Bogofilter Basics

I use bogofilter with procmail to counteract spam. There's a good web page here:

To use bogofilter effectively, you need two things:

You need to register all this email as spam or ham with bogofilter so it can build up its word lists. Then, when another email comes in, bogofilter consults its word lists to determine whether the new email is spam or ham.

If you tell it to, bogofilter will insert an appropriate header (X-Bogosity) with an indication of the chances that the email in question is spam. You can then use the third-party tools of your choosing to act on this information. I use procmail myself (See Procmail Integration).

Let's say you have all you junk mail in a maildir called Junk. You can then run the following in the directory where your Junk mail resides:

bogofilter -s -B Junk

The -s options means to register as spam. The -B options tells bogofilter to register the "objects" listed on the command line. In the above example, the object is a maildir box/folder. bogofilter knows enough about the maildir format to register the emails in the folder correctly.

For the rest of your legitimate emails, register them as ham. You'll need to provide all your folder/mailboxes on the command line:

bogofilter -n -B INBOX INBOX.mailing-list01 INBOX.mailing-list02

The -n tells bogofilter to register as ham.

If your only spam maildir is named Junk, and the rest of your maildirs are ham, you can do this:

$ cd ~/Mail # wherever your mail folder are
$ ls | grep -v Junk | xargs bogofilter -n -B

bogofilter stores its word lists under ~/.bogofilter. If you ever want to recompute your word lists, you can delete the contents of ~/.bogofilter and start again.

Procmail Integration

Bogofilter will insert an X-Bogosity header into the emails it classifies if you tell it to. My procmail setup takes advantage of this to dump spam into a special folder.

#### start spam conditions
| bogofilter -e -p --spam-cutoff=0.95


* ^X-Bogosity: Spam, tests=bogofilter
#### end spam conditions

The -p (passthrough) option to bogofilter tells it to tack on the X-Bogosity to the email.

The -e will cause bogofilter to exit with a 0 on success, instead of different statuses depending on the classification results. This simplifies procmail integration.

The --spam-cutoff option tells bogofilter that all emails with a spam score higher then this will be considered spam.


BogofilterBasics (last edited 2010-04-17 16:15:47 by DesmondRivet)