Transformer Models for Smishing Detection: Benchmarking BERT and ELECTRA

Abstract

Today, the field of artificial intelligence is becoming increasingly popular. This popularity continues to drive the introduction of new models for diverse tasks. BERT (Bidirectional Encoder Representations from Transformers) and ELECTRA (Efficiently Learning an Encoder that Classifies Token Replacements Accurately) models are also two of the popular and advanced transformer models. The aim of this study is to comparatively analyze the performance of BERT and ELECTRA models in detecting smishing attacks. For this purpose, the dataset used is compared for the first time in the literature. The tests show that the BERT model achieves a higher accuracy rate than the ELECTRA model in detecting smishing attacks. However, the ELECTRA model produces results with comparable accuracy while being faster and more resource-efficient. The results of this study show that both models were able to produce reliable classification results. © 2025 IEEE.