases?
-- /''' Bas Meijer mailto:email@example.com
c-OO http://antraciet.com Web Services
\ > Kerkstraat 19 Postbus 256 1400 AG Bussum
\&& t. +31 35 7502100 f. +31 35 7502111
Received on Tue Oct 31 10:43:38 2000