03-02-2004, 02:32 PM
|
|
|
Confirmed User
Join Date: Nov 2002
Location: Belgium
Posts: 7,383
|
Quote:
Originally posted by raymor
Let me see if I understood you right.
You got 360 spams. Their filter detected
300 of those spams, or 83%, and let 17% though?
If that impresses you, take a look at one of the Bayesian filters.
There are several free ones and commercial ones.
Using Bayesian filtering, I get about a 98.6% detection rate,
so my filter would miss 5 out of 360 spams rather than 60.
So far I've had one false positive (out of roughly 20,000 emails).
That false positive was a hotmail using who wrote the message
in all caps, used a lot of cuss words, and wasn't talking about business.
(With bayesian filtering, the filter learns which words indicate
a legit email, so if they mentioned words that my legit emails
have in them, like "script", "security", or even
"GFY" the filter would factor that word in, suggesting that
it may be a legit email.
|
can you recommend me one? the one you use maybe? sounds intresting! |
|
|