Re: spam filter rules

From: Justin Higgins <justin_at_thruport.com>
Date: Wed 3 May 2000 13:49:39 -0400 (EDT)

ANN O'DONNELL WROTE:
> Does anyone know of any research that has been done
> about which rules are generally used to filter spam? I
> understand that if the word 'free' is included in the
> subject heading, that mail will probably be filtered
> out as spam - should we consider that if 'free' is in
> the email body, that might also get filtered out? Any
> thoughts / links to research much appreciated.

It is unlikely that most spam filters would do
filtering based on as loose a criteria as a single
common word like "free" being in the subject
heading....this would cause a real headache...i.e. an
innocent message like "are you free for lunch this
afternoon?" or something would be filtered out.

Many spam filters would only filter subjects out based
on some specific phrases, etc. "i.e. MAKE MONEY FAST",
etc. But in general, this kind of spam filtering has
been seen as pretty weak, and shouldn't be in use
nearly as much.

A lot of spam filtering is done now more by analyzing
the entire contents of a message, structure of a
message, etc. as well as comparing it with databases
such as ORBS to see if it comes from either a known
spammer, or has been sent through a service that is
known to be spammer-friendly. In addition, most spam
filters will often try to filter out fake e-mail
addresses, etc. as those are big indicators of
spam....i.e. if the message has ysad8yk_at_yahoo.com in
the subject, it's a big red flag for most spam
detection systems. They will also look to see if it
has been sent through multiple mail relays, etc. in an
attempt to try and mask where it was originally sent
from.

So in general, when spam filtering is in place, it will
normally be a combination of much more sophisticated
techniques than just looking for a keyword in the
subject.

Justin Higgins
ThruPort Technologies





Received on Wed May 03 2000 - 12:49:39 CDT