At 20:08 2004-02-03 +0100, Peter Rosa wrote:
How is it possible, the Internet providers running antispam/antivirus
protection SW, have accuracy about 99 % with their anti-spam rules ?
Hmm, where do you get this figure, and what's the correlating FALSE
POSITIVE figure (that indicates the amount of legitimate email that gets
trashed as if it were spam).
Does anybody know EXACT SEQUENCE what software/daemons/rules to setup on
mailserver to finally get spam out ? Or is it still trial / failure method
for anybody who need it ?
If there were some magical "exact sequence" then there wouldn't be so much
spam, would there?
I use weighting in my own recipes - this significantly reduces false
positives, by assigning a variable weight to different characteristics
found in spam (a single Received: header for foreign sender for instance),
and after all the tests have been run, tallies up the weight and determines
if it is above the threshold. This allows for individuals to send a
message From: hotmail, when they're not actually USING hotmail to send it
(not entirely uncommon), but that, taken into consideration with a number
of other factors, weighs in - only a spammer will have an abundance of
spammy characteristics, and thus only spam gets flagged (my only exceptions
have been spam-like message content, and foreign character sets, both of
which I assign a high score to). Search the procmail list archives for
"SPAMMISHNESS" to read more. The approach isn't really all that different
than a SpamAssassin score.
Of course, I also use DNSBLs (including some which I administer for local
use only), and they avoid much of the traffic and processor load associated
with dealing with spam. No doubt, some of those DNSBL-rejected messages
are legit now and again, but those users have alternate methods to contact
people, and at DNSBL time, they at least receive a bounce (which for my
servers, includes a URL to a page with explanations and further links).
Sorry for my english, I still can not speak as I would like to.
Your English is superior to my furrin' language skills. <g>
---
Sean B. Straw / Professional Software Engineering
Procmail disclaimer: <http://www.professional.org/procmail/disclaimer.html>
Please DO NOT carbon me on list replies. I'll get my copy from the list.
_______________________________________________
procmail mailing list
procmail(_at_)lists(_dot_)RWTH-Aachen(_dot_)DE
http://MailMan.RWTH-Aachen.DE/mailman/listinfo/procmail