Are Bayes tokens used on all domains?

Use this forum for discussions about SpamAssassin and anti-spam in general.
Post Reply
palinka
Senior user
Senior user
Posts: 2186
Joined: 2017-09-12 17:57

Are Bayes tokens used on all domains?

Post by palinka » 2018-08-11 15:43

Situation: I have a global rule that forwards ALL spam, including those above the HMS delete threshold, to an account on a domain that I set up specifically to review and sort spam for Bayes learning. This is the only account on the domain and it's not used for sending email, etc - only for sorting spam & ham for Bayes learning.

Question: do the tokens learned from this account get used in scoring every incoming message to every domain? I *think* it does, but I'm not positive and if the answer is no, then I have to rethink my Bayes learning strategy. Thanks.

palinka
Senior user
Senior user
Posts: 2186
Joined: 2017-09-12 17:57

Re: Are Bayes tokens used on all domains?

Post by palinka » 2018-08-11 15:50

Question 2 (just popped into my head): is it worth keeping an archive of all spam messages? I am currently using the "backup & clear down" script to delete old messages in trash and spam folders. The only reason I can think of would be if I had to rebuild the Bayes db from scratch, but the only event I could see that happening is if I had a catastrophic crash and also lost my HMS backups - in which case I'd have lost my spam archive as well. ;)

Just curious if there are any good reasons because I can't think of any but my gut feeling tells me it's a good idea to save them.

User avatar
SorenR
Senior user
Senior user
Posts: 3835
Joined: 2006-08-21 15:38
Location: Denmark

Re: Are Bayes tokens used on all domains?

Post by SorenR » 2018-08-11 17:26

Allthough SpamAssassin DO support users and differentiated scoring, the hMailServer implementation is NOT user-centric.

I clean 30+ days old SPAM from all users EXCEPT my Spam user... Sometimes you need to do a historic search. Currently just over 10,700 mails covering 2 years. Just about maximum of what I would keep...
SørenR.

" I will initiate self-destruct. " — IG-11.

palinka
Senior user
Senior user
Posts: 2186
Joined: 2017-09-12 17:57

Re: Are Bayes tokens used on all domains?

Post by palinka » 2018-08-11 18:36

SorenR wrote:
2018-08-11 17:26
Allthough SpamAssassin DO support users and differentiated scoring, the hMailServer implementation is NOT user-centric.

I clean 30+ days old SPAM from all users EXCEPT my Spam user... Sometimes you need to do a historic search. Currently just over 10,700 mails covering 2 years. Just about maximum of what I would keep...
Well that makes sense. My gut it usually right. :wink:

Thanks.

Post Reply