When (re)training POPFile to improve classification, it would help very much to see, for a wide swath of messages, how they would now classify.
I realize this might be an 'expensive' (cpu) column to generate all the time, but it would be of huge value while (re)training the app!!