Mitigating the Impact of Spams by Internet Content Pollution

In recent years, there has been a steep rise in the amount of unsolicited-emails (spams) [11]. Such mails overwhelm users’ mailboxes, consume server resources and cause delays to mail delivery. Many techniques [2, 10, 12, 5, 13] have been used for mitigating spams. Despite the plethora of schemes proposed, all of them have the cardinal problem ofalse positives which compromises the reliability of emails. Furthermore, many of such schemes are plagued with security, privacy, deployability and transparency issues. In this project, we propose a new spam-mitigation approach that is orthogonal and complementary to the previous schemes: it reduces the spams reaching the mailboxes of real users by misleading spammers into spamming non-existing mailboxes. To send spams, spammers need a set of victim addresses ( V ). From the perspective of spammers’ resource utilization, it is imperative that th e setV consists largely of valid addresses. It has been observed that to create the set V , spammers primarily use two techniques: (1) crawling the Internet (homepages, newsgroups) [9], and (2) guessing email addresses with the hope of hitting on valid ones [11]. In this paper, we (1) hypothesize that majority of spams reaching a user’s mailbox is because the user’s address is harvested; (2) perform experiments to confirm hypothesis; and (3) propose a simple, false positives free scheme that mitigates the impact of spam on individual mailboxes by poisoning the address harvesting of spammers.