Spam filter is a piece of software that will block spam messages. What
is essential about a spam filter do not delete or filter out legit messages.
For most people it is very important not to loose messages that where
sent to them. To what degree the spam filter actually manage do this varies
from what spam filter software and what methods it uses to categorize
messages.
How the software actually blocks spam is very much up to the creators
of the software.
Content filtering - Rule based
One of the simplest way of filtering out spam is by checking the email
message for a set of keywords. This can be a list already defined from
the software creators that you can add your keywords to.
Content filtering - Bayesian filtering
Bayesian filtering is more advanced way of doing content filtering. You
will tell the spam filter what is a spam message and what is a legit message.
From this selection the bayesian algorithm will learn what content and
words and parts of sentences that are likely to be spam. The spam filter
will gradually improve to be more correct as you teach it with more spam
messages.
Blacklisting
Blacklisting is defining all mails coming from a list of servers as spam.
This method has been around a long time, but it do not seem to be as effective
as you would want. Studies of the the MAPS RBL show that this blacklist
only catches 24 % of all spam messages with 34 % false positives.
Whitelisting
Whitelisting is defining a list of email addresses, servers or words to
not be spam. A whitelisted message is very likely not to be spam. To send
spam you can exploit whitelisting by first getting an email address whitelisted,
then sending spam messages to it. But this is unlikely to happen on a
large scale since most spamming is done automatically. A spam program
that is not specially programmed to be able to perform the steps to whitelist
will not get through.