On Wednesday, January 5, 2005, 10:55:35 AM, MCSE wrote: SSMOE> Just wanted to chime in here that the original changes seem to have helped SSMOE> my rulebase as it has gone down from 11,355,504 bytes to 4,871,976 bytes but SSMOE> I believe I'm still at the 'standard' rulebase strength.
This is true - only ineffective rules have been removed. It was always the case that ineffective rules would be removed, but that process was less efficient before. As a result of the new process, all rulebases should be somewhat smaller and all rulebases will become more effective (relative to rule strength setting) over time. One of the changes that have been put in place is that we now have a differential rule boost analysis in place. This process uses messages that hit our spamtraps and compares the scan results for the standard rulebase and the "fullbase" which includes ever rule ever put in our system. If a message arrives at a spamtrap, passes through the standard rulebase, but is captured by the fullbase, then the rules involved in that message are "boosted" by recording an accelerated hit rate. This pushes the rule strength of these rules above the 1.0 threshold defined for the "standard" settings. This mechanism tends to push marginal rules back into the mainstream as long as they are still viable as measured by our spamtraps. In the past these rules would have normally remained in the 0.1 to 1.0 range along with other much less effective rules. (Sorry for writing another book...) _M This E-Mail came from the Message Sniffer mailing list. For information and (un)subscription instructions go to http://www.sortmonster.com/MessageSniffer/Help/Help.html