Using Live Spam Beater (LiSB) Framework for Spam Filtering during SMTP Transactions

Abstract

This study introduces the Live Spam Beater (LiSB) framework for the execution of email filtering techniques during SMTP (Simple Mail Transfer Protocol) transactions. It aims to increase the effectiveness and efficiency of existing proactive filtering mechanisms, mainly based on simple blacklists. Since it implements some proactive filtering schemes (during SMTP transaction), when an email message is classified as spam, the sender can be notified by an SMTP response code as a result of the transaction itself. The presented framework is written in Python programming language, works as an MTA (Mail Transfer Agent) server that implements an SMTP (Simple Mail Transfer Protocol) reverse proxy and allows the use of plugins to easily incorporate new filtering techniques designed to operate proactively. We also include a plugin to perform proactive content-based filtering through the analysis of words included in the body of the email message. Finally, we measured the performance of the plugin and the framework (time required for operation and accuracy) obtaining values suitable for their use during SMTP transactions.

Keywords

• spam filtering
• proactive spam filtering
• SMTP transactions
• machine learning
• profile-based filtering
• anomaly detection