Hi, I have a question regarding mass-check parameter -learn. I want mass-check to learn from 50% of my database, and then to apply bayes rules on the other half of my corpus.
So the command running mass-check looks like this : ./mass-check --progress --bayes -j 2 --all --learn=50 --restart=400 ham:detect:../../../HAM/ spam:detect:../../../SPAM the user_prefs file looks like this : bayes_auto_learn 0 lock_method flock bayes_store_module Mail::SpamAssassin::BayesStore::SDBM use_auto_whitelist 0 (I am actually proceeding according to the manual on http://wiki.apache.org/spamassassin/RescoreDetails) So I expected mass-check to finish with one half of the database ranked as "set=0" and the other half as "set=2", but the results were different. So I want to ask, if someone knows, what could be the problem. -- View this message in context: http://old.nabble.com/mass-check--learn-tp31602414p31602414.html Sent from the SpamAssassin - Dev mailing list archive at Nabble.com.
