Improving Efficiency of E-mail Classification Through On-Demand Spam Filtering
Shafiya Afzal Sheikh, Mohammad Tariq Banday · 2020
E-mail systems have been witnessing a race between e-mail security researchers and spammers; wherein modern spam filtering systems have been built that constantly evolve and use modern learning techniques to filter spam and likewise spammers explore tricks and ways that to a larger extent bypass these spam filters. Once messages from spammers enter into inbox of a user, they can pose a significant threat, even if subsequent spam messages from these spammers are blocked by the filter. These include activity monitoring of e-mail addresses and subsequent e-mail address harvesting, phishing, spread of viruses and strengthening botnets. This study through experiments explores some commercial as well as corporate e-mail systems and finds this vulnerability a reality. The paper proposes an on-demand spam filtering mechanism which allows an e-mail client to notify its users about the presence of spam messages in their inboxes and moves them to junk folders as a solution to this vulnerability. Experiments with the proposed on-demand filtering mechanism have demonstrated its effectiveness in removing spam messages from inboxes and improve classification efficiency to classify subsequent spam messages from these spammers as spam messages.