| Re-education problem [message #526] |
Wed, 13 July 2005 15:43 |
psynnott
Messages: 2
Registered: July 2005
|
Junior Member |
|
|
Hello,
I'm having a problem with re-educating the Bayesian filter to recognise genuine mail that has previously been identified s spam. I've followed the instructions for educating with Exchange mail, importing to mbox before running sa-learn etc, but many of the messages that are being identified as spam are messages that I have already "sa-learned" as spam two or three times. Am I missing something?
Thanks,
Paul
|
|
|
|
| Re: Re-education problem [message #527 is a reply to message #526] |
Thu, 14 July 2005 14:49  |
support
Messages: 925
Registered: April 2004
|
Senior Member |
|
|
psynnott wrote:
> I'm having a problem with re-educating the Bayesian filter to
> recognise genuine mail that has previously been identified s
> spam. I've followed the instructions for educating with
> Exchange mail, importing to mbox before running sa-learn etc,
> but many of the messages that are being identified as spam are
> messages that I have already "sa-learned" as spam two or three
> times. Am I missing something?
A typo, or a misunderstanding? You have to learn false positives as HAM, using the --ham option of sa-learn. Learning them as spam has the contrary effect.
Customer Support
Byteplant GmbH
|
|
|
|
| Re: Re-education problem [message #528 is a reply to message #526] |
Tue, 19 July 2005 12:21 |
psynnott
Messages: 2
Registered: July 2005
|
Junior Member |
|
|
Apologies, that was indeed a silly typo. The good mails were being re-learned as ham, rather than spam. The problem was actually due to a different silly oversight and (re)education is now working correctly.
Thanks,
Paul
|
|
|
|
Members
Search
Help
Register
Login
Home
