SMS spam detection using BERT and multi-graph convolutional networks

The surge in smartphone usage has significantly increased Short Message Service (SMS) traffic and, consequently, SMS spam, posing risks such as phishing, financial losses, and privacy breaches. Traditional rule-based and blacklist methods fail against evolving spamming techniques, prompting the adop...

Full description

Saved in:
Bibliographic Details
Main Authors: Linjie Shen, Yanbin Wang, Zhao Li, Wenrui Ma
Format: Article
Language:English
Published: KeAi Communications Co., Ltd. 2025-01-01
Series:International Journal of Intelligent Networks
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S2666603025000089
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850070211721428992
author Linjie Shen
Yanbin Wang
Zhao Li
Wenrui Ma
author_facet Linjie Shen
Yanbin Wang
Zhao Li
Wenrui Ma
author_sort Linjie Shen
collection DOAJ
description The surge in smartphone usage has significantly increased Short Message Service (SMS) traffic and, consequently, SMS spam, posing risks such as phishing, financial losses, and privacy breaches. Traditional rule-based and blacklist methods fail against evolving spamming techniques, prompting the adoption of machine learning and deep learning approaches. However, models like Convolutional Neural Networks and Recurrent Neural Networks struggle to capture global co-occurrence patterns and complex semantics, while transformer-based models like Bidirectional Encoder Representations from Transformers (BERT) lack explicit syntactic and co-occurrence modeling. To address these limitations, we propose the BERT with Triple-Graph Convolutional Networks (BERT-G3CN) model, the first framework to integrate BERT word embeddings with graph embeddings from Co-occurrence, Heterogeneous, and Integrated Syntactic Graphs. This multigraph approach captures diverse features and models both global and local structures using tailored Graph Convolutional Networks. Experiments on two benchmark datasets demonstrate that BERT-G3CN achieves superior accuracy of 99.28 % and 93.78 %, representing an improvement of approximately 2–3 % over competitive baselines.
format Article
id doaj-art-f5f1810a2c0d43fe8c0ec10a5b9c47e3
institution DOAJ
issn 2666-6030
language English
publishDate 2025-01-01
publisher KeAi Communications Co., Ltd.
record_format Article
series International Journal of Intelligent Networks
spelling doaj-art-f5f1810a2c0d43fe8c0ec10a5b9c47e32025-08-20T02:47:36ZengKeAi Communications Co., Ltd.International Journal of Intelligent Networks2666-60302025-01-016798810.1016/j.ijin.2025.06.002SMS spam detection using BERT and multi-graph convolutional networksLinjie Shen0Yanbin Wang1Zhao Li2Wenrui Ma3School of Computer Science and Technology, Zhejiang Gongshang University, Hangzhou, Zhejiang, ChinaDepartment of Engineering, Shenzhen MSU-BIT University, Shenzhen, 518172, Guangdong, ChinaZhejiang Lab, Hangzhou, Zhejiang, ChinaSchool of Computer Science and Technology, Zhejiang Gongshang University, Hangzhou, Zhejiang, China; Department of Big Data and Future E-Commerce Technology, Zhejiang Key Laboratory, Hangzhou, Zhejiang, China; Corresponding author. School of Computer Science and Technology, Zhejiang Gongshang University, Hangzhou, Zhejiang, ChinaThe surge in smartphone usage has significantly increased Short Message Service (SMS) traffic and, consequently, SMS spam, posing risks such as phishing, financial losses, and privacy breaches. Traditional rule-based and blacklist methods fail against evolving spamming techniques, prompting the adoption of machine learning and deep learning approaches. However, models like Convolutional Neural Networks and Recurrent Neural Networks struggle to capture global co-occurrence patterns and complex semantics, while transformer-based models like Bidirectional Encoder Representations from Transformers (BERT) lack explicit syntactic and co-occurrence modeling. To address these limitations, we propose the BERT with Triple-Graph Convolutional Networks (BERT-G3CN) model, the first framework to integrate BERT word embeddings with graph embeddings from Co-occurrence, Heterogeneous, and Integrated Syntactic Graphs. This multigraph approach captures diverse features and models both global and local structures using tailored Graph Convolutional Networks. Experiments on two benchmark datasets demonstrate that BERT-G3CN achieves superior accuracy of 99.28 % and 93.78 %, representing an improvement of approximately 2–3 % over competitive baselines.http://www.sciencedirect.com/science/article/pii/S2666603025000089SMS spam detectionBERTGraph convolutional networks
spellingShingle Linjie Shen
Yanbin Wang
Zhao Li
Wenrui Ma
SMS spam detection using BERT and multi-graph convolutional networks
International Journal of Intelligent Networks
SMS spam detection
BERT
Graph convolutional networks
title SMS spam detection using BERT and multi-graph convolutional networks
title_full SMS spam detection using BERT and multi-graph convolutional networks
title_fullStr SMS spam detection using BERT and multi-graph convolutional networks
title_full_unstemmed SMS spam detection using BERT and multi-graph convolutional networks
title_short SMS spam detection using BERT and multi-graph convolutional networks
title_sort sms spam detection using bert and multi graph convolutional networks
topic SMS spam detection
BERT
Graph convolutional networks
url http://www.sciencedirect.com/science/article/pii/S2666603025000089
work_keys_str_mv AT linjieshen smsspamdetectionusingbertandmultigraphconvolutionalnetworks
AT yanbinwang smsspamdetectionusingbertandmultigraphconvolutionalnetworks
AT zhaoli smsspamdetectionusingbertandmultigraphconvolutionalnetworks
AT wenruima smsspamdetectionusingbertandmultigraphconvolutionalnetworks