But you were talking about extracting the bayes learning results some how instead of actually saving the entire email... then later you would train bayes based on the matches it made or something like that if i was correctly understanding you.