Statistical filters don't need any maintenance,
which means that you as the user don't have to
set rules and configure blacklist servers etc.
Spammers find them much more difficult to circumvent
as the filter learns what is non spam from your
personal email which a spammer has no access to.
What
makes Spamatak different from other Bayesian anti-spam
filters?
It uses a Markov Model to calculate the Bayesian
probability of an email being spam or not spam.
This filter is sensitive to word and character
order, and to random text which improves its detection
rate when compared to naïve Bayesian filters.
Using a character based filter also allows Spamatak
to filter languages that do not use spaces to
delineate words.
I
don’t understand the concept of false-negative
and false-positive. False
Negative
This is an email that Spamatak classified as non
spam when it was in fact spam. So Spamatak was
incorrect (False) and the error was classifying
the email as not (Negative) spam.
False
Positive
This is an email that Spamatak classified as spam
when it was in fact not spam. So Spamatak was
incorrect (False) and the error was classifying
the email as (Positive) spam.
