| PrismEmail
's 'normal' filters utilise
traditional filtering methods. These methods are utilised
across the industry, and are well known by the public and
spammers.
These filters are largely based on looking for keywords or
phrases. For example, a message that contains the word "SEX"
or "PORNO" is probably a pornographic spam that can be rejected.
Likewise, known websites belonging to spammers or that are
referenced by spammers can usually be rejected as spam as
well. This simple filtering approach to spam is easy to implement
and will reduce spam significantly--good, well-maintained
keyword filters can catch around 90% of spam.
The problems with these simple word or phrase filters are:
- Spammer adaptation. Spammers will look for other
ways to express the same idea. We've all seen it. The word
"VIAGRA" is expressed as "V!AGRA". Spammers do this so that
simple spam filters looking for "VIAGRA" will not filter
their message.
- Reactive appoach. These kind of filters are reactive.
Someone generally has to receive an offending spam and recognize
a potential filter before the filter works. That means spammers
trying new variations to words or phrases will get a few
spams through before a filter is added to block them. Then
they'll just change their text again to get around the new
filter. It's a never-ending battle.
- False positives. As more and more filters are added
the possibility of a "false positive" increases. At some
point you will have so many filters that you inadvertently
block a legitimate message because it contained some word
that is usually used in spam. For example, "SEX" is commonly
an indicator of spam, but it is entirely possible to have
a conversation with someone that mentions that word without
it being spam.
- Everyone is different. The fact is, everyone is
different and everyone's email is different. Language or
terms used by one person might be considered vulgar and
a clear indication of spam for others. Many people might
be able to filter on "4-letter words" as an indication of
spam while others might use such language in their everyday
email. Since everyone is different it makes it very hard
to agree on what terms should be filtered.
These limitations are why we offer you
Bayesian Filtering in addition to the Normal filtering.
Find out what normal
filters we offer
|
