AI-Enhanced SMS Spam Detection Using Hybrid NLP and Machine Learning Techniques

Abstract: Short Message Service (SMS) spam has emerged as a significant cybersecurity threat due to the rapid growth of mobile communication systems. With billions of SMS messages exchanged daily, malicious actors exploit this platform to distribute phishing links, fraudulent advertisements, fake financial alerts, and malware. Traditional rule-based spam filtering techniques, which rely on predefined keywords and patterns, have become ineffective against modern spam strategies that continuously evolve [1].

Recent advancements in Machine Learning (ML), Deep Learning (DL), and Natural Language Processing (NLP) have significantly improved the capability to detect spam messages with higher accuracy. Transformer-based models such as BERT further enhance semantic understanding of short text messages [10].

This research proposes a Hybrid Adaptive SMS Spam Detection Model that integrates TF-IDF, word embeddings, and transformer-based contextual representations. Additionally, it incorporates adaptive learning mechanisms to handle concept drift and evolving spam patterns. The proposed system not only improves classification accuracy but also reduces false positives, ensuring better user experience and system reliability.

Keywords: SMS Spam Detection, Machine Learning, NLP, BERT, Cybersecurity.