Kevin,
The statistical filter is checked only if all previous tests fail. This is done since performing a statistical analysis of an incoming message can be very process intensive and we try to avoid it if possible. If you have very good rules that catch a lot of spam, very little will get past them, and only whatever does is then checked against the corpus database.
In theory you could eliminate all other rules and start relying only on the statistical filter after it becomes active, but that is something we are still testing.
If you run the following query against the quarantne database you'll get an idea of what messages are being blocked by what rule (change the date/time as you see fit):
SELECT tblQuarantine.RejectID, tblRejectCodes.RejectDesc, COUNT(tblQuarantine.RejectID) AS Total
FROM tblQuarantine INNER JOIN
tblRejectCodes ON tblQuarantine.RejectID = tblRejectCodes.RejectID
WHERE (tblQuarantine.MsgDate > CONVERT(DATETIME, '2004-1-20 00:00', 102))
GROUP BY tblQuarantine.RejectID, tblRejectCodes.RejectDesc
Roberto F. LogSat Software
|