Spam Email Detection

Electronic mail (email) is one of the most important and widely used forms of communication today in the digital world. As the Internet has become more popular, and web communication is becoming more widespread, the amount of spam e-mail has also risen sharply. Spam e-mails are unwanted or malicious e-mails that may include advertisements, phishing links, fraudulent information, malware, or attempts to obtain sensitive user information. These emails are not only a productivity issue for users, but they also present significant cybersecurity threats for individuals and organisations alike, and can also waste network bandwidth.The traditional spam filtering methods, including manual filters and signature-based filters, are ineffective against modern-day spam attacks. Spammers are constantly changing their approach to messages, words, and delivery methods in order to evade standard security measures. Therefore, intelligent and adaptive spam detection techniques must be developed to detect and filter out spam emails in real-time.The contribution of this research is the spam email detection system using machine learning classification technique that automatically determines whether the email is spam or legitimate (ham). The proposed system performs Natural Language Processing (NLP) and machine learning on the content of emails to detect its textual and structural properties. To enhance text quality in the pre-processing, unwanted elements like punctuation marks, stop words, URLs, HTML tags and so on are removed. The TF-IDF (Term Frequency–Inverse Document Frequency) technique is used to extract features from the text and transform it into numerical feature vectors for training the model.The study applies and tests Naive Bayes (NB) and Support Vector Machine (SVM) classifiers and a hybrid model called Naive Bayes-SVM (NB-SVM) which is a fusion of NAIVE BAYES and SVM to boost the classification accuracy. The experimental results prove the proposed model to be efficient and high accuracy, precision, recall, and F1-score.