Spamassassins Bayesian learning filter

Hey everyone,

I was wondering how the learning filter for spam works. We had some reports from differen domains of receiving a lot of spam. We made some adjustments to the spamfilter which made the spam mails become way less. However we noticed that the learning filter only works for the specific domain. Instead we would like the learning filter to work for all domains on the server. It is a lot of work to go through all domains by hand to make sure the learning filter works the way we want to. If the learning filter works for all domains you only need to configure it for one domain and the rest can use it as well. Otherwise you still kind of have the problem that specific kinds of spam will be received by the other domains on the server.


It depends on how you have SpamAssassin configured. If you’ve configured it to allow “per-domain” settings, then each domain will have to establish it’s patterns.

If you’ve configured it server wide, then all reports will affect all domains respectively.

Keep in mind, it takes time for SpamAssassin to learn about spam in order to prevent as much false positives. Make sure everyone is reporting spam regularly.

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.