[The Great Flamingo]

Bayesian Filtering uses user input to determine the probability of certain words turning up in "spam" e-mail, and then tries to determine the probability that an e-mail with those words will be spam.

Every time a mail user (you) receives a spam e-mail, you will mark it using a "mark as spam" button. The filter than picks up on that and assigns values to each word in the spam e-mail. If a certain word keeps turning up in e-mails that you mark as spam, then those e-mails will eventually be filtered out.

The advantage is that it allows the e-mail user to select the types of mail that will or won't be marked as spam.