Spam
Recently, I’ve been investigating spam filters.
I did use Cloudmark’s product SpamNet, which used a community to filter spam. The idea is that if enough people block a message, then everyone else that gets the same message also has it block.
Good idea, but spam has moved past that.
Next I read about SpamAssassin, thinking it might be useful to install at work. We get a lot o’ spam. In reading about SA’s Bayesian filter, I stumbled across the K9 POP3 spam filter. It stands in between your POP3 client (I use Outlook). You point your client at K9, request mail, it takes the request, contacts your server, downloads the mail, then processes it.
It parses all the tokens and adds them to a database based on whether or not the email is spam or not. So, you have to hand sort a number of meesages to build your database of spam tokens. I’m still in my first week of using it, but it’s doing a great job so far. I’m still getting a couple spams a day through the filter, but it’s almost eliminated the mail from the -owners list.
I’ve noticed that the messages that DO get through are putting random words at the bottom of the message. Supposedly, these words are typically found in a non-spam database and would lower the probability that a message is spam.
Enough on spam for now.
