This DLL accepts each line of an email body (plain text only), breaks it down into single words and then checks each word against a list of "bad words". It can then return a percentage of bad words in the email body. The bad word list is quite brief. This is user generated. The user teaches the dll which emails are good and the words from that go into a good word list. The user teaches the dll which emails are bad and the bad words go into the bad word list. Then the user "verifies" the bad word list - the dll strips words in the bad word list out if they're in the good word list. Hence only bad words are held and the user is assured of accurate spam identification.