Bogofilter seems to not be working

[David Relson]

At 02:49 PM 3/25/03, daniel wrote:

>I have set up bogofilter with the procmail recipies in the man page:
>
>:0fw
>| bogofilter -u -e -p
>
>:0e
>{ EXITCODE=75 HOST }
>
>:0:
>* ^X-Bogosity: Yes, tests=bogofilter
>spam-bogofilter
>
>
>Initially I was getting spamicity readings such as 0.415000, 0.000002, 
>spamicity=0.081142.  Now however all my new mail has a spamicity of 
>0.000000 despite the fact that I think some of it is blatant spam.  I have
>  bogoutil -w ~/.bogofilter .MSG_COUNT
>                        spam   good
>.MSG_COUNT              180    442
>
>so far and bogofilter still has not filtered anything into 
>spam-bogofilter.  I am wondering if it is working correctly or if it is 
>failing and if there is something I can do to check?

Hello Daniel,

There are several things you can do to check.  First, run bogofilter from 
the command line with differing numbers of "-v" (verbose) flags to learn more:

bogofilter -v < message --- will print the "X-Bogosity line for the message"
bogofilter -vv < message --- will print a histogram showing token counts 
vs. spam scores.
bogofilter -vvv < messagee --- will list _all_ the tokens of the message 
and their spam scores.

For words that you think are highly spammish, use bogoutil to display their 
spam/good counts and spam scores, i.e.:

bogoutil -w ~/.bogofilter word1 word2 word3

The value "0.415000" is the score of a token not in either the ham or spam 
wordlists.  When it appears as the score of a message, it indicates that 
bogofilter isn't finding the message's tokens in the wordlists.  Use 
"bogofilter -qv" to see what parameters bogofilter is working with and use 
"bogofilter -x d -v </dev/null" to print some debug info (which will 
include the names of the wordlists).

Hope this helps ...

David