Perbandingan Embedding IndoBERT dan FastText dalam Deteksi Cyberbullying Berbahasa Indonesia
- Muhammad Ihksan Alfiansyah Universitas Bina Darma, Palembang, Indonesia
-
Diana Diana
*
Universitas Bina Darma, Palembang,
Indonesia
(*) Corresponding Author
Abstract
The escalation of cyberbullying on social media necessitates automated content moderation systems capable of deciphering the characteristics of non-standard digital language. This study compares the effectiveness of contextual word embeddings (IndoBERT) and static subword embeddings (FastText) in multi-class text classification (Safe, Mild, Moderate, Severe, and Very Dangerous) for Indonesian-language social media comments. The research focuses on analyzing embedding architectures: IndoBERT is integrated with a Dense Layer, while FastText is evaluated using Logistic Regression and Random Forest (15 trees). A dataset of 16,643 samples was extracted from social media platforms and processed through text cleaning and dictionary-based slang normalization stages. Test results indicate that the bidirectional contextual architecture of IndoBERT achieved the highest performance, with an accuracy of 82.15% and a Macro F1-Score of 74.97%. Regarding FastText representations, the linear Logistic Regression approach (78.70% accuracy) proved more stable than the Random Forest model (65.50% accuracy). These findings demonstrate that contextual embeddings are more adaptive than static embeddings at distinguishing sarcasm and mutated lexicons in digital text. As a research contribution, this study not only offers a precise five-level severity classification architecture for informal Indonesian but also provides a prototype automated moderation system (CyberGuard AI) ready for implementation in real-time risk mitigation.
Downloads
References
D. Riswanto and R. Marsinun, “Perilaku Cyberbullying Remaja di Media Sosial,” ANALITIKA, vol. 12, no. 2, pp. 98–111, 2020, doi: 10.31289/analitika.v12i2.3704.
A. N. A. W, C. Santika, I. Wahyudi, and K. M. Istiqomah, “Perundungan Dunia Maya Menjadi Faktor Utama Depresi Berujung Bunuh Diri,” De Juncto Delicti Journal of Law, vol. 4, no. 2, pp. 15–29, 2024, doi: 10.35706/djd.v4i2.7972.
V. Auriemma, G. Iorio, G. Roberti, and R. Morese, “Cyberbullying and Empathy in the Age of Hyperconnection: An Interdisciplinary Approach,” Frontiers in Sociology, vol. 5, p. 551881, 2020, doi: 10.3389/fsoc.2020.551881.
M. D. Sánchez-Hernández, M. C. Herrera, and F. Expósito, “Is Online Disinhibition Related to Cyberdating Abuse Perpetration through Moral Disengagement? The Moderating Role of Gender, Sexism, and Cybervictimization,” Sex Roles, vol. 90, no. 7, pp. 938–959, 2024, doi: 10.1007/s11199-024-01480-3.
S. Govindankutty and A. Kumar, “Design and Implementation of Automated Content Moderation Systems in Social Media,” Integrated Journal for Research in Arts and Humanities, vol. 4, no. 6, pp. 380–402, 2024, doi: 10.55544/ijrah.4.6.27.
M. O. Ibrohim and I. Budi, “Multi-label Hate Speech and Abusive Language Detection in Indonesian Twitter,” pp. 46–57, 2019, doi: 10.18653/v1/w19-3506.
H. Han, M. Asif, E. M. Awwad, N. Sarhan, Y. Y. Ghadi, and B. Xu, “Innovative deep learning techniques for monitoring aggressive behavior in social media posts,” Journal of Cloud Computing Advances Systems and Applications, vol. 13, no. 1, 2024, doi: 10.1186/s13677-023-00577-6.
A. Muzakir and K. Adi, “Classification of Hate Speech Language Detection on Social Media : Preliminary Study for Improvement,” vol. 4, pp. 1–11.
P. Bojanowski, E. Grave, A. Joulin, and T. Mikolov, “Enriching Word Vectors with Subword Information,” vol. 5, pp. 135–146, 2020.
J. Devlin, “BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding,” no. Mlm, 2019.
A. Onan and M. A. Toçoğlu, “A Term Weighted Neural Language Model and Stacked Bidirectional LSTM Based Framework for Sarcasm Identification,” IEEE Access, vol. 9, pp. 7701–7722, 2021, doi: 10.1109/access.2021.3049734.
A. Vaswani, “Attention Is All You Need,” no. Nips, 2017.
B. Wilie et al., IndoNLU: Benchmark and Resources for Evaluating Indonesian Natural Language Understanding. 2020. doi: 10.48550/arXiv.2009.05387.
F. Koto and T. Baldwin, “IndoLEM and IndoBERT : A Benchmark Dataset and Pre-trained Language Model for Indonesian NLP,” pp. 757–770, 2020.
A. Joulin, “Bag of Tricks for Efficient Text Classification,” vol. 2, pp. 427–431, 2017.
E. Grave, P. Bojanowski, P. Gupta, A. Joulin, and T. Mikolov, “Learning Word Vectors for 157 Languages,” Feb. 2018, doi: 10.48550/arXiv.1802.06893.
M. Ghosh and G. Sanyal, “An ensemble approach to stabilize the features for multi-domain sentiment analysis using supervised machine learning,” J. Big Data, vol. 5, no. 1, 2018, doi: 10.1186/s40537-018-0152-5.
B. Ali, T. Id, and D. O. Sullivan, “Cyberbullying severity detection : A machine learning approach,” pp. 1–19, 2020, doi: 10.1371/journal.pone.0240924.
D. Purwitasari, A. S. S. Ansyah, A. P. Kurniawan, and A. N. Kholifah, “A Hybrid Method on Emotion Detection for Indonesian Tweets of COVID-19,” Jurnal RESTI (Rekayasa Sistem dan Teknologi Informasi), vol. 7, no. 2, pp. 254–262, 2023, doi: 10.29207/resti.v7i2.4816.
G. G. Setiaji, J. Suntoro, and A. Rifa’i, “Random State Parameter Undersampling untuk Penanganan Data dengan Kelas Tidak Seimbang pada Algoritme Random Forest,” Jurnal Transformatika, vol. 21, no. 2, pp. 73–83, 2024, doi: 10.26623/transformatika.v21i2.8901.
Bila bermanfaat silahkan share artikel ini
Berikan Komentar Anda terhadap artikel Perbandingan Embedding IndoBERT dan FastText dalam Deteksi Cyberbullying Berbahasa Indonesia
Pages: 997-1007
Copyright (c) 2026 Muhammad Ihksan Alfiansyah, Diana Diana
This work is licensed under a Creative Commons Attribution 4.0 International License.
Authors who publish with this journal agree to the following terms:
- Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under Creative Commons Attribution 4.0 International License that allows others to share the work with an acknowledgment of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgment of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (Refer to The Effect of Open Access).
