Spam Mail Filtering through Dynamically Updating URL Statistics

This paper presents a unique spam mail filtering technique based on a deep analysis of statistics on URL’s included in various e-mails gathered from a laboratory in a university for about six months. Since the proposed mail filtering technique searches only URL’s in mail, the overhead introduced by searching all mail contents or black list utilized by many other mail filtering algorithms is significantly reduced. In addition, the proposed filtering technique dynamically updates URL list through client feedback, and the bias possibly introduced by selecting bad training mail set can be eliminated as the filtering process is progressed.