Automated spam learning on Linux with SpamAssassin

First of all, create a new user account, or virtual mail account for your spamtrap. I suggest a mailbox name that has zero chance of receiving legitimate mail. I am also assuming that this mailbox will be located in /var/spool/mail/usernamehere. Finally, you will need a fairly new Spamassassin, since the older ones in many linux distros do not include the bayesian filter.

So next off, we have to create a cron job that will run the spam learning. Mine is here (and a little crude, but it works):

#!/bin/bash
if [[ `cat /var/spool/mail/algoholic | wc -c` -ge 10 ]] ; then
sa-learn –spam –showdots –mbox /var/spool/mail/algoholic
rm -f /var/spool/mail/algoholic

fi

Now that we have the spam learner ready, we can start placing links all over the place which point at our decoy user. If you want, you can add a style="display:none" parameter inside of the a href so that the link is invisible to people, but visible to computers.

That is all. It takes a couple of weeks for you to get fully indexed by the spam bots, although you can also hit some reply to unsubscribe links with this email address substituted for your real addy if you want.


This entry was posted on Tuesday, August 29th, 2006 at 9:36 am and is filed under Hacks. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.

One Response to “Automated spam learning on Linux with SpamAssassin”

  1. bsdwork Says:

    Honestly, I have to say that amazingly, this still works. Pretty cool, eh?

Leave a Reply