[ Date Index ] [ Thread Index ] [ <= Previous by date / thread ] [ Next by date / thread => ]
James Fidell wrote: > I feed sa-learn regularly, but some of the latest ones just don't seem > to be triggering the filters :( Perhaps clearing the bayesian database > and retraining would help. I don't think I'd do that. The database contains accumulated wisdom which is still making useful discriminations for you and I wouldn't chuck that. Keep training-up the database you've already got. Myself, I feed /all/ messages identified as spam (whether by spamassassin or by me) to sa-learn --spam, and all the messages I wanted to receive (similarly) to sa-learn --ham. I do this daily, or more often, and don't throw away any received message till it's been through sa-learn at least once. The autolearned ones don't /have/ to go through this process, but I find it's too much trouble to separate them. You might also study the values you can adjust in ~/.spamassassin/user_prefs : but make changes singly and by small steps ! John -- John Palmer Preston near Weymouth, Dorset, England e-mail: johnp@xxxxxxxxxx (plain text preferred) website: http://www.palmyra.uklinux.net/ -- The Mailing List for the Devon & Cornwall LUG http://mailman.dclug.org.uk/listinfo/list FAQ: http://www.dcglug.org.uk/linux_adm/list-faq.html