bogominitrain.pl -f(n)

[Boris 'pi' Piwinger]

Hi!

Today I observed for the first time that a message was used
more than once for training in bogominitrain.pl -f. I
investigated and found that the same spam (received twice)
once made it to the spam training set and once to the ham
set. So not using -n can be a very useful tool to find those
mistakes.

Recall that -s can be used to create mboxes with the
messages used for training. Mistakes in the training set
likely show up there.

pi