[BUGS] trapping spam from headers

[Martin Barry]

$quoted_author = "jonathan michaels" ;
> 
> even so i don't think that rbl's are worth much beyond 'feel
> good' esentially . .anyway

some are better than others.

i use zen.spamhaus.org and dnsbl.njabl.org at SMTP session time. i.e. this
refuses connections from IPs on those RBLs.

other rbls are still useful but i don't trust them enough to use in this
regard so...

  
> > Currently, one of the best anti-spam measures is greylisting.  This
> > will get rid of most of the low-hanging fruit and saves you having to
> > accept the spam.  Some sort of Bayesian filter should sop up most of
> > what is left.
> 
> i have been doing some reading on Bayesian filtering of late
> and would like to do it that way, but havent found much reading
> material on this topic in teh freebsd/sendmail world, so i am
> thinking perhaps postfix might be a better way forward ???

use spamassassin. it bundles in a whole bunch of tests (including bayes and
rbl checks), adding scores for each one and then giving the email itself a
total score. it means that no single test can push an email to be flagged as
spam, so less reliable tests can be utilised.

e.g.
from an email in my spam folder at work...
  X-Spam-Report: BAYES_99=3.5, DOS_OE_TO_MX=2.75, FH_HELO_EQ_D_D_D_D=0.001,
  FH_HOST_EQ_VERIZON_P=0.001, HELO_DYNAMIC_IPADDR=2.426, HTML_MESSAGE=0.001,         
        RCVD_IN_SORBS_DUL=0.877, RDNS_DYNAMIC=0.1                                                                                                            
  X-Spam-Score: 9.7 (+++++++++)                       

http://spamassassin.apache.org

cheers
marty

-- 
with usenet gone, we just don't teach our kids entertainment-level hyperbole
any more. --Paul Vixie

http://www.merit.edu/mail.archives/nanog/2006-01/msg00593.html