Tuesday, February 10, 2004

I'll Bet on the Baye

A Plan for Spam by Paul Graham describes simply how software that recognizes the message sent by the spammers (using an algorithm Better Bayesian Filtering) can catch 995 spam out of a 1000 with few false positives.  Graham uses the spam itself to create statistical filtering based on the words that spammers use that our regular e-mail does not contain.