Any Server-Side Anti-Spam that works at the user level?

[IT_Architect]

Any Server-Side Anti-Spam that works at the virtual E-Mail user level?

We get rid of 90% of the spam using SpamAssassin, but 10% is a big number. For PCs, I use PopFile for POP3 on the client. Nothing works better than that. The problems are webmail and phones. In order to have good filtering, it must be able to be trained by the user, and filtered and categorized on the server. Is there a solution for this problem that we can use?

 

[SeLLeRoNe]

This may be interesting for you: http://help.directadmin.com/item.php?id=188

Regards

 

[IT_Architect]

Good find. That is interesting. The way I read that is in practice, the virtual users are collectively determining what is spam, and what isn't spam, for the entire domain. Is that how you read it?

 

[SeLLeRoNe]

Well as far as i did understood that they learn for the entire spamassassin system, but i may be wrong.

Perhaps someone else may give a better response.

Regards

 

[IT_Architect]

Well as far as i did understood that they learn for the entire spamassassin system, but i may be wrong.

I see a .spamassassin directory at the user level with bayes_* files in it. The script runs at the user level. Thus, I would assume from that, that it is at least down to the user level.

 

[SeLLeRoNe]

Well since nowhere in the script (except for the script itself position) is defined the domain (at least on sa-learn command), i thot it was server-level, but maybe the command itself recognize the receiver...

I made a script that run every two hours for all mail account on the server based on that script.

Regards

 

[SeLLeRoNe]

I'd noticed that files get created in /root/.spamassassin, so i presume that happend because scripts i made is running at root level, so there would be a way to tell sa-learn a specific user/path where to save informations instead of create a single server-wide file

Any hint?

Regards