
Training your mailbox about SPAM
Spam is the bain of our lives and there’s no simple solution to a massive problem.
One tool in the toolbox is what’s known as a bayes filter. In an oversimplification, Bayes works by you training the filter with ham (not spam mail) and spam. The Bayes filter takes statistical views of these emails and then uses these statistics to determine if future mail is spam or ham.
How do I use Bayes with my mailbox?
To use bayes with your mailbox on a WebCP server, you’ll need to set up your email box as an IMAP mailbox. When you’ve created your IMAP account you’ll notice a Spam folder under the inbox.
To train your spam filter simply move your spam emails into the spam folder, rather than deleting them.
That’s it!
I’ve put spam email in my Spam folder but I’m still getting spam
Firstly, spam filtering is not a science, its an art. As such no spam technique will work 100% of the time. Another very important factor to note with bayes is that it will not do any statistical analysis until you have at least 200 emails in the ham folder (your inbox) and 200 spam emails in your spam folder, so initially spam will not be learnt by bayes until you have quite a lot of spam emails.
How do I speed up the time to bayes training?
You’ve just learnt that bayes doesn’t work until you have 200 spams. So how do you get 200 spams? Well, in WebCP you have a setting for spam thresholds. This link is under “Emails->Spam Assassin”. Each email that comes into your mailbox has a percentage score. The higher the percentage score the higher the chances are that its spam.
In the Spam Assassin settings, set your spam threshold to 100 so that almost no spam is blocked. I know, I’m recommending that you actively turn off spam filtering and get spammed… but, its just for a little while. Once you have 200 spam emails in the Spam folder you can set the spam threshold back down.
I made a mistake classifying an email as spam with bayes
That’s ok, just move it back to your inbox and spam assassin will relearn it as ham.. No biggie!
Comments