You can train your mail filters on actual messages you receive. To do this:
, non-spam -
,
or not recognized at all -
. If you have
already trained your spam filter on a message and the
results were recorded in the Bayesian database, an icon
is shown in the
right column.A spam filter database can become corrupt, or the data inside it unreliable. This may happen when many non-spam e-mails have been learned as spam, or vice versa, and the spam filter subsequently produces incorrect results. If the spam filter appears to be working incorrectly, clear its database and start afresh. To do this, click the ‘Clear Spam Filter Database’ button.