Enhancing Spam Filtering: A Comparative Study of Modern Advanced Machine Learning Techniques

Abstract

Spam remains a persistent issue that not only consumes time and bandwidth but also poses significant cybersecurity threats. As a result, effective spam filtering has become essential. With an emphasis on Naïve Bayes (NB), Decision Trees (DT), and Support Vector Machines (SVM), this study offers a thorough analysis of the major machine learning techniques utilized in contemporary spam filtering. This paper investigates underlying principles of these methods, compares their performance through extensive experiments conducted on the Kaggle dataset, and discusses the cunent challenges and future directions for spam filtering technology. The study reveals that SVM is particularly effective for handling high-dimensional data. DT offers superior interpretability, and NB simplifies probabilistic classification. Experimental results demonstrate that while each method has its strengths and weaknesses, combining SVM with NB notably enhances classification accuracy. Despite these advances, spam filters still face challenges due to evolving spamming tactics. In order to address these persistent problems, the conclusion part highlights the need for more reliable and flexible spam filtering teclmologies and makes recommendations for future research directions.