Bayes scoring priority

[Amir 'CG' Caspi <ce...@3phase.com>]

Hi all,

	So, I think I've gotten my Bayes DB largely under control... 
most of the FN spam I'm getting is getting high Bayes scores and 
simply not a large enough aggregate score to count as spam.  So, now 
I'm wondering if I should increase the points assigned to high Bayes 
scores.

For example, here is the header of one spam I just received:

X-Spam-Status: No, score=3.5 required=5.0 tests=BAYES_99,HTML_EXTRA_CLOSE,
      HTML_MESSAGE autolearn=no version=3.3.1

So, it's hitting Bayes_99, but is not "spammy" enough according to 
the other tests.  In fact, I got 13 (!!) FN spams this morning in my 
inbox which hit Bayes_99 but did not hit enough other rules to pass 
my required 5.0 spam score threshold.  Most of these mails had scores 
above 4, just not quite enough to hit 5.  (One or two had scores in 
the 3s.)

My Bayes scores are set to the default, i.e. Bayes_99 adds 3.5 points 
(I have network tests enabled).  I'm considering just making this 
5.0, so ANYTHING with Bayes_99 is automatically spam.  I looked 
through my inbox and I have no true ham with Bayes_99.

I could, of course, make it a bit less severe, e.g. setting to 4.5... 
this would still require some other rules to pop before getting 
counted as spam, but would take care of all of these FNs with scores 
of 4+ (but not the 3+ ones, though these are much more rare).

Has anyone modified their Bayes scoring priority, and if so, what 
were your experiences?  What scores did you assign?

Cheers.
						--- Amir

Re: Bayes scoring priority

[RW <rw...@googlemail.com>]

On Mon, 24 Jun 2013 11:29:20 -0600
Amir 'CG' Caspi wrote:

> Hi all,


> My Bayes scores are set to the default, i.e. Bayes_99 adds 3.5 points 
> (I have network tests enabled).  I'm considering just making this 
> 5.0, so ANYTHING with Bayes_99 is automatically spam.  I looked 
> through my inbox and I have no true ham with Bayes_99.

SA score generation targets an overall FP rate of 1 in 2500. IIWY I'd
wait until I'd established that there are no BAYES_99 FPs in a few
thousand hams.

Re: Bayes scoring priority

[Ben Johnson <be...@indietorrent.org>]

On 6/24/2013 1:29 PM, Amir 'CG' Caspi wrote:
> Has anyone modified their Bayes scoring priority, and if so, what were
> your experiences?  What scores did you assign?

This has been discussed at length; perhaps start with this archived topic:

http://spamassassin.1065346.n5.nabble.com/BAYES-99-and-ham-td38832.html

The short answer is that you can, and probably should, increase the
BAYES_99 score value to 4 or 4.5. Setting it to 5 puts you at risk
(albeit very slight) for false-positives.

-Ben