Header Based Email Spam Detection Framework Using Support Vector Machine (SVM) Technique

Abstract

Email spam is continuously a major problem, especially on the Internet. Spam consists of malicious malwares which attack user’s machine to steal information, destroy the user’s machine and trick the user into buying their products. Although spam detection or email spam filtering was developed, there is still a rising number of emails that contain spam. The study of this research is to identify the potentially useful email header features for email spam detection by analyzing two (2) email datasets, the Anomaly Detection Challenges and Cyber Security Data Mining from website. By analyzing the datasets, the main objective of this research is to extract the suitable features of the email header and examine the features to classify the features using Support Vector Machine (SVM) using RapidMiner Studio and Weka 3.9.2. There are five (5) phases in the methodology which are Data Collection, data Pre-Processing, Features Selection, Classification and Detection. Classification of the email header using Support Vector Machine (SVM) for CSDM2010 is higher than the Anomaly Detection Challenges datasets at 88.80% and 87.20% respectively. Thus, SVM proves as a good classifier which produced above 80% accuracy rate for both datasets.