SpamBayes: Open-source antispam tool for Microsoft Outlook

If you don't have an Exchange server, only individual Microsoft Outlook desktop installations, the open-source antispam tool SpamBayes is a handy way to manage spam without dropping a lot of money.

For organizations that don't have an Exchange server, only individual Microsoft Outlook desktop installations, SpamBayes is a handy way to manage spam without dropping a lot of money.

SpamBayes is an open-source antispam product written in Python. It can run cross-platform, but its Windows implementation and Microsoft Outlook plug-in version have been very successful. The product integrates with Microsoft Outlook, and can either replace Microsoft Outlook's native spam filtering or work in conjunction with it.

The SpamBayes team recently released a public alpha 3 release of version 1.1, which is stable enough to be used on a desktop, since most of the changes are "under the hood" (in the classification system). The updates make this latest version of SpamBayes even more accurate and discriminating.

If you upgrade from an older release of SpamBayes, it's probably a good idea to rebuild the spam database by retraining the system using recent good and bad email.

Some of the new features in SpamBayes 1.1 include:

  • More sophisticated trapping of "word salad" spam -- i.e., email containing nonsense phrases or gibberish -- to disguise an image attachment that contains the actual spam. This has become a fairly major variety of spam, and not easy to filter.
  • Looking for the presence of Habeas headers to determine if email is valid.
  • Experimental support for image extraction and optical character recognition (OCR) to determine if images contain textual spam (also known as the "crack-images" option). This option relies on the ocrad and Python Imaging libraries and is only being minimally supported at this time; there's controversy about whether this is a valid way to handle the problem.
  • Many small fixes to the Microsoft Outlook plug-in.

To integrate SpamBayes with Microsoft Outlook, I've found that it's best to make the following changes:

  1. Keep a fair amount of spam in the Junk E-mail folder to train SpamBayes with when you first install it. Retrain SpamBayes from scratch (this process is largely hands-off) about once every three or four months.
  2. Set SpamBayes to move email that is certain to be spam to the Junk E-mail folder and mark it as read.
  3. Edit the Junk E-mail folder's properties. Under General, set "Show number of unread items" instead of "Show total number of items," so Junk E-mail doesn't show as bold in the folder tree view. Under AutoArchive, set the folder to permanently delete items older than three months (so that you have a future repository of spam to train from as needed).
  4. For fast hotkey access to the Spam/Not Spam function, go to the Inbox and then right-click on the Spam button in the SpamBayes toolbar, and select Customize -> Toolbars. Right-click on the button again and set the name to "S&pam" (to allow Alt-P to mark mail as spam). Close the Customize window, then open the Junk E-mail folder and repeat the same process with the "Not Spam" button -- set it to "Not S&pam" to again allow Alt-P to be used in that folder to unmark email as spam.

About the author:
Serdar Yegulalp is editor of Windows Insight, a newsletter devoted to hints, tips, tricks, news and goodies for all flavors of Windows users.

Related information from SearchExchange.com:

Dig Deeper on Exchange Server setup and troubleshooting