Skip to main content

Scores

As mentioned in the tags and scores section, the spam filter's analysis of incoming messages yields a series of tags, each with a corresponding score. To determine the email's overall classification (spam or ham), scores associated with these tags play a crucial role.

The configuration of these scores is done in the spam-filter.list.scores setting. Each entry in this list denotes a tag and its corresponding action or score. The format used is:

TAG [score|"discard"|"reject"]
  • Score: Tags can be assigned either a positive or negative score. A positive score suggests that the presence of this tag increases the likelihood of the message being spam. Conversely, a negative score indicates that the tag's presence makes the message more likely to be legitimate or ham.
  • Discard and Reject Flags: Instead of a numeric score, tags can also be assigned one of two special flags:
    • "discard": If an email contains a tag with this flag, the entire message will be discarded, meaning it won't reach the recipient's mailbox.
    • "reject": When an email is found with a tag having this flag, the message will be rejected. Typically, the sending server will be notified of this rejection.

For example:

[spam-filter.list]
scores = {
   "BAYES_HAM" = "-3.0",
   RBL_SPAMHAUS_DROP = "7.0",
   "SPAM_TRAP" = "discard"
}

In this example, BAYES_HAM = -3.0 indicates that if the Bayesian filter determines the email to be ham (not spam), a score of -3.0 will be applied, reducing the overall likelihood of the message being classified as spam. RBL_SPAMHAUS_DROP = 7.0 indicates that, if the email's sender IP is found on the SPAMHAUS_DROP list, a score of 7.0 is added, increasing the chance of the email being categorized as spam. And finally, SPAM_TRAP = discard indicates that if the message triggers a spam trap, the action taken is to discard the message, preventing it from reaching the intended recipient.