I noted that I get quite different scores for spamassassin -D --lint on different machines depending on Bayes. On one system I get BAYES_10 (a well trained system with lots of mail in the db), on another system I get BAYES_70 (with only a few hundred mails in the db). What does sa use for testing here and what should it score approximately, as ham or spam?
Well, a "ideal" bayes database training will always give you 0 for ham, and 100 for spam.
Of course, reality isn't ideal, but the lower the score is for a ham message, and the higher it is for spam, the better the training of the database.
------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
