On Mon, Oct 04, 2004 at 11:37:35AM -0700, [EMAIL PROTECTED] wrote: > 0.000 0 2563 0 non-token data: nspam > 0.000 0 2963 0 non-token data: nham > > Once the third number in each of these lines is over 200, SpamAssassin will > turn on the BAYES_## rules > There are threshold spam-scores that SpamAssassin uses to auto-train Bayes... > these are from a Bayes-less scoreset to avoid feedback loops.
Also note that just because Bayes is used for the scan, doesn't mean there'll actually be a result. If there aren't enough tokens in the message that are also in the Bayes DB, the Bayes code aborts since it won't be able to generate a probability. 200/200 is a bare minimum. -- Randomly Generated Tagline: I did this 'cause Linux gives me a woody. It doesn't generate revenue. (Dave '-ddt->` Taylor, announcing DOOM for Linux)
pgp7kLzvmm9MA.pgp
Description: PGP signature
