Bayespam is a qmail spam filter written in Perl, using Bayesian
classification to filter out unsolicited commercial email. It is
written with ease of installation and use in mind, and it is
encouraged that you give it a try. Bayespam actually learns as you
give it more spam to process, so it should become better and better
the longer you use it.

Bayespam is based on a paper written by Paul Graham called A Plan
for Spam. In this paper, Mr. Graham talked about a spam filter he
is working on that used Bayesian classification to determine if a
particular piece of email is spam or not.