K9 email anti-SPAM utility for Microsoft Windows

K9

K9 is a Windows client based tool. It is written in C and so is much smaller than the competing clients which are written in Perl or Python. It also seems to have a much lower memory and processing requirement, certainly when compared against most other anti-spam tools (especially SAproxy).

K9 primarily uses Bayesian analysis but also has optional public blacklist checking and black/white lists, it automatically learns what SPAM is based on the emails that you receive and is very accurate. I've now done several installations and typically get one or two false positives in the first few days and about 97-99% accuracy overall on 50 business emails per day 75% of which is SPAM.

When an email is wrongly identified, you simply use the K9 interface to re-classify it.

There is active discussion on POPfile (on which K9 was originally based) and K9 on the grc.spam newsgroup on news.grc.com.

None Windows Use

For none-Windows users, you will not be able to use K9 but you can use POPfile which is available for both Windows and Linux platforms. POPfile is a PERL based solution so is not as fast as K9 but I am told that its accuracy is very similar.

Recommended Settings

These are the settings I use with an explanation in case you want to do things differently.

The first thing to note is that you probably do not need the black and white lists. Reserve these as a last resort. If an email is miss-identified, all you need to do is to start up the main K9 interface (double click on the icon in the system tray) and in the "recent emails" page, select the email and click on the Spam icon.

... t.b.c. - settings and screen shots (Note that the default settings work fine!) ...

What you don't get

Unlike SA, you do not get much information as to why something is classified as SPAM nor do you have the chance to write your own rules or reclassify the score on those rules. However, the automatic analysis that K9 does more than makes up for this and the only thing from SAproxy that I miss is the header: X-Spam-Level which has an asterix for each point score. This meant that I could set a rule in Outlook to give a different highlight (or automatically delete) SPAMs with a very high score.

The other thing that you don't get with K9 is local help. To get the installation and configuration options, you need to go to the keir.net web site. This can occasionally be a little frustrating when you want to tweak one setting.

Conclusion

Use it! This tool gives very high levels of accuracy, is fast and takes little resources. This is one of the best tools that you can use.