Enhanced Topic-based Vector Space Model for semantics-aware spam filtering
Expert Systems with Applications: An International Journal
Word sense disambiguation for spam filtering
Electronic Commerce Research and Applications
Reversing the effects of tokenisation attacks against content-based spam filters
International Journal of Security and Networks
| Hi-index | 0.00 |
The underlying problem with spam detection is how to define spam. Simon Heron of Network Box examines current techniques for defining and detecting spam and how spamming itself has evolved in order to avoid detection. From early whitelisting and blacklisting systems through to the most up-to-date defence solutions based on relationship management, a vast array of detection techniques have been deployed to try to fix the spam problem. Spam is a fast-moving target, however, and none of these techniques work in isolation. Spammers' financial rewards mean that they will always work to keep ahead of protection systems. The only real solution is to combine many of the techniques laid out here, that adapt and 'learn' from new threats, and to continuously develop new solutions. The underlying problem with spam detection is how to define what spam is. One person's spam is another person's newsletter. Although definition may be easy for end users (''I don't want it and I didn't ask for it.''), it isn't that simple legally. In the USA, the spam definition in the CAN-SPAM Act of 2003 runs to 21 pages.^1 Definition of spam is vital to effective spam detection. This article examines techniques used to define and detect (not avoid) spam and also how spam has evolved to avoid detection.