On Tue, 30 May 2017, Robert Kudyba wrote:
I note that message hit BAYES_00. If content like that is getting a
"strong ham" Bayes score, you should review your training processes and
Bayes corpora - you *do* keep copies of messages you train Bayes with,
right? :)
Yes just re-synced.
Did you do any review before re-training? Re-training with
misclassifications in the corpora will not correct the problem.
But: fixing your Bayes and getting a non-forwarding DNS server for your
mail system so that you're not hitting RBL query limits are the biggest
things you need to do to address this.
It’s enabled and looks like it’s working based on this and that use_bayes 1 in
local.cf
sa-learn --dump magic
0.000 0 3 0 non-token data: bayes db version
0.000 0 688 0 non-token data: nspam
0.000 0 80012 0 non-token data: nham
That seems somewhat out-of-balance, and might lead to FNs due to Bayes.
You should try to get more spam to train.
--
John Hardin KA7OHZ http://www.impsec.org/~jhardin/
jhar...@impsec.org FALaholic #11174 pgpk -a jhar...@impsec.org
key: 0xB8732E79 -- 2D8C 34F4 6411 F507 136C AF76 D822 E6E6 B873 2E79
-----------------------------------------------------------------------
People think they're trading chaos for order [by ceding more and
more power to the Government], but they're just trading normal
human evil for the really dangerous organized kind of evil, the
kind that simply does not give a shit. Only bureaucrats can give
you true evil. -- Larry Correia
-----------------------------------------------------------------------
7 days until the 73rd anniversary of D-Day
