[this is a repost. yesterday i posted this and it never showed up on the
list. lost on sf.net? sorry if it's a duplicate. -cb]

Chris Santerre Sent: Tuesday, November 04, 2003 4:42 PM

> THe idea is that you WILL stop more spam in total if you share them.
> Sorry, I got a little excited! Share your rules!

My rules are attached. Here's the story behind them. Over the past month I
got about 15 FN's in my personal account that all were clearly from the same
spammer. They all had the same format... red headline, black body, blue
link. They all obfuscated with periods and they all had a bunch of WEIRD
sentences at the bottom that were probably to throw off bayes.

I searched the web for these sentences and found them all on some high
school girl's web page. They were all vocab sentences for some homework
assignment. This spammer had clearly stolen the text of her page and was
adding random sentences from this list to his emails. I just made a list of
rules to match each sentence (Which are REALLY obscure and I doubt would
ever match ham. Admittedly, I didn't test them as I'm still working out my
Outlook corpus issues). I scored them super high in a fit of rage. You may
want to adjust scores so they are not so drastic.  They have worked great
for me in the past few days to catch a handful of messages that would have
slipped through otherwise.

YMMV but enjoy. And by all means: I know about this much || about regex so
all suggestions are MORE than welcomed. I don't even know if I escaped the
right characters.

cheers,
Colin

Colin A. Bartlett
Kinetic Web Solutions
www.kineticweb.biz

Attachment: spamsentence.cf
Description: Binary data

Reply via email to