Bayes filter training with POP3 accounts

<< Click to Display Table of Contents >>

Navigation:  Spam Filter (SpamAssassin) >

Bayes filter training with POP3 accounts

It is possible to train the Bayes filter via mails sent to specific POP3 accounts and applying rules to them.

For this, two additional POP3 accounts are required as seen in the following picture:


One account to train as "Ham" and one to train as "Spam". You can name the accounts as you like but they should be recognizable for better usability. The same goes for the mail address belonging to the account.

Each account should be configured like the following picture shows.


A SMTP server address is not needed because you won't forward any mails.

Now you have to create a rule for each of those two accounts which determins that any message sent to one of those accounts will be trained either as "Ham" or "Spam".

An example is seen in the following picture.


In case you are using an existing SMTP server you should also enable the "Do not deliver"-Action.