-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Theo Van Dinter writes: >On Mon, Jan 12, 2004 at 11:26:46AM -0500, Jack Gostl wrote: >> I sometimes wonder if the whole system wouldn't be smaller and stabler if >> it were entirely Bayes based. > >Well, sure it would be. But frankly, if you want Bayes only, there >are several other projects out there to look at... SpamBayes, CRM114, >etc, etc. > >Bayes works fairly well on its own, assuming decent learning, until >spammers start attacking the algorithm or the code. Then it kind of >falls on its face until the attack is countered. > >The idea of SA is to use multiple detection systems such that even if >one or two don't work, the message will still probably be caught from >the others. BTW I've attempted to come up with ways to use Bayes as the core in the past, without much luck; the current mechanism gives the highest degree of accuracy in my tests. - --j. -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.2 (GNU/Linux) Comment: Exmh CVS iD8DBQFAAumxQTcbUG5Y7woRAj64AJ0ROK7JL3eLTrqb4q4ek7XayRqPyQCgi9wu hFcrnT9VDbl6h8bPA8bCR6I= =nXgD -----END PGP SIGNATURE----- ------------------------------------------------------- This SF.net email is sponsored by: Perforce Software. Perforce is the Fast Software Configuration Management System offering advanced branching capabilities and atomic changes on 50+ platforms. Free Eval! http://www.perforce.com/perforce/loadprog.html _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
