Self-learning Spam Detector
Spam Detector is a self-learning artificial intelligence
that calculates spam probability for all incoming
messages. To calculate spam probability ratings
Spam Detector uses a special algorithm (optimized
Bayes probability). The algorithm applies to the
results of statistical analysis of YOUR email
messages and YOUR meaning of wanted and unwanted
correspondence, that empower to determine spam
with very high accuracy.
When, How and What does it learn?
Spam Detector observes incoming correspondence
and extracts textual words, useful numbers and
other information (such as presence of long words,
absurd alpha-numerical combinations, domains and
much other) using advanced tokenization mechanism.
When you mark messages for deletion and click
Process, Self-learning Spam Detector learns the
deleted messages as unwanted and saves all the
received info of unwanted messages into the 'spam'
dictionary. Info of other messages goes into the
'good' dictionary.
If you delete a message that have been learned
as 'good', the intelligent self-learning Spam
Detector unlearns it and then learns it as 'spam'.
When you exit from the program, Spam Detector
learns all messages that left in your mail boxes
as 'good'.
You can see word and dictionary statistics via
Spam Detector tab of the Options menu (Tools >
Options).
Learning takes time
As you know learning takes time. Spam Detector
always learns. It even doesn't show you spam probability
until it learns 20 'good' and 20 'spam' messages
to not confuse you with its initial naive forecasts.
But with every new learned message Spam Detector
improves its detection abilities and will tell
you more truthful spam probability rating for
incoming email messages.
Self-learning Spam Detector is one of an excellent
Mail Box Dispatcher features that help you to
filter spam without downloading it.
