-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Matt Kettler writes: >At 06:48 PM 12/10/2003, stan wrote: >>Did I do wrong by teaching it with lots of _good_ messages? Should I reset >>it to the base rules, and start over? BTW how can I do that? > >Idealisticaly you want to train it with something "realistic" in terms of >spam/ham ratio.. ie: something close to what you get in reality. Bayes will >tolerate a lot of variance from the "ideal" ratio, but you can't go too far >off course with it... 200 spams and 200,000 hams will result in a bad database. There's also a possibility that you trained all of those messages as "ham" - -- ham is *non*-spam. You need to train on *both* ham and spam for it to work. - --j. -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.2 (GNU/Linux) Comment: Exmh CVS iD8DBQE/17ZLQTcbUG5Y7woRAhoYAKCh+5rJP/FB12N6oPbKBTh9G3DEoACg12qQ Xm5hvAZ58o2dNZ4r2TrpzFE= =4BSF -----END PGP SIGNATURE----- ------------------------------------------------------- This SF.net email is sponsored by: IBM Linux Tutorials. Become an expert in LINUX or just sharpen your skills. Sign up for IBM's Free Linux Tutorials. Learn everything from the bash shell to sys admin. Click now! http://ads.osdn.com/?ad_id=1278&alloc_id=3371&op=click _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
