[spambayes-bugs] Spambayes repeatedly classifies messages frommailing list as SPAM

Tim Peters tim.one at comcast.net
Thu Sep 11 13:46:11 EDT 2003


[Ryan Malayter]
> What exactly are '*H*' and '*S*' then? I thought they represented the
> ratio of incoming ham to spam...

spambayes creates two scores internally, one for ham (*H*), the other for
spam (*S*).  The final score spambayes delivers is a combination of those
two:

    final score = (1 + *S* - *H*) / 2

*S* and *H* both range from 0 to 1 inclusive.  The combined score is
basically *S*-*H*, linearly rescaled so that it too always lies 0 through 1.

The number of ham, number of spam, and ratio of ham/spam trained on can't be
deduced from them, and both typically differ between any pair of messages
you look at.




More information about the Spambayes-bugs mailing list