Analyzing Words in Spam Emails
We recently did some analysis on our bayesian spam filter corpus (spam assassin token database), and came up with a list of words with a high spam/ham ratio.
By using the spam/ham ratio, and not the spam count, we came up with a better list of words to avoid. Most lists would have you avoid words like
here, but they are used so much in legitimate email, that they have a lot spam/ham ratio.
You might also like:
- Battling Comment Spam - January 31, 2007
- Trick or Treat - Web 2.0 Goodies for ColdFusion - October 31, 2006
- Spammers now using ASCII Art - April 21, 2005
- ReturnPath aquires BondedSender - April 12, 2005
- Another Trick for Avoiding Email Harvesters - March 20, 2005