this is obsolete doc -- see http://doc.nethence.com/ instead

Setting up SpamProbe bayesian spam filtering on FreeBSD 

Installation 

Make sure it is installed, 

  pkg_info | grep spamprobe 

Configuration 

Create a folder for SpamProbe to store it's Berkeley DB databases, 

  cd ~/ 

  mkdir -p .spamprobe/ 

Setup your procmail Local Delivery Agent so it calls SpamProbe to filter incomming messages, 

cd ~/ 

vi .procmailrc 

at some point at the beginning of the rule sets, add, 

# Invoke spamprobe 

:0 

SCORE=| spamprobe train 

:0 wf 

| formail -I "X-SpamProbe: $SCORE" 

:0 a: 

*^X-SpamProbe: SPAM 

.spam.spamprobe/ 

Ready to go 

Send a test email to the recipient while reading the procmail logs e.g., 

  cd ~/ 

  tail -F .procmailrc.log.2013 

Learn, unlearn and long-term maintenance 

Prepare a script to be executed nightly by the cron daemon, 

mkdir -p ~/bin/ 

cd ~/bin/ 

cat > spamprobe.sh <EOF9 

#!/bin/sh 

IMAPDIR=$HOME/Maildir 

spamprobe remove $IMAPDIR/.remove/ 

spamprobe good $IMAPDIR/.nonspam/ 

spamprobe spam $IMAPDIR/.spam/ 

spamprobe train-spam $IMAPDIR/.spamprobe/ 

EOF9 

It's also good to proceed with some clean-ups of words that haven't been seen as spam in the last two weeks (first option, commented out) or even six monthes (hereby enabled). 

Configure the cron tab for the mail user, 

crontab -e 

PATH=/sbin:/sbin:/usr/sbin:/usr/bin:/usr/local/sbin:/usr/local/bin:$HOME/bin 

02 00 * * * spamprobe.sh 

#03 00 * * * spamprobe cleanup 2 14 

03 00 * * * spamprobe cleanup 1000 180 2 14 

References 

Welcome to SpamProbe - A Fast Bayesian Spam Filter: http://spamprobe.sourceforge.net/ 

spamprobe(1) 

SpamProbe v1.4 README: http://spamprobe.sourceforge.net/README.txt