I've been using this technique for several years now, making sure that my honey-pot addresses are widely available. Any email that matches a whitelist gets through immediately. Anything that matches my blacklist goes straight to spam.
Everything else is put through a comparison filter that looks to see if it's similar to something that's been blacklisted. If so, to goes in the spam folder.
If something is then still unclassified it goes through a Bayesian filter trained on the explicitly black/white messages, and prioritised accordingly. Anything that's closer to black than white gets delayed a day or two, anything that's nearly white goes in my inbox.
Works a treat, and in my experience, even with only a comparatively small amount of email, it beats the gmail filters, because it's trained on my data.
Everything else is put through a comparison filter that looks to see if it's similar to something that's been blacklisted. If so, to goes in the spam folder.
If something is then still unclassified it goes through a Bayesian filter trained on the explicitly black/white messages, and prioritised accordingly. Anything that's closer to black than white gets delayed a day or two, anything that's nearly white goes in my inbox.
Works a treat, and in my experience, even with only a comparatively small amount of email, it beats the gmail filters, because it's trained on my data.