A Theoretical Taxonomy and Analysis of Anti-Spam Technologies
Charles Duan · 2004
The problem of unsolicited, unwanted e-mail messages, otherwise known as spam, is growing to uncontrollable proportions, and there is a clear need for directed research in spam prevention and control. Current research has identified many different schemes for controlling spam, but the general direction of this research has been in improving individual schemes without consideration for the various strengths and weaknesses of those schemes as compared to others. The goal of this paper is to identify and categorize the various antispam technologies. This categorization will differ from more common ad-hoc anti-spam taxonomies in that it will attempt to cover the entire theoretical space of anti-spam solutions rather than merely those that are implemented or those that are popular. Using such a taxonomy, then, we can investigate the relative strengths and merits of each category. We will demonstrate that many commonly used solutions, such as content filtering, public-key authentication, and computational-cost based prevention, are vulnerable to various forms of attacks and thus ineffective in this problem space. Some of the key conclusions are that a system successful in preventing spam requires sender intervention, and that intervention must be in the form of either identity verification or payment of some cost. This research will aid future studies in anti-spam technology, steering them to greater efficacy.