Improoving spamassassin

I’ve got a mailserver running that fetches mails from different accounts via fetchmail, scans them for viruses using clamav, filters for SPAM using spamassassin and then deliveres it to the mailbox and sorts them via procmail. Now Spamassassin just filtered arround 35-40% of all SPAM I recieved. My local.cf includes

use_bayes 1
bayes_auto_learn 1
bayes_auto_lern_threshold_nonspam 1
bayes_auto_lern_threshold_spam 14.00
bayes_ignore_header X-Spam-Flag
bayes_ignore_header X-Spam-Status
In order to recognize more SPAM i created a folder SPAM in my Mailbox and moved all SPAM I recieved into it. Now simply change to that folder and run a sa-learn –spam –progress * there, as well as a sa-learn –ham –progress * in your inbox and voila… SPAM-recognition by spamassassin starts improoving very fast :-)

(Thanks Diederik for the hint)