Skip to main content.
home | support | download

Back to List Archive

Re: Rogue stop word, "and"

From: Bill Moseley <moseley(at)not-real.hank.org>
Date: Sun Feb 06 2000 - 18:40:00 GMT
>Bill Moseley wrote:
>> "And" is a boolean term and part of the search language.

At 09:49 AM 02/06/00 -0800, Edwin Whitelaw wrote:
>OK, I'll buy that.
>
>All of the boolean flags are also listed as stop words.   What's the point
>of listing them if they don't return the stop word message, "word too
>common, etc..."?

Swish was written by humans.  Why list "(" and ")" in the wordcharacters
setting?  They are also part of the search language.  Why does swish split
words using the wordcharactes settings on indexing, but not while
searching?  Why does swish stem already stemmed words in its index before
doing a wild card search resulting in missed hits?


>I'm further puzzled that a search on "not" returns a hit
>even though it's in the stop list. (I recognize it's not a boolean, but it
>is a rule word.)

Bad program logic.  Searching for "not" probably really searches for
everything that doesn't match nothing, which is everything.  So it works
perfectly ;)


>>From the user's perspective though, they have tried to search on a trivial
>word so I'd like to return the same message as if they'd searched for
>"the", for instance.

Two options, I'd guess.  Either filter on the front end, or hack the source.

>I'm just puzzled a bit by the choice of logic here.

Luckily, Swish is rather easy to change.  That's the nice thing about open
source.


Bill Moseley
mailto:moseley@hank.org
Received on Sun Feb 6 13:46:46 2000