On Wed, 9 Jul 2014, Asai wrote:
Greetings,
We've been running Spamassassin (3.3.1 currently, concurrently with Amavis)
using MySQL as a backend for many years now and we have 1 million + entries
in the Bayes table. At this time, there seems to be a lot of spam getting
through the filters and we currently have our spam level set to 2.5 points
for users with the most spam.
A couple questions:
Does 2.5 seem excessively low?
Yes. The base rules are scored on the assumption that 5.0 is the spam
threshold. Reducing the threshold will likely increase false positives.
Is it advisable to clear out the Bayes table and start from scratch?
That depends. Are spams getting low Bayes scores? If so, that's an
indication of mistraining and a need to retrain.
If so, would it be wise to raise the level to 4.0 while the Bayes data
retrains?
If Bayes is generating FPs, yes.
--
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
-----------------------------------------------------------------------
False is the idea of utility that sacrifices a thousand real
advantages for one imaginary or trifling inconvenience; that would
take fire from men because it burns, and water because one may drown
in it; that has no remedy for evils except destruction. The laws
that forbid the carrying of arms are laws of such a nature. They
disarm only those who are neither inclined nor determined to commit
crime. -- Cesare Beccaria, quoted by Thomas Jefferson
-----------------------------------------------------------------------
11 days until the 45th anniversary of Apollo 11 landing on the Moon
