7 Mar
2006
7 Mar
'06
09:14
Jim Flanagan wrote:
Still, some spam gets thru, and I know there is a way to select which emails are spam and which are ham using the baysean filter. But I'm not sure how to do this.
A better place to ask is probably on one of the spamassassin lists, but the basic idea is you train the bayes database using known spam and known nonspam. After that you tell spamassassin to use the database to determine a probability of an email being spam. Spamassassin also has some auto-learning features and such. The actual method of training may vary - in some instances, using train-on-error is preferable to e.g. just training a standard corpus of 100K emails. /Per Jessen, Zürich