[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: spamassassin updated on sanpietro
From: |
B. W. Fitzpatrick |
Subject: |
Re: spamassassin updated on sanpietro |
Date: |
Sun, 17 Aug 2003 09:11:49 -0500 |
Noel Cragg <address@hidden> writes:
> The bayesian filtering routines are disabled by default, because
> folks on the net at large haven't yet come to consensus whether it
> works out of the box or not.
I started using the bayesian routines about 3 months ago, and I took
the time to train spamassassin using about 2300 messages from my spam
folder, and about 2500 messages taken from my inbox, my personal mail,
and all the mailing lists I belong to (including svn/cvs commit
lists), and it's turned out to be uncannily accurate. The only
problem is that spamassassin inexplicably doesn't run certain mails (
< .5% ) through the bayesian filter, and those wind up in my inbox.
Anyway, I tuned my prefs to give a lot of weight to the Bayesian
routines.
Here's my ~/.spamassassin/user_prefs:
required_hits 6
score BAYES_99 9.0
score BAYES_90 8.0
score BAYES_80 7.0
score BAYES_70 6.0
score BAYES_60 5.0
score ADVERT_CODE 3
score USER_AGENT_PINE -0.5
score USER_AGENT_MOZILLA_UA -0.5
Thanks for updating, Noel.
-Fitz