jeff covey wrote:

on Wed, Jun 18, 2003 at 10:50:22AM -0400%, Ross Vandegrift said:

ross> if you retrain Bayes with the messages in question

apparently, this thread has wandered far enough that my original
message has been forgotten.  it's here:

here's the question again:

    jeff> 75 spam messages have been recorded as ham.  i don't have
    jeff> the original messages, so i can't use sa-learn to unlearn
    jeff> them.  how can i keep the spam i've given the filter but
    jeff> wipe out all the ham, so i can start training ham from
    jeff> scratch?

AFAIK you can't. Either wipe out your Bayes database and begin again, or wait for the token bias to cure itself - which, given time, it will, of course. Depending on how much spam and non-spam you have in the database and the daily volume of mail, the bias for 75 false spam messages shouldn't take too long to decrease.



Tony Earnshaw

Working to get a life

This SF.Net email is sponsored by: INetU
Attention Web Developers & Consultants: Become An INetU Hosting Partner.
Refer Dedicated Servers. We Manage Them. You Get 10% Monthly Commission!
INetU Dedicated Managed Hosting
Spamassassin-talk mailing list

Reply via email to