http://bugzilla.spamassassin.org/show_bug.cgi?id=3847





------- Additional Comments From [EMAIL PROTECTED]  2004-09-30 21:01 -------
Subject: Re:  Consider removing RFCI tests from SA 3.0

RFCI tests are more than accurate enough.  I don't think we want to
decide to fundamentally reject our own model of "we can use somewhat
inaccurate tests by assigning weights".  We can and we do.

OVERALL%   SPAM%     HAM%     S/O    RANK   SCORE  NAME
 766289   506205   260084    0.661   0.00    0.00  (all messages)
100.000  66.0593  33.9407    0.661   0.00    0.00  (all messages as %)
  1.473   2.2123   0.0346    0.985   0.66    1.46  DNS_FROM_RFC_BOGUSMX
 10.564  15.9113   0.1553    0.990   0.59    1.14  RCVD_IN_RFC_IPWHOIS
  1.466   2.1792   0.0773    0.966   0.56    0.49  DNS_FROM_RFC_WHOIS
  4.549   6.7726   0.2207    0.968   0.49    0.00  DNS_FROM_RFC_DSN
 15.276  22.8917   0.4537    0.981   0.44    1.38  DNS_FROM_RFC_POST
 21.889  26.9543  12.0292    0.691   0.32    0.37  DNS_FROM_RFC_ABUSE

Given an equal 50/50 mix of spam and ham, When POST hits it is spam
98.1% of the time.  We'd be nuts to not let the perceptron try using it.
And the perceptron likes it too!  (It doesn't like DSN, though.)

If you want to be more productive, I suggest trying to come up with a
way to reduce the FP rate, perhaps in discussion with the RFC-Ignorant
maintainer, or via some other means.

So, in conclusion, I'm still very much -1 on removing the tests, but I'm
open to any improvements anyone can come up with.  Maybe someone should
look at those ham hits and see whether RFC-Ignorant should change their
policies?  I don't know, but this bug seems like the wrong place to
discuss it.

Cheers.





------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

Reply via email to