Detecting Spam Emails Using Machine Learning and Lemmatization vs Traditional Methods

Emails are a common method of communication and used by many companies for public facing systems. As malicious actors send spam emails to try to gain access to the systems, filters must be put in place to filter them out. Although traditional email filtering methods exist, a machine learning approach can be employed to attain a higher accuracy of filtering. This paper employs a machine learning method Extra Trees Classifier with lemmatization to determine the benefit that such a method will have compared to methods not using machine learning. Through the results of our testing, this method achieved an accuracy of 98.16% and precision of 94.30% with a lemmatized dataset detecting spam emails. This is an improvement of 10.07<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup> compared to previous study employing traditional method of email filtering using active domain name service data and email reception log. This method of Extra Trees Classifier with lemmatization also achieves better accuracy than other machine learning methods for the same dataset.