Autofiler is a server-side e-mail classifier based on Bayesian text analysis. In combination with an IMAP server, autofiler can file e-mail messages in the appropriate folder automatically.
Autofiler is self-learning and can handle an arbitrary large number of folders. Most importantly, autofiler is transparent to end-users.
The program is licensed under the terms of the GPL. It uses (customized versions of) algorithms described in A Statistical Approach to the Spam Problem and is implemented in the perl programming language. Autofiler further requires access to a correctly configured PostgreSQL database.
Autofiler is a Mail Delivery Agent (MDA) that understands the concept of folders. It is meant to be used with a mail server such as Exim or Postfix on a UNIX or Linux platform. Upon arrival of a new message the mail server executes the af_deliver_message program which examines the folders in the users mail box and determines the appropriate folder for the message. It then places the message in this folder.
Mail folders are analyzed by means of a seperate program, af_build_database, which looks through the user's mail folders and builds a token database. This program should be run periodically to keep the database up to date.
The author uses the Maildir format to store mail folders on disk but theoretically the program can work with any format that the perl Mail::Box module supports and that recognizes the notion of folders.
Documentation is included with the release and is available online as well.
You can reach the author by sending an e-mail to: All e-mail correspondence will be read and answered if possible. This is however not a commercial-grade support service for the software.