A ROBUST SUPERVISED MACHINE LEARNING APPROACH FOR SPAM DETECTION IN DIGITAL COMMUNICATION

Abstract

Email and message filtering systems perform better and are more dependable when supervised learning techniques are applied. This study examines an effective method for identifying spam. Because so many people communicate online, spam is a major issue for both businesses and individuals. However, traditional filtering techniques may not always be sufficient to identify emerging trends in trash. In order to distinguish between spam and legitimate messages, supervised machine learning techniques such Naïve Bayes, Support Vector Machine, Decision Tree, and Random Forest are employed in this work. We utilize a marked-up sample to train the models and identify key elements in the text. Preprocessing techniques that improve the model's performance include tokenization, stop-word removal, and feature extraction. Many classifiers are evaluated based on their F1-score, accuracy, precision, and recall using the recommended method. To determine the most effective technique for identifying spam, a comparison study is conducted. Supervised learning techniques significantly improve detection accuracy and reduce false positives, according to empirical evidence. Regular training can teach the algorithm to identify new types of spam.