Currently, about 30% of the "spam" that sneaks through our setup of spamassassin 3.0.4 is flagged as RAZOR2_CHECKED, but still scores below 5.0 ( it varies from 2-4) . We have the rules_du_jour script running updates regularly, and bayes is trained.

Does it make sense to "bump up" the RAZOR2_CHECKED score to 5-7 to get this portion of spam, yet allow those emails with enough "non- spam" characteristics that might still be flagged as RAZOR2_CHECKED to trim the score down to hopefully <5 if the email is actually "not" spam?

Currently, the avg spam score is around 16, and the avg non-spam score is around -3.5.

Reply via email to