On the use of cross-domain statistics in spam-filtering
Against well crafted spam there is very little defence to be expected from the traditional ways of content scanning. For this reason the fight against spam and viruses is moving towards centralised anti-spam services that can use meaningful real-time statistics to recognise malicious content. Extensive use of techniques like honey-potting, statistical anomaly detection and cross-domain content analysis give these services a significant edge over stand-alone solutions in the market. The presentation will go into the advantages of several of these techniques and give insight in how they are used by Onlinespamfilter.