|SYSTEM INFORMATION| |
|**OS type and version:CentOS Linux 7.9.2009
|**Related products version:SpamAssassin version 3.4.0
Using: spamc (Client for SpamAssassin filter server spamd)
I’ve been working with this platform for a couple of years, and everything is doing fine. However, I wanted to understand further how SpamAssassin is being handled by Virtualmin/Webmin system. Specifically, the Bayesian Learning process.
I’ve read this virtualmin guide:
as well as some instructions from SPAMASSASIN portal directly.
According to the Virtualmin page, I understood that this application inside Virtualmin is learning from 2 methods:
- By marking the spam email as “SPAM” under USERMIN interface
- By forwarding the Spam and Ham emails to the Trap Email Aliases
But I don’t find more explanation about the script or “auto-Learn” process for the Bayesian classifier.
For example: What about if a user sees some unsolicited emails in the INBOX that were not considered as spam by SA, so the user moves them to the “spam” folder. (using IMAP client for example). How is Spamassassin “learning” about the spam emails moved to that specific folder. I remember using long time ago a script via CRON JOBS using the “sa-learn --spam" command (every week for example), but not sure if this would crash with some inner configuration that Virtualmin/Webmin has with Spamassassin installation (for example, I don’t find where it is the “auto-learn” parameter for the Global/Server wide configuration of SpamAssassin -if there is any-, and if it’s tweakable or not). And if it’s auto-learning, is it doing it from the score assigned to all emails arrived to INBOX or just from the messages in the “spam” folder already?. What about if the user changed the name of the spam folder and using a custom name like “JUNK” instead of “spam”.
I would like to properly understand the whole Spamassassin learning process without messing with other configurations inside Virtualmin.
Thanks in advance for any guidance, and to help me to understand this.