Hi, I'm concerned that my bayes database may contain incorrect information. I performed a search on all of the messages in the quarantine, and pulled out the ones that contained BAYES_00 in their score. There weren't all that many of them, but enough that I want to investigate further. Simply deleting the database and starting over isn't really the best option.
Is it possible to "unlearn" the tokens in these messages from the database, and then re-learn them as spam messages? How should this really be handled? Thanks, Alex