Posted Feb 23, 2006 17:30 UTC (Thu) by vmole
In reply to: User interfaces
Parent article: The Grumpy Editor's guide to bayesian spam filters
The problem with feeding back the message the user sees is that by that time the user sees it, Outlook has already destroyed the headers - there is no
standard way to extract an undamaged, as-delivered message from Outlook and send it back to the anti-spam system.
Dspam work around this by inserting a hash into the ongoing message. When it's sent back for "rescoring", it extracts the hash code and then just reassigns the tokens associated with that hash. The downside is that the users wonder about that weird code in all their e-mails.
Another method is used by Maia Mailguard, which just keeps a cache of all the mail. The user needs to go to a web interface to mark false negatives, but it also let's them look through the spam cache to find and release false positives. (Maia is "just" a front-end to amavisd/spamassassin).
to post comments)