DSPAM: Single aliases for spam/ham reporting, opt-in/out-flags in the database
My mail server uses DSPAM (http://sourceforge.net/projects/dspam/) to filter incoming mail. I use the following set-up:
- spam@xxxxx for retraining false negatives.
- ham@xxxxx for retraining false positives.
- Virtual users in the dspam database.
- Opt-in/out flags from the database.
- User IDs in the X-DSPAM-Signature header.
This set-up means that any of my mail users can forward an incorrectly classified message using any of their aliases as the ‘From’ address and DSPAM will work out the correct user from the X-DSPAM-Signature header. I was pretty sure this set-up was working, until I upgraded to version 3.9.0 and the detection efficacy dropped off.
(more…)