Black words
The black
words are effective for filtering spam by subjects of the messages. This method
is
especially convenient for filtering spam on the server, because it requires message
headers only.
Every word in a list of black words has a spam ratio. It is an integer that equals
to a number of messages with this words in subject, marked as spam. The plug-in
is optionally updates a list of black words automatically on training. If some
of the black words will appear in normal messages, then it will be deleted from
a list.
The plug-in counts the black words with spam ratio greater or equal to defined
value in settings.
By default the message is recognized as spam if it has at least one black word
in subject, with a spam ratio that is large enough. If you raise the minimum number
of black words, required for recognizing message as spam, then the probability
of mistakes will be lower.
A list of black words can be edited manually or updated automatically on learning.
You can export and import a list of stop-words as UTF8 strings.