Rules
The lists
of black and white rules allow the plug-in to recognize spam and non-spam by specific combinations
of characters in message bodies and headers. These rules are an optional supplement
to the very effective trainable classification algorithm. But carefully compiled
lists of rules may improve the classification quality in those cases when the algorithm
cannot select a certain class for a message with high probability.
A rule consists of one ore more conditions.
It is applied when all conditions are met. It is also applied when at
least one of its Strong conditions is true.
White rules recognize non-spam and they are checked first. A message is classified
as non-spam with the rating of 0 if at least one white rule is applied to it.
Black rules recognize spam. A message is classified as spam with the rating of 100
if at least one black rule is applied to it.
If the "Learn the message" checkbox is selected, a message will
be marked as spam or non-spam depending on the type of the rule.
You can disable rules by clearing their checkboxes in the list. The
same feature is available for conditions.