Open access Jurnal SINTA 2
Deteksi Spam Berbahasa Indonesia Berbasis Teks Menggunakan Model Bert
Jurnal Teknologi Informasi dan Ilmu Komputer (JTIIK) · Vol. 11 No. 6 (2024) · pp. 1291–1302
- ISSN
- 2528-6579, 2355-7699
- Published
- 2024-12-10
- Publisher
- Fakultas Ilmu Komputer, Universitas Brawijaya
- Language
- id
Abstract
Bahasa Indonesia
Spam pada SMS dan Email menyebabkan pengalaman kurang menyenangkan bagi pengguna dalam pemanfaatan teknologi. Spam secara umum merupakan sebuah tindakan mengirim pesan yang tidak diinginkan atau tidak diminta kepada sejumlah besar orang. Spam kini dapat ditemui dalam berbagai bentuk, seperti web maupun multimedia. Penelitian ini bertujuan untuk mengevaluasi model berbasis BERT, khususnya IndoBERT dan MultilingualBERT, dalam mendeteksi dan mengklasifikasi spam berbahasa Indonesia pada pesan SMS dan Email. Model yang dipilih kemudian dilatih untuk mengidentifikasi perbedaan antara pesan spam dan bukan spam. Hasil evaluasi pada percobaan menggunakan dataset SMS dan Email memiliki nilai akurasi sebesar 98% pada model IndoBERT dan 95% pada model MultilingualBERT, yang menunjukkan tingkat akurasi yang tinggi. Hasil ini menunjukkan bahwa model BERT efektif dalam mendeteksi pesan spam dalam Bahasa Indonesia. Abstract Spam on SMS and Email causes an unpleasant experience for users in using technology. Spam in general is the act of sending unwanted or unsolicited messages to a large number of people. Spam can now be found in various forms, such as web and multimedia. This research aims to evaluate BERT-based models, specifically IndoBERT and MultilingualBERT, in detecting and classifying Indonesian spam in SMS and Email messages. The selected model is then trained to identify the differences between spam and non-spam messages. Evaluation results in experiments using SMS and Email datasets have an accuracy value of 98% in the IndoBERT model and 95% in the MultilingualBERT model, which shows a high level of accuracy. These results indicate that the BERT model is effective in detecting spam messages in Indonesian.
English
Spam pada SMS dan Email menyebabkan pengalaman kurang menyenangkan bagi pengguna dalam pemanfaatan teknologi. Spam secara umum merupakan sebuah tindakan mengirim pesan yang tidak diinginkan atau tidak diminta kepada sejumlah besar orang. Spam kini dapat ditemui dalam berbagai bentuk, seperti web maupun multimedia. Penelitian ini bertujuan untuk mengevaluasi model berbasis BERT, khususnya IndoBERT dan MultilingualBERT, dalam mendeteksi dan mengklasifikasi spam berbahasa Indonesia pada pesan SMS dan Email. Model yang dipilih kemudian dilatih untuk mengidentifikasi perbedaan antara pesan spam dan bukan spam. Hasil evaluasi pada percobaan menggunakan dataset SMS dan Email memiliki nilai akurasi sebesar 98% pada model IndoBERT dan 95% pada model MultilingualBERT, yang menunjukkan tingkat akurasi yang tinggi. Hasil ini menunjukkan bahwa model BERT efektif dalam mendeteksi pesan spam dalam Bahasa Indonesia. Abstract Spam on SMS and Email causes an unpleasant experience for users in using technology. Spam in general is the act of sending unwanted or unsolicited messages to a large number of people. Spam can now be found in various forms, such as web and multimedia. This research aims to evaluate BERT-based models, specifically IndoBERT and MultilingualBERT, in detecting and classifying Indonesian spam in SMS and Email messages. The selected model is then trained to identify the differences between spam and non-spam messages. Evaluation results in experiments using SMS and Email datasets have an accuracy value of 98% in the IndoBERT model and 95% in the MultilingualBERT model, which shows a high level of accuracy. These results indicate that the BERT model is effective in detecting spam messages in Indonesian.
Keywords
Collaboration network of the authors
40 authors · 110 co-authorship links · showing the 40 most connected
Hover or tap a name to highlight its co-authors. Double-click to open the author's page.
- University of Brawijaya
- Affiliation unknown
- Different organisations
- Same organisation
- Affiliation unknown
- Thicker line = more joint articles · bigger circle = more articles
Show as table
| Co-author | Organisation | Joint articles |
|---|---|---|
| Fitra Abdurrachman Bachtiar | University of Brawijaya | 2 |
| Achmad Ridok | University of Brawijaya | 2 |
| Yuita Arum Sari | University of Brawijaya | 1 |
| Fadhil Yusuf Rahadika | University of Brawijaya | 1 |
| Agus Wahyu Widodo | University of Brawijaya | 1 |
| Bayu Rahayudi | University of Brawijaya | 1 |
| M. Sofyan Irwanto | University of Brawijaya | 1 |
| Candra Dewi | University of Brawijaya | 1 |
| Nurfatima Selle | University of Brawijaya | 1 |
| Zainal Arifien | University of Brawijaya | 1 |
| Fitra Abdurrahman Bachtiar | Affiliation unknown | 1 |
| Alvin Tarisa Akbar | University of Brawijaya | 1 |
| Michael Stephen Lui | University of Brawijaya | 1 |
| Daniel Geoffrey Manurung | University of Brawijaya | 1 |
| Mohammad Ryan Pinasthika | University of Brawijaya | 1 |
| Muhammad Azka Obila Vasya | University of Brawijaya | 1 |
| Rania Aprilia Dwi Setya Putri | University of Brawijaya | 1 |
| Agustinus Parasian Tampubolon | University of Brawijaya | 1 |
| Rakan Fadhil Prayata | University of Brawijaya | 1 |
| Septia Khoirin Nisa | University of Brawijaya | 1 |
| Assyfa Rasida Hanum | University of Brawijaya | 1 |
| Ivykaeyla Adriana Zetha | University of Brawijaya | 1 |
| Salwa Cahyani Putri | University of Brawijaya | 1 |
| Rafifah Ayud Wulandari | University of Brawijaya | 1 |
| Sherla Puspa Andina | University of Brawijaya | 1 |
| Julia Nur Fajrina | University of Brawijaya | 1 |
| Mohammad Alfiano Rizky Mahardika | University of Brawijaya | 1 |
| Yehezkiel Stephanus Austin | University of Brawijaya | 1 |
| Haikal Irfano | University of Brawijaya | 1 |
| Juan Young Christopher | University of Brawijaya | 1 |
| Lintang Cahyaning Sukma | University of Brawijaya | 1 |
| Octo Perdana Putra | University of Brawijaya | 1 |
| Lailil Muflikhah | University of Brawijaya | 1 |
Affiliations come from OpenAlex; journal OAI-PMH metadata carries none. Authors without an OpenAlex ID are matched by name, which can merge or split people.