How to train Spamassassin?

I’m having trouble understanding what spam assassin is doing, if anything at all.
If I move an email to spam, the training should be, the next email received from that reported email’s address should go directly to spam.

I even tried blacklists on postfix. Nothing works. Everything still arrives on the Inbox.

Are there any commands to check the SA “training”?

This isn’t quite how the Bayes filter works. It has to get a large sampling of messages. Many corporate admins have large files full of spam messages that are used to train a new mail server using command line tools.

If a blacklist on postfix is not working, there is something not correct about the blacklist configuration or its application.

This can sometimes be a client problem. Some clients will undeclare spam based on their own rules.

Thank you for your quick reply.

I understand it uses a statistical model based on its continuous built DB.

But that should be for email content from body and subject words.

Somehow (suggestion: via interface) one could submit an email address as a key to automatically be filtered by SA.
Might try to understand how all this is connected and submit a PR.


This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.