spamassassin bayes

[Red Squirrel]

I just setup all my mail folders so that any spam not making it past the filters is put in a folder which is scanned daily by sa-learn, my question is, what is a quick way of checking that bayes is even working, other then the fact that spam will be reduced? Just wondering since I want to ensure that it is infact working.

 

[Nothinman]

You really want sa-learn to 'learn' from the same mailbox every day on it's own?

 

[Red Squirrel]

Yeah since throughout the day I add new spam to it, as time goes by when I find myself with less spam making it through I'll probably set it up to only do it once a week. I could probably also make it so the script clears the folder, but I think the folders have to have the permissions of the mail user, so that would mean I'd also have to make the script chown each folder, though it is an option. But given that sa-learn reconizes messages it already learned, it's not a big issue, and I can simply manually clear them once in a while. The script works great, it must of learned at least 500 messages in a few days, I just don't know how to tell if bayes is actually applied to filters.

 

[Nothinman]

spamassassin modifies the message however you tell it, usually either adding somethign to the subject or at least adding the score to the headers so if you want to know if it's being applied to your filters you'll have to look at your filters to know if they're setup to look for that.

 

[Red Squirrel]

So it should show in the headers then? Since it does not in mine. So how do I turn bayes on?

Received: from localhost by borg
with SpamAssassin (version 3.0.2);
Tue, 10 May 2005 01:30:10 -0400
From: "Katelyn Chavez" <rrtyoirr@assassins.ch>
To: <redsquirrel@iceteks.com>
Subject: (11.0) [SPAM] Special report reveals this SmallCap
Date: Tue, 10 May 2005 08:21:00 +0200
Message-Id: <910906592456.FPY46799@deferent.born.org>
X-Spam-Flag: YES
X-Spam-Checker-Version: SpamAssassin 3.0.2 (2004-11-16) on borg
X-Spam-Level: ***********
X-Spam-Status: Yes, score=11.0 required=4.0 tests=AGENMEDS0,
AGEN_offer_1107628832,AGEN_profit_1107977741,AGEN_stock_1107629635,
AGEN_stock_1107807524,AGENinvest2216022332,AGENmarket4432044104,
AGENopportunit0,CUST_TEXTMIME,US_DOLLARS_3 autolearn=no version=3.0.2
X-Spam-Report:
* -0.1 CUST_TEXTMIME CUST_TEXTMIME
* 0.2 AGENmarket4432044104 BODY: market often found in spam
* 3.0 AGENMEDS0 BODY: meds
* 0.9 AGEN_profit_1107977741 BODY: profit related
* 0.9 AGENinvest2216022332 BODY: invest used in spam
* 1.0 AGEN_stock_1107629635 BODY: stock related
* 3.0 AGEN_stock_1107807524 BODY: stock related
* 0.4 US_DOLLARS_3 BODY: Mentions millions of $ ($NN,NNN,NNN.NN)
* 0.9 AGEN_offer_1107628832 BODY: talks about special offers
* 0.8 AGENopportunit0 BODY: opportunit found in spam in various combinations
MIME-Version: 1.0
Content-Type: multipart/mixed; boundary="----------=_428046E2.D4929BCA"
Status: RO
X-UID: 47
Content-Length: 6823
X-Keywords:

 

[Nothinman]

X-Spam-Flag: YES
X-Spam-Checker-Version: SpamAssassin 3.0.2 (2004-11-16) on borg
X-Spam-Level: ***********
X-Spam-Status: Yes, score=11.0 required=4.0 tests=AGENMEDS0,
AGEN_offer_1107628832,AGEN_profit_1107977741,AGEN_stock_1107629635,
AGEN_stock_1107807524,AGENinvest2216022332,AGENmarket4432044104,
AGENopportunit0,CUST_TEXTMIME,US_DOLLARS_3 autolearn=no version=3.0.2
X-Spam-Report:
* -0.1 CUST_TEXTMIME CUST_TEXTMIME
* 0.2 AGENmarket4432044104 BODY: market often found in spam
* 3.0 AGENMEDS0 BODY: meds
* 0.9 AGEN_profit_1107977741 BODY: profit related
* 0.9 AGENinvest2216022332 BODY: invest used in spam
* 1.0 AGEN_stock_1107629635 BODY: stock related
* 3.0 AGEN_stock_1107807524 BODY: stock related
* 0.4 US_DOLLARS_3 BODY: Mentions millions of $ ($NN,NNN,NNN.NN)
* 0.9 AGEN_offer_1107628832 BODY: talks about special offers
* 0.8 AGENopportunit0 BODY: opportunit found in spam in various combinations

All those are headers added by spamassassin.

 

[Red Squirrel]

Yeah, but those are specific to rules added, how can I tell if bayes is also being run on these messages? Since I don't think bayes actually generates rules, or does it?

 

[Nothinman]

bayes is being used if you have use_bayes set. And you should see a line BAYES_XX, with the XX being a spam percentage guess, in the tests line so I don't think you enabled bayes.

 

[Red Squirrel]

Is the command suppost to be use_bayes 1 in the local.cf file? Since I got that. But I think it needs 200 of spam and ham before it's activated, so maybe I just have to wait. Got more then 200 spam but for ham I probably don't have that much yet.

 

[Nothinman]

Yea, it's just use_bays in local.cf

 

[Red Squirrel]

Oh ok so I guess there's not 200 ham yet then, so I just have to wait, as it only works when there's over 200, for acuracy reasons.