So about a year ago I was starting to get a lot of spam so I tried different techniques to combat the problem. Long story short, the solution I found to be the most effective was simply to use bogofilter. After using it for about a year without any retraining I finally had to retrain my filter as there was 2-3 specific kinds of mails it kept misclassifying as non-spam. During that period only 100-200 spam mails had been registered as legitimate mail and 45000 spam mails had been correctly classified. I stopped checking for real mails classified as spam after about 2 weeks of no false positives. That’s pretty damn impressive. All hail bogofilter 🙂