Spam increase with number of files ?

[David Relson]

On Thu, 4 May 2006 09:50:58 +0200
Christophe Journel wrote:

> Hello
> 
> I am using the last version of bogofilter and i have a problem.
> 
> Indeed, when i send a mail with 5 attached files, it's tagged spam
> ( 90%)
> 
> and when there are 10 attached files, the spam rate is about 100% !!
> 
> I need some help
> 
> Thx

H'lo Christophe,

First, may I suggest you subscribe to the mailing list?  As list admin,
I only check once a day for messages from non-subscribers.  As a
subscriber your message posts immediately, which results in getting an
answer much sooner.

Unfortunately, your message lacks much info.  Bogofilter knows about
multi-part messages and will score each part (although it ignores
attachments of type image, audio, and video).  Bogofilter has no checks
for number of attachments.  Bogofilter just looks at the content of the
message and of the attached files.  So, there must be something about
the files that causes the spammish (high) score.

Using option "-vv" you can get a histogram and see how many tokens are
hammish and spammish.  Using "-vvv" bogofilter will show you the exact
score of each token in the message.  With "-vvv" bogofilter also sorts
the tokens from lowest scoring to highest scoring.  Try commands like
the following (which generate the v* results):

   bogofilter -vv  < msg  > msg.vv
   bogofilter -vvv < msg  > msg.vvv

If you can't learn enough from the above commands, gzip your 5 and 10
attachment messages and email them directory to me.   I'll score them
with _my_ wordlist and let you know if I see anything weird happening.

Regards,

David