On Sat, Jun 14, 2003 at 04:38:58PM +0100, Mathew Hendry wrote: > Jim Ford wrote: > > > If auto-learn is on (default setting), is there any point in putting > > correctly identified spam and ham through sa-learn? Is the only use > > for sa-learn to teach SA when spam or ham is incorrectly identified? > > No, auto-learn learns only from mails with particularly high or low > scores. The default thresholds are > > auto_learn_threshold_nonspam -0.5 > auto_learn_threshold_spam 10 > > You could override those thresholds, but any false positives/negatives > would skew the statistics. I'd follow the suggestion in the sa-learn > manual and train it periodically with hand-classified mail. > > (So to answer one of your other questions, it's a good idea to keep > caughtspam for sa-learning, at least for a month or two).
Thanks - question fully answered! I seem to recall others have suggested using cron to periodicaly run sa-learn. But it seems to me that this is wrong without hand checking the spam/ham mailboxes first. I might set up cron to just remind me to do it - maybe once a month. Regards: Jim Ford -- Spam poison - don't use! ---> [EMAIL PROTECTED] <--- ------------------------------------------------------- This SF.NET email is sponsored by: eBay Great deals on office technology -- on eBay now! Click here: http://adfarm.mediaplex.com/ad/ck/711-11697-6916-5 _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
