Hi there,

I contributed some notes on the server-side implementation of SpamBayes we are using on our exim email server:

Filtering is achieved by processing each incoming email twice in exim:
The first time, the email is analyzed by the SpamBayes filter and tagged as "ham", "unsure" or "spam".
The email is then resubmitted to exim and the delivery itself takes place.
Emails tagged as "ham" or "unsure" are delivered to the user's mailbox whereas emails tagged as "spam" are not.

By default, a system-wide database is being used to filter the emails.
But a personalized database can be used to filter each user's emails.

A dedicated web interface, the "SpamViewer" allows each user to:
  • view how his/her recent emails were classified,
  • have recent "spam" emails (normally retained by the filter) delivered to his/her mailbox (false positives!),
  • use his/her corpus of recent emails to train his/her own filter.
Details of implementation can be found here: https://wiki.bri.nrc.ca/wiki/tiki-index.php?page=SpamViewer+installation

Cheers.
-- 
Christophe Deprez                         [EMAIL PROTECTED]
----------------------------------------------------------------------
Institut de Recherche en Biotechnologies / Biotech. Research Institute
6100 Royalmount, Montréal (QC) H4P 2R2, Canada     Tel: (514) 496-6164 


_______________________________________________
[email protected]
http://mail.python.org/mailman/listinfo/spambayes
Check the FAQ before asking: http://spambayes.sf.net/faq.html

Reply via email to