You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@spamassassin.apache.org by David B Funk <db...@engineering.uiowa.edu> on 2017/05/10 16:33:44 UTC
block Bayes autolearn for specific messages
Is there any way to use Bayes autolearn in general but prevent it from learning
specific messages?
I have a specific source of messages (Office-365) which I would like to prevent
from being autolearn (with out scoring them as spam).
I still want those messages to be SA scored using the normal methods, just not
be considered -at-all- for autolearning.
--
Dave Funk University of Iowa
<dbfunk (at) engineering.uiowa.edu> College of Engineering
319/335-5751 FAX: 319/384-0549 1256 Seamans Center
Sys_admin/Postmaster/cell_admin Iowa City, IA 52242-1527
#include <std_disclaimer.h>
Better is not better, 'standard' is better. B{
Re: block Bayes autolearn for specific messages
Posted by John Hardin <jh...@impsec.org>.
On Wed, 10 May 2017, David B Funk wrote:
> On Wed, 10 May 2017, John Hardin wrote:
>
>> On Wed, 10 May 2017, David B Funk wrote:
>>
>> > Is there any way to use Bayes autolearn in general but prevent it from
>> > learning specific messages?
>> >
>> > I have a specific source of messages (Office-365) which I would like to
>> > prevent from being autolearn (with out scoring them as spam).
>> >
>> > I still want those messages to be SA scored using the normal methods,
>> > just not be considered -at-all- for autolearning.
>>
>> bayes_ignore_from user@example.com
>>
>> bayes_ignore_to user@example.com
>
> John,
> Thanks for the suggestion but I still want Bayes classifier run on those
> messages, just no autolearning.
>
> bayes_ignore_(to|from) prevents both.
Yeah, I was wondering about that being a deal killer.
> Is there some kind of score calculation rule that does something along the
> line of "if total score is less than N, add M"
Nope.
I think you may be stuck doing something like saving those messages and
running them through sa-learn --forget.
You might want to open a bugzilla feature request for something like
tflags RULENAME suppress_autolearn
and/or
bayes_noautolearn_(to|from)
--
John Hardin KA7OHZ http://www.impsec.org/~jhardin/
jhardin@impsec.org FALaholic #11174 pgpk -a jhardin@impsec.org
key: 0xB8732E79 -- 2D8C 34F4 6411 F507 136C AF76 D822 E6E6 B873 2E79
-----------------------------------------------------------------------
If guns kill people, then...
-- pencils miss spel words.
-- cars make people drive drunk.
-- spoons make people fat.
-----------------------------------------------------------------------
41 days since the first commercial re-flight of an orbital booster (SpaceX)
Re: block Bayes autolearn for specific messages
Posted by David B Funk <db...@engineering.uiowa.edu>.
On Wed, 10 May 2017, John Hardin wrote:
> On Wed, 10 May 2017, David B Funk wrote:
>
>> Is there any way to use Bayes autolearn in general but prevent it from
>> learning specific messages?
>>
>> I have a specific source of messages (Office-365) which I would like to
>> prevent from being autolearn (with out scoring them as spam).
>>
>> I still want those messages to be SA scored using the normal methods, just
>> not be considered -at-all- for autolearning.
>
> bayes_ignore_from user@example.com
>
> bayes_ignore_to user@example.com
John,
Thanks for the suggestion but I still want Bayes classifier run on those
messages, just no autolearning.
bayes_ignore_(to|from) prevents both.
I've already got a rule that adds a small score (0.3) to those messages but
unfortunately they hit minus-score rules (EG: RCVD_IN_MSPIKE_*,
KHOP_RCVD_TRUST, etc) often enough that they still get learned.
I could jack up the local score add but then I run the risk of FPing O365
messages that don't hit the minus-score rules.
Is there some kind of score calculation rule that does something along the line
of "if total score is less than N, add M"
Dave
--
Dave Funk University of Iowa
<dbfunk (at) engineering.uiowa.edu> College of Engineering
319/335-5751 FAX: 319/384-0549 1256 Seamans Center
Sys_admin/Postmaster/cell_admin Iowa City, IA 52242-1527
#include <std_disclaimer.h>
Better is not better, 'standard' is better. B{
Re: block Bayes autolearn for specific messages
Posted by John Hardin <jh...@impsec.org>.
On Wed, 10 May 2017, David B Funk wrote:
> Is there any way to use Bayes autolearn in general but prevent it from
> learning specific messages?
>
> I have a specific source of messages (Office-365) which I would like to
> prevent from being autolearn (with out scoring them as spam).
>
> I still want those messages to be SA scored using the normal methods, just
> not be considered -at-all- for autolearning.
bayes_ignore_from user@example.com
bayes_ignore_to user@example.com
--
John Hardin KA7OHZ http://www.impsec.org/~jhardin/
jhardin@impsec.org FALaholic #11174 pgpk -a jhardin@impsec.org
key: 0xB8732E79 -- 2D8C 34F4 6411 F507 136C AF76 D822 E6E6 B873 2E79
-----------------------------------------------------------------------
The philosophy of gun control: Teenagers are roaring through
town at 90MPH, where the speed limit is 25. Your solution is to
lower the speed limit to 20. -- Sam Cohen
-----------------------------------------------------------------------
41 days since the first commercial re-flight of an orbital booster (SpaceX)
Re: block Bayes autolearn for specific messages
Posted by RW <rw...@googlemail.com>.
On Wed, 10 May 2017 11:33:44 -0500 (CDT)
David B Funk wrote:
> Is there any way to use Bayes autolearn in general but prevent it
> from learning specific messages?
>
> I have a specific source of messages (Office-365) which I would like
> to prevent from being autolearn (with out scoring them as spam).
Perhaps, but it's a bit cheesy. If you are trying to prevent them
being learned as ham you could make a rule that scores something like
6 6 0.001 0.001
i.e. that adds little to the classification score, but adds enough to
the autolearning score to sabotage ham learning.
If this to keep the