All mails have spamicity=0.5200000
Robin Bowes
robin-lists at robinbowes.com
Mon Dec 12 17:05:51 CET 2005
Matthias Andree said the following on 12/12/2005 15:13:
>
> 1. -Ns is only correct if the spam message(s) had previously been registered as
> "good" (ham). If it was "unsure", use only "-s" without "-N".
OK.
> 2. you absolutely must train bogofilter what "good" (ham) messages look like,
> and the ratio of ham and spam messages you trained should roughly match the
> ratio of ham vs. spam you get. In doubt, train more ham messages to avoid false
> "spam" tagging.
>
> Looking at your statistics, it seems that bogofilter has not been trained for
> "good" mails at all.
Yes, it appears that way.
What was *supposed* to happen, is that all mail is treated as "good"
initially. If I then drop it in the Spam/Undetected folder, it is
re-classified as "spam".
So, it looks like I need to modify my bogofilter command line to not do
tri-state classification.? "-o ,0" should so the job, if my perusal of
the man page is correct.
R.
More information about the Bogofilter
mailing list