Does "Edit Header Field" impact Baynesian training?

Does "Edit Header Field" impact Baynesian training?

Postby ehavemann » Sat Jan 19, 2008 12:44 am

I am trying to improve the accuracy of my spam filters. When certain rules are triggered, I want to indicate as such by inserting a new custom line in the header. This allows me to examine the email source after delivery to see which events are logged in the email header. When I go to add that email to either a "Good Mail" or "Spam" folder for subsequent Baynesian filter training, does the presence of these new headers impact the scoring and evaluation of the email?
ehavemann
 
Posts: 26
Joined: Fri Dec 14, 2007 6:15 pm

Re: Does "Edit Header Field" impact Baynesian training?

Postby Code Crafters » Mon Jan 21, 2008 11:51 am

Altering the subject to "<SPAM> ####subject####" where ####subject#### will be the original subject in this content filter action will have some but very minimal effect on Bayesian training. All words are used as tokens with good and bad counts in the Bayesian database. If a word appears in one more than the other it may be used for Bayesian scoring but only the strongest words that appear nearly always only in SPAM / non-SPAM will be used for scoring and for this reason it probably will never use these words anyway and will usually more use other parts of the header or body of the mail that are more recognisable with SPAM mails always. Therfore, in short, it's fine to use these and they won't adversely affect your Bayesian training at all.
Code Crafters
 
Posts: 933
Joined: Mon Sep 10, 2007 2:35 pm


Return to General

Who is online

Users browsing this forum: No registered users and 15 guests

cron