Strength of MailScanner vs w/SpamAssasin

[Chris W. Parker]

Mark Steudel <mailto:msteudel at PANICWARE.COM>
    on Thursday, April 15, 2004 5:03 PM said:

> I was wondering if MailScanner can be as tough on spam as
> MailScanner/wSpamAssin?

hmm... that's a bit of a strange comparison. you're assuming MailScanner
Plain and MailScanner w/ Spammassassin. :)

spamassassin is just another program that MailScanner utilizes to combat
spam. if 45% of your daily mail is found to be spam without using
spamassassin then that number will only go up once you start using
spamassassin. (i'm no mailscanner expert in the slightest so i didn't
even know it was possible to use mailscanner without spamassassin.)

> I've also heard people talk about bayes but am unsure of exactly what
> it is and whether or not I can implement it without spamassasin.

bayesian filtering (aka bayes) is a method by which known spam and known
non-spam (called ham) are analyzed. bayes uses the words (called tokens)
within each email and gives it a score. the idea being that a known spam
message will consistently have different tokens than a known ham
message.

for example, the word pe*ni_s (without the * and _) is much more likely
to appear in a spam message than in a ham message. this token is given a
higher score than say the word "cat". so after a while you have a *HUGE*
list of words and scores for each of those words. when a new message
comes in, it's tokens (words) are added up and the resulting score is
used to determine whether or not it's spam.


hope this helps!

chris.

p.s. everyone, if i'm wrong about something please correct me.