Sciam (Scientific American) on Spam

The latest issue of Scientific American has an article on stopping spam. Sciam has published prior articles on the spam issue, but this one is the most comprehensive to date.

It's a pretty nice overview of various technologies that are in use in the battle on unwanted email. It covers in detail various machine learning approaches such as naive Bayesian filtering. SpamButcher's own fuzzy-logic system is not entirely unlike a Bayesian system. In fact, it's been proven that Bayesian filtering systems can be simulated using fuzzy sets.

Apparently some new anti-spam efforts attempt to analyze image content within the spam. Algorithms to detect pornographic pictures have been under development for some time. This is the first I've heard of them being deployed as part of an email filter.

The SpamButcher spam blocker software examines information about images in messages, but never downloads image content from remote servers. Downloading an image could give a spammer the impression that the message was successfully delivered and cause the user to receive more junk mail.

The article also covers "upstream" efforts to prevent spam such as changing the protocols behind email transfers, and legal efforts. They detail the mechanisms behind systems like sender authentication, challenge response schemes and electronic postage. A diagram is presented containing a hierarchy of all the different spam fighting efforts. It's the first time I've seen client-side spam filters presented in context to efforts like sender-ID and anti-spam laws. In the end, it may be a combination of such efforts that take a 'byte' out of spam email.

Resource: Scientific American - Stopping Spam

Back