I cannot express in words how happy I am. A picture is worth a thousand words right? How many words is a dynamic picture worth? Using these instructions you too can be as SPAM free as I, all while running Mac OS X. Special Thanks to Neil Lee on his help on the config.in file. Just a warning. This requires the instillation of the Developer Tools, which comes with OS X these days. The new version of POPFile uses the BerklyDB perl module, these files need to be compiled. But don't fret, this is very easy to do even if you are a nOOb at the command line. If all goes well you can just copy the commands from Neil's site and paste them into the Terminal.
The end result is more then worth the hassle. Let me say that again. The end result of running POPFile on Mac OS X is more then worth the hassle of compiling some files in the command line. Why? POPFile offers a superior bayesian SPAM filtering system with "buckets" and "magnets" to provide the best SPAM filtering system known to man.
Mac OS X Mail does not have bayesian filtering in it (score one for Outlook 2003, that is IF you want to pay the steep upgrade fee) and that is not good enough for me. Mac OS X Mail is pretty good after you train it for a while. But it is nowhere near as accurate as using POPFile. I have used POPFile for a full 8 months on my PC, and it got up to 99% accurate, and to the point to where there were NO false positives.
Article comments
1 - Phillip Winn
Actually, I believe that Mail.app does include Bayesian filtering, just not a very good implementation of it. Something about not including certain headers, or something like that.
2 - Ken Edwards
Mail does not include Bayesian filtering. I checked on this when the new Junk Mail filter was intro'd in 10.2. The Panther improvement to the junk mail filter is that it now supports ISP junk mail headers. it also has a preference section to it self, but thats not really anything new on the classification of junk mail.
do a google for "adaptive latent semantic analysis" that is what Apple uses for Mail.