are you saving it with the signature, or the original message? we
need to see some debug output to see what's going on.
Jonathan
On Oct 23, 2005, at 3:23 AM, Guillaume Laurent wrote:
> On Sunday 23 October 2005 01:54, Jonathan Zdziarski wrote:
>
>> Are we still in Markov land, or are we talking about the standard
>> graham-bayes implementation?
>>
>
> No, I've only been speaking about Markov here.
>
>
>> Markov usually learns very quickly, and
>> as evidenced by your testing it only needed to retrain one message in
>> that manor.
>>
>
> This is not my experience. I take a false positive, save it as a
> seperate
> message, retrain it as innocent with
>
> dspam --class=innocent --source=error < msg
>
> then I check if dspam has fixed its mistake with
>
> dspam --stdout --deliver=spam,innocent < msg | less
>
> and in the output I see a 2nd set of dspam headers which still mark
> it as
> spam. Retraining it again doesn't seem to change anything.
>
> --
> Guillaume.
> http://www.telegraph-road.org
>
>
>
Received on Sun Oct 23 08:54:00 2005
This archive was generated by hypermail 2.1.8 : Wed Nov 02 2005 - 00:00:01 EST