"Monique Y. Herman" <[EMAIL PROTECTED]> writes:
> Well, I've only started using SA in the last few months. You say
> you've been using it for years ... perhaps it's become much more
> accurate with time?
I've never actually used SA since I don't generally trust heuristic
spam filters (hence TMDA). Gmane runs SA against the messages in
its groups by default, and crossposts the message to
gmane.spam.detected if it thinks it's spam. Since I read Gmane allot,
I've come across many SA false positives. Here is some of the logic
from some recent legit messages it has falsely tagged:
* 2.4 -- BODY: Contains at least 3 dollar signs in a row
* 1.9 -- BODY: message body is 25-50% uppercase
The message was a corrected encoded Japanese message in a Japanese
newsgroup.
* 1.5 -- BODY: Asks you to click below
* 2.4 -- Contains phrases frequently found in spam
[score: 13, hits: click here, com subject, com]
[with, for more, help you, here mailto, list]
[please, mailing list, more information, please]
[send, reply this, that you, when you, you for,]
[you need, your name]
The message was an ezmlm (a mailing list manager) subscription
confirmation request.
* 2.1 -- BODY: FONT Size +2 and up or 3 and up
The message was a CVS commit notice, and thus contained program source
code.
etc, etc.. Pretty amusing.
_____________________________________________
tmda-users mailing list ([EMAIL PROTECTED])
http://tmda.net/lists/listinfo/tmda-users
