email containing token with high spamcount only gets an unsure

[David Relson]

On Tue, 18 Jul 2006 14:07:46 +0200
Gerrit Thede wrote:

> Hi,
> 
> I am receiving annoying spam emails with differing content, but
> always with the same last lines that go like this:

...[snip]...

Hi Gerrit,

You can learn more about how bogofilter is scoring these messages by
using the -vv option (to generate a histogram) or by using -vvv to show
each token's spamicity.  For example, using my wordlist, the histogram
for your message is:

relson# bogofilter -C -d /var/spool/bogofilter -vv < msg.7443
X-Bogosity: Ham, tests=bogofilter, spamicity=0.000000, version=1.0.2
 int cnt   prob  spamicity histogram
0.00 56 0.022153 0.020743 ###################################
0.10  2 0.122604 0.024829 ##
0.20  0 0.000000 0.024829 
0.30  0 0.000000 0.024829 
0.40  0 0.000000 0.024829 
0.50  0 0.000000 0.024829 
0.60  0 0.000000 0.024829 
0.70  0 0.000000 0.024829 
0.80  0 0.000000 0.024829 
0.90  2 0.991605 0.145378 ##

The FAQ has more info on the -vv and -vvv options.

It's possible that the headers of the messages have hammish tokens and
they are counter-balancing the spammish tokens.  "-vvv" will show if
that is so.  Assuming this is the case, you _could_ create an ignore
database to tell bogofilter to ignore certain tokens when scoring
messages.

Regards,

David