SMSHIELD: Sistem Deteksi Pesan Berbahaya Phishing (Smishing) secara Real-Time Berbasis Aplikasi Mobile dengan Pendekatan Deep Learning

Rahmapuri, Annisa (2025) SMSHIELD: Sistem Deteksi Pesan Berbahaya Phishing (Smishing) secara Real-Time Berbasis Aplikasi Mobile dengan Pendekatan Deep Learning. Other thesis, Institut Teknologi Sepuluh Nopember.

[thumbnail of 5027211018-Undergraduate_Thesis.pdf] Text
5027211018-Undergraduate_Thesis.pdf - Accepted Version
Restricted to Repository staff only

Download (25MB) | Request a copy

Abstract

Kemajuan teknologi informasi dan komunikasi telah mendorong efisiensi dalam komunikasi digital, namun juga membuka peluang bagi kejahatan siber, seperti smishing (SMS phishing). Meskipun pendekatan deep learning telah digunakan untuk mendeteksi pesan penipuan, sebagian besar penelitian masih terbatas pada bahasa inggris dan klasifikasi biner, (phishing dan non-phishing). Penelitian ini bertujuan untuk mengembangkan sistem deteksi pesan penipuan dalam Bahasa Indonesia dengan pendekatan klasifikasi multikelas (normal, promo, dan phishing) menggunakan model deep learning, yaitu Long Short-Term Memory (LSTM), Bidirectional Long Short-Term Memory (BiLSTM), dan Recurrent Neural Network (RNN). Dataset yang digunakan dibangun dalam tiga versi dengan teknik balancing dan data augmentation berbasis IndoBERT untuk mengatasi ketimpangan kelas. Evaluasi dilakukan dengan metrik akurasi, presisi, recall, dan F1-score serta analisis confusion matrix. Hasil menunjukkan bahwa distribusi data yang seimbang secara signifikan meningkatkan performa model, khususnya dalam klasifikasi kelas promo. BiLSTM menghasilkan performa terbaik secara konsisten dengan akurasi tertinggi sebesar 97,27% dan F1-score yang mendekati sempurna, yaitu 0,97 untuk kelas promo dan normal, serta 0,98 untuk kelas phishing. Penelitian ini juga menghasilkan aplikasi mobile berbasis Kotlin yang mengimplementasikan sistem deteksi secara praktis. Temuan ini menunjukkan efektivitas pendekatan BiLSTM dalam mendeteksi pesan berbahaya berbahasa Indonesia, serta pentingnya distribusi data yang seimbang dalam pelatihan model.
======================================================================================================================================
Rapid advances in information and communication technology have driven efficiency in digital communication, but also opened up opportunities for cybercrime, such as smishing (SMS phishing). Although deep learning approaches have been used to detect fraudulent messages, most studies are still limited to English and binary classification, (phishing and non phishing). This research aims to develop a fraud message detection system in Bahasa Indonesia with a multi-class classification approach (normal, promo, and phishing) using deep learning models, namely Long Short-Term Memory (LSTM), Bidirectional Long Short-Term Memory (BiLSTM), and Recurrent Neural Network (RNN). The dataset used was built in three versions with IndoBERT-based balancing and data augmentation techniques to overcome class inequality. The evaluation was conducted with accuracy, precision, recall, and F1-score metrics as well as confusion matrix analysis. Results show that balanced data distribution significantly improves model performance, especially in promo class classification. BiLSTM produced the best performance consistently with the highest accuracy of 97.27% and near-perfect F1-score of 0.97 for promo and normal classes, and 0.98 for phishing class. This research also produced a Kotlin-based mobile application that implements the detection system in a practical way. The findings demonstrate the effectiveness of the BiLSTM approach in detecting malicious messages in Indonesian, as well as the importance of balanced data distribution in model training.

Item Type: Thesis (Other)
Uncontrolled Keywords: Smishing, Deteksi Pesan, Klasifikasi Multikelas, Deep Learning, Bidirectional Long Short Term Memory (BiLSTM) Smishing, Message Detection, Multiclass Classification, Deep Learning, Bidirectional Long Short-Term Memory (BiLSTM)
Subjects: T Technology > T Technology (General)
Divisions: Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information Technology > 59201-(S1) Undergraduate Thesis
Depositing User: Annisa Rahmapuri
Date Deposited: 29 Jul 2025 08:23
Last Modified: 29 Jul 2025 08:23
URI: http://repository.its.ac.id/id/eprint/123443

Actions (login required)

View Item View Item