Email Spam Detection using Deep Learning Approach

Email is one of the most popular and efficient methods of internet communication and data or message sharing. Considering the significance and high usage of emails, the number of spam emails has also increased rapidly. Spam emails are unwanted emails that contains various contents like advertisements, offers, malicious links, malware, trojan, etc. Spammers send junk mails with an intention of committing email fraud, thus it is important to filter spam emails from emails. The motivation of this research is to build email spam detection models by using machine learning and deep learning techniques so that spam emails can be distinguished from legitimate emails with high accuracy. The Enron email dataset has been used and deep learning models are developed to detect and classify new email spam using LSTM and BERT. NLP approach was applied to analyze and perform data preprocessing of the text of the email. The results are compared to the previous models in email spam detection. The proposed deep learning approach obtained the highest accuracy of 99.14% using BERT, 98.34% using BiLSTM and 97.15% using LSTM. Python is utilized for all implementations.