The VPOP3 spamfilter contains several rules which test for certain things, eg 'adult words', or entries in real-time blacklist databases, or whitelisted email addresses etc.
Each rule has a weight associated with it. This weight determines how important that rule is, and you can adjust the weights depending on your situation. For instance, if you are a pharmaceutical company, you may want to reduce the weight of the 'drugs' rule which checks for words like 'viagra' or 'prescription'.
Typically each rule will have two weight values associated with it. The first is if the Bayesian filter gives a spam probability of ⇐90%, and the second is if the spam probability is >90%
The spamfilter rules will give a value to the message for that rule, let's use 27 as an example. The Rule Weight then acts as a multiplier. So, if the weight for this example rule was '2', then the rule would contribute a value of 54 (27 x 2) to the final spam score for the message. The higher the spam score, the more likely it is to be treated as spam.
In the table, the numbers in brackets after the entered value are the default values defined in the spamfilter rule definition.