How are the scores assigned?

In [SpamAssassin] 2.x, the scores are assigned using a genetic algorithm (GA).  In [SpamAssassin] 3.x, the scores are assigned using a neural network trained with error backpropagation (\["Perceptron"\]).  Both systems attempt to optimise the efficiency of the rules which are run, while also minimising the number of false positives and false negatives. More information can be found on the 'Tests' page. Note that you can help this system by providing statistics on your mail spool.

Some DNS blacklist rules are distributed with scores of 0. These generally request or require payment, and as such are disabled by default. Feel free to enable the lookups, if you've paid for them.

A score of 0 will stop a rule from being run.

The scores are assigned using a neural network trained with error back propagation (Perceptron). Both systems attempt to optimize the efficiency of the rules that are run in terms of minimizing the number of false positives and false negatives.

You can find all of the currently active rules and their scores in the Subversion repository under /trunk/rules or by downloading the latest published set using the sa-update tool. 

You can help this system by providing statistics on your mail spool via NightlyMassCheck and RescoreMassCheck.

Confusing scores

Scores for "learn" rules (example the various BAYES_?? rules) Note: Scores for "learn" rules, such as BAYES_*, that rate the probability that a message is spam, are scored using the same method. This can produce "confusing" scores, for instance, that have scores which seem incorrect (example BAYES_80 with a higher score than BAYES_99. There are a few reasons for this. 1) The score generation system does not understand that BAYES_* are ). This is due to the fact that rules are not related to one another, they're separate rules that need have separate scores. 2) More importantly, the higher the

Messages with high probability from a "learn" rule , the higher likelihood that the message also hit a bunch of will most likely match other rules. This lets the score generation system lower the "learn" rule score due to the inevitable false positive, while also still marking the message as spam via preventing false positives. The message still is recognized as spam due to the sum of all rule scores.

Some DNS blacklist rules are distributed with scores of 0. These generally request or require payment are disabled by default. Feel free to enable the lookups, if you've paid for them.

A score of 0 will stop a rule from being run.

In version 2.x, the scores are assigned using a genetic algorithm (GA).