Uma análise comparativa entre técnicas de detecção de spam com base na captura do tráfego da rede

Abstract

With the popularization of the Internet and the use of e-mail, the sending of spam has become a common pratice. Several techniques of e-mail spam detection exist and have been studied as countermeasures to spam and to the harm caused by them. Network administrators constantly analyze network traffic through trace analysis. This work has the objective of performing a comparative analysis of spam detection techniques, based on the capture of the network traffic. In this work we create traces using the ENRON dataset and apply blacklists, keywords and content analysis using Naive Bayes and SVM techniques in the traces evaluation. We conclude that it is possible to perform spams detection from network traffic capture and that content analysis technique using SVM presents the best accuracy.