Smishing Detection Using Fine-Tuned BERT

Abstract

Nowadays, the advancement of technology has made communication much easier. However, this convenience also brings some disadvantages. One of these disadvantages is smishing attacks. Although there are numerous studies aimed at preventing smishing attacks in the fields of email and web, there are only a few studies focusing on SMS. The increasing number of mobile devices and their use by people of all ages clearly reveal the necessity of taking precautions in this area as well. This study aims to detect whether a message is a smishing attack by analyzing messages through Natural Language Processing (NLP). To the best of our knowledge, this study is the first in the literature where the BERT-base model has been fine-tuned on the utilized dataset. The experiments were conducted by randomly splitting the dataset into training and test subsets. The proposed model achieved a reliable result with an accuracy rate of up to 99.82%. © 2025 IEEE.