At Fri Nov 14 23:03:42 2003, Alan Munday wrote: > > Having just read the FAQ's can I just check the requirements for the > source SPAM/HAM mail for sa-learn? > > Do the contents of the sources messages need to be false positives > for the HAM file and false negatives for the SPAM file?
No. To work well, sa-learn needs a good representative sample of all your spam and ham, not just those that are FP/FNs. You do need to be careful, particularly in the early days of using Bayes, that you correct any erroneous auto-learning of messages (i.e. where a spam message is auto-learned as ham, and vice versa), by using sa-learn to learn those messages with the correct type. > I can see that I'll be able to assemble 200 false negatives easily, however > I get very few false positives. > > Once you make it over the 200 limit will Bayesian filtering activate for > testing e.g. false negatives before enough false positives have been > learned? Bayes won't start working until it has learned 200 hams and 200 spams. Martin -- Martin Radford | "Only wimps use tape backup: _real_ [EMAIL PROTECTED] | men just upload their important stuff -o) Registered Linux user #9257 | on ftp and let the rest of the world /\\ - see http://counter.li.org | mirror it ;)" - Linus Torvalds _\_V ------------------------------------------------------- This SF. Net email is sponsored by: GoToMyPC GoToMyPC is the fast, easy and secure way to access your computer from any Web browser or wireless device. Click here to Try it Free! https://www.gotomypc.com/tr/OSDN/AW/Q4_2003/t/g22lp?Target=mm/g22lp.tmpl _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
