When using the standard disk-based Bayesian database, SpamAssassin keeps a Bayesian database for each user account it is run under, not for each user it processes email for. The alternative is to use SQL; this is covered in Chapter 14.

Although Bayesian filtering is enabled by default, SpamAssassin will not use the filter until it has learned enough spam and ham emails to be able to make a decision when processing an email. SpamAssassin has an auto-learn facility, which allows emails to be used to train the Bayesian filter automatically. It also has command-line tools to allow a user or system administrator to train the Bayesian filter manually.

To use automatic learning, set the bayes_auto_learn flag to 1. This can be configured site-wide in the /etc/mail/spamassassin/local.cf file, and can be overridden in a user's ~/.spamassassin/user_prefs file. Two other configuration flags also affect auto-learning, and are the thresholds for learning ham and spam. These values are in the...