COMPARATIVE ANALYSIS OF MACHINE LEARNING MODELS FOR DETECTING EVASIVE SMS SPAM

This article analyzes different machine learning methods to identify deceptive SMS spam. Evasive spam communications are famously challenging to identify due to their use of obfuscation to circumvent conventional filters.   A variety of models are assessed, including Deep Learning, Naïve Bayes, Decision Trees, and Support Vector Machines. The collection comprises preprocessed spam and ham messages derived from real-world sources.  The evaluative metrics employed for comparison are F1-score, recall, accuracy, and precision.  The experiment's results demonstrate the advantages and disadvantages of each paradigm.   In the presence of intricate patterns, deep learning models surpass conventional methods.To enhance detection, feature engineering and data augmentation are necessary. The article offers various tips to enhance spam detection models.  In response to the evolving tactics of spam, models will be enhanced in the future.