On Friday 22 November 2002 11:08 pm, [EMAIL PROTECTED] wrote: > Hi, > > I've setup another server to try out version 2.5.0. I already trained > it to identify spam but not yet non-spam messages. How do I know it's > working?
You can't, not until you train it with a non-spam corpus as well. How to train and test: - Use tools/split-corpora to split your spam corpora and your non-spam corpora into two (giving you spam.1, spam.2, non-spam.1, non-spam.2). - Remove the current Bayes databases, which are probably in ~/.spamassassin - Train Bayes on spam.1 and non-spam.1 - Copy the Bayes dbs into masses/spamassassin/, or make softlinks from that directory to the dbs; if you don't, mass-check won't do Bayes. - Run masses/mass-check on spam.2 and non-spam.2 - run "masses/hit-frequencies -p -m BAYES" to see how the Bayes rules are working. To make mass-check run more quickly, make a second copy of the spamassassin tree, then remove all of the ".cf" files from the rules/ directory, except for 10_misc.cf, 23_bayes.cf and 50_scores.cf, so that *only* the Bayes tests will be run during by mass-check. -- Give a man a match, and he'll be warm for a minute, but set him on fire, and he'll be warm for the rest of his life. Advanced SPAM filtering software: http://spamassassin.org ------------------------------------------------------- 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
