Article:
  Bayesian Filtering with bogofilter and Sylpheed Claws
Subject:   Great!
Date:   2003-07-04 04:27:41
From:   anonymous2
I just spent an hour learning this article and got sylpheed-claws right. And as homework adapted the script to Evolution (mbox file format):



#!/bin/sh
# http://linux.oreillynet.com/pub/a/linux/synd/2003/01/30/bogofilter.html

BOGOFILTER="/usr/bin/bogofilter";
EVOGOODDIR="/home/lagc/evolution/local/SPAM";
EVOSPAMDIR="/home/lagc/evolution/local/SPAMFREE";

cd $EVOSPAMDIR;
echo Spam:
echo Processing Mail Box;
bogofilter -M -s -v < mbox;

cd $EVOGOODDIR;

echo NonSpam:
echo Processing Mail Box;
bogofilter -M -n -v < mbox;



Just a few tests and man pages consults and it worked. In evolution just set a filter with "pipe message to a shell command" ang set "/usr/bin/bogofilter" and select to move to the SPAM folder.


Two things more:


1. Remember to set the filter high (up, up) to be the first filter applied to the messages (both on sylpheed-claws and evo)


2. Could it be good if the bogolearn script remove de messages after pass them to bogofilter? Or should it have a learn session only when some false positive or span confuses bogo?


Thanks,


lagc