bayes ham/spam is not working

[mejay]

When I receive spam (or not), it doesn't seem to ever use bayes to determin ham/spam. It is only using the other non-bayes rules.

In the log, I have this:
"The ham class needs more training samples. Currently: 0; minimum 200 required"

I've been using rspamd for awhile. What exactly do I need to do to make bayes work with rspamd in directadmin? Do the emails that it determines are spam and not spam, are they not used as training samples?

Is there some directadmin rspam documentation that I am overlooking to make this work?

 

[zEitEr]

Hello,

"The ham class needs more training samples. Currently: 0; minimum 200 required"

This is the answer. You should train the antispam software. This subject was already discussed here many times, as well as the forums already contain instructions on how to complete it. Search the forums.

Related:

- https://www.google.com/search?q=train+antispam+site:directadmin.com
- https://docs.directadmin.com/other-...omatically-marking-moved-mail-to-junk-as-spam

 

[mejay]

I already have those files, but the "sa-learn.sh" triggers the file "/usr/bin/sa-learn" which does not exist in rspamd, only with spamassassin. So, that document does not work for rspamd.

 

[zEitEr]

So, that document does not work for rspamd.

Follow instructions for rSpamd. There are instructions for rSpamd on directadmin forums

 

[Cordarex]

I came across the following project:

Automating Spam Training with rspamd-learn-helper

Fighting spam is a constant battle. Even with powerful tools like Rspamd, it’s easy to forget to regularly train your filters, especially in multi-user or low-maintenance environments.

www.filipnet.de

Github: https://github.com/filipnet/rspamd-learn-helper

It makes sense to make the training a little more elegant.
Currently, I only have one folder, and that is spam. Unfortunately, important emails have also been classified as spam. AND, of course, I have entered this as HAM. Unfortunately, this does not seem to work properly, or the learning process is a little slow...

This Project may help to make learning and permanent sharpening more targeted.