Source Trust Prediction looks like an interesting way to fight spam. Instead of looking at the contents of the messages, it looks at the ip address of the server sending the mail, and predicts the likelihood that the message is spam.
This should be a faster way to determine that a message is spam, especially since now content filtering really requires OCR to handle the messages where the text is rendered as a graphic.
[via Slashdot]
No comments:
Post a Comment