On Dec 4, 2003, at 8:52 PM, Mikael Byström wrote: > I just installed SpamSieve 2.02 Demo and fed it 300 spams and 200 or so > good mesages before I used it for new messages resulting on over 4000 > lines in the corpus. It mistakenly let trough about 69 messages (127 > good > messages arrived, som on accounts not affe) from an account with 99% > spam. I then selected them and ran the "Add Spam" script. > > Yet, when I check statistics it says that it filtered 36 good messages > and 2 spams (both accurate), 0 false positives and only 1 false > negative. > Shouldn't the statistics function regard the 69 messages as false > positives? What am I missing here?
The 69 messages are false *negatives* since they're spam messages mistakenly thought to be good. As to why the statistics say only 1, that could be because the messages are duplicates. Or it could be that SpamSieve didn't actually predict that 68 of them were good; maybe another filter stopped the processing before the SpamSieve filter executed. To tell for sure, I'd have to see your SpamSieve Log file. Mail it to me at <[EMAIL PROTECTED]>, if you like. -- Michael Tsai <http://www.c-command.com>
