Re: Learning and verifying ham

On Dec 7, 2007 10:13 PM, Matt Kettler [EMAIL PROTECTED] wrote: Asif Iqbal wrote: Hi All I took a message and learned it as ham like this cat email-with-headers | sa-learn --ham Now I should expect the exact same email to be considered as ham. Correct? No. You'd expect the

Re: Learning and verifying ham

Asif Iqbal wrote: On Dec 7, 2007 10:13 PM, Matt Kettler [EMAIL PROTECTED] wrote: Asif Iqbal wrote: Hi All I took a message and learned it as ham like this cat email-with-headers | sa-learn --ham Now I should expect the exact same email to be considered as ham. Correct?

Re: Learning and verifying ham

On Fri, Dec 07, 2007 at 02:34:51PM -0500, Asif Iqbal wrote: I took a message and learned it as ham like this cat email-with-headers | sa-learn --ham (Useless use of cat!) Now I should expect the exact same email to be considered as ham. Correct? Not necessarily. You should expect that

Learning and verifying ham

Hi All I took a message and learned it as ham like this cat email-with-headers | sa-learn --ham Now I should expect the exact same email to be considered as ham. Correct? But it does not. When I pipe it through spamassassin like following it shows it as spam cat email-with-headers |

Re: Learning and verifying ham

Asif Iqbal wrote: Hi All I took a message and learned it as ham like this cat email-with-headers | sa-learn --ham Now I should expect the exact same email to be considered as ham. Correct? No. You'd expect the bayes score to go down, but that alone might not be enough. But it does