Automatic mining of Threatening e-mail using Ad Infinitum algorithm
S. Balamurugan · 2008
We have proposed a new classifier named Ad Infinitum to identify e-mails containing terrorist threats. This Ad Infinitum is an enhancement of decision tree induction algorithm which will generate reliable rules to discriminate emails from potentially dangerous to those that are safe, with the help of agent. In the preprocessing stage, we have used deception theory which states that deceptive writing is characterized by reduced frequency of first person pronouns, exclusive words, elevated frequency of negative emotion words and action verbs. We have demonstrated the effectiveness of the proposed approach on a real world email corpus.