I have noticed recently that BAYES is giving me some funny results and i
had a look at what was causing it.
The emails concerned all come in to an account which does receive
probably 98% spam. As BAYES uses the received headers and from and to
headers it finds the actual email address many times, and adds it as a
spammy token. This is pretty bad, because this means that when a genuine
email comes into that account with very hammy tokens, they are just
cancelled out by the token which is the email itself which is ranked as
a very highly spammy token.
I don't want to ignore all received headers, as these can be really
helpful, but i would like to be able to specify which tokens are
ignored, and even remove them from BAYES if possible. Also is it
possible to ignore just certain Received headers (for example internal
mail transfers which happen to every ham and spam message and are
therefore not helpful to BAYES?
Are any of these things possible?
Am using SpamAssassin 3.1.2 on Windows 2003 server.
Thanks
Ben
