> Seems my message got no anwser ... It's forth on my list of ones to do. It's a busy time at moment.
> some email messages (evidently spam) have this structure (One example > received this morning) - The last part of the message contains some > words completly mixed hasardous : [...] > I don't want to add these message in the database of SpamBayes cause I > think there can be words in the last part of message that I don't want > to STOP!We receive this type of message more and more often > these days... > > I don't know if this as been discuted here before, and if there is an > answer for stopping these type of emails messages? This sort of thing is commonly called "word salad". If you continue training on the messages as normal (i.e. just unsures and mistakes) it shouldn't effect classification results at all. Many of these words will have never been seen in ham, so will be correctly spammy, and those that have will just be less strongly ham. A ham message should easily have enough ham clues to still be classified correctly. All in all, it's not really a problem. If you actually find that you're getting classification errors as a result, please send us a copy of the spam clues for a message that gets incorrectly classified (before training). =Tony.Meyer -- Please always include the list ([email protected]) in your replies (reply-all), and please don't send me personal mail about SpamBayes. http://www.massey.ac.nz/~tameyer/writing/reply_all.html explains this. _______________________________________________ [email protected] http://mail.python.org/mailman/listinfo/spambayes Check the FAQ before asking: http://spambayes.sf.net/faq.html
