Mercury/32 provides several methods you can use to control spam. Here are some of my favorites:
POPFile - When run as a POP3 or SMTP proxy, POPFile accepts email BEFORE Mercury gets it. A Bayesian classifier that attempts to classify mail into one of several "buckets" based on the content of the message. Whenever it makes a mistake, you correct it using a web interface. This simple process teaches POPFile how to classify messages. It is incredibly accurate, easily attaining over 99.9% accuracy. For an explanation of Bayesian filtering, and some idea of how accurate it can be, read this page:
http://www.paulgraham.com/antispam.html. A technical discussion of how bayesian filtering works:
WikiPedia:Bayesian_filtering. If you want to use DNSBLs and POPFile, then you should use the POPFileD daemon for Mercury/32.
POPFileD - POPFileD is a daemon for Mercury/32 that integrates POPFile message classification into the Mercury/32 Core process, therefor permitting Mercury to do front-end DNSBL lookups and Transaction Filtering.
SpamWall - SpamWall is a daemon for Mercury/32 that tags messages as SPAM using a Bayesian filter. This program has received many good reviews from its users.
Spamhaus,
NJABL, and
ORDB. IP addresses make it onto the list because they are open relays, open proxies, known spam sources, or whatever other criteria the list owner decides to use. When choosing DNSBLs to use, make sure you know what criteria the list uses for listing IP addresses. Some lists are very aggressive, listing wide ranges of IP addresses for little or no reason.| |
|
|---|---|
| MailWiki | February 24, 2005 10:11 am |
| Mercury32 | June 28, 2006 8:39 am |
| MercuryDaemons | April 11, 2006 9:10 am |
| MessageProcessingFlowChart | May 12, 2006 10:11 am |