KOMPARASI SVM-LR BERBASIS INDOBERT DAN SMOTE-TOMEK PADA DETEKSI BUZZER
DOI:
https://doi.org/10.24076/joism.2026v7i2.2435Keywords:
Deteksi Buzzer, IndoBERT, SMOTE-Tomek, Imbalanced Data, TikTokAbstract
Dominasi aktivitas buzzer politik di TikTok mengancam orisinalitas aspirasi publik melalui manipulasi opini, yang diperburuk oleh ketimpangan distribusi data (imbalanced data) sehingga menyulitkan deteksi akurat. Penelitian ini bertujuan mengomparasikan performa Support Vector Machine (SVM) dan Logistic Regression (LR) berbasis fitur IndoBERT yang dioptimasi dengan hybrid resampling SMOTE-Tomek untuk menangani ketidakseimbangan kelas ekstrem. Studi kasus dilakukan pada 1.502 komentar terkait isu penculikan mahasiswa. Hasil pengujian menunjukkan integrasi IndoBERT dan SMOTE-Tomek efektif menghasilkan Area Under Curve (AUC) >0,90 pada kedua model. Meskipun SVM unggul marginal dalam sensitivitas (AUC 0,9193), Logistic Regression terbukti lebih superior dalam menjaga validitas pengguna organik dengan spesifisitas 79,2% dan F1-Score kelas minoritas 0,72. Penelitian ini merekomendasikan LR sebagai model terbaik karena karakteristik probabilistiknya lebih adaptif meminimalisir False Positive, sehingga lebih akurat dalam memisahkan narasi organik dari propaganda buzzer.
References
[1] V. Tanuwijaya and S. C. M, “FENOMENA JINGLE " OKE GAS ": STUDI NETNOGRAFI KAMPANYE POLITIK DI TIKTOK Abstrak " OKE GAS " JINGLE PHENOMENON : A NETNOGRAPHIC STUDY OF,” vol. 8, no. 2, pp. 165–184, 2024.
[2] M. R. Hasanah, F. A. Putri, P. K. Nisa, U. Islam, N. Syarif, and H. Jakarta, “Retorika Politik Anies Baswedan pada Kampanye Pemilu 2024 Terhadap Persepsi Publik di TikTok,” vol. 5, no. 2, pp. 566–572, 2025.
[3] P. J. Widodo, “Penggunaan Buzzer dalam Hegemoni Pemerintahan Joko Widodo,” vol. 1, no. 2, pp. 70–81, 2022.
[4] K. S. Id, M. Girnyk, and L. D. Id, “Short text classification with machine learning in the social sciences : The case of climate change on Twitter,” pp. 1–26, 2023, doi: 10.1371/journal.pone.0290762.
[5] S. F. Taskiran, B. Turkoglu, E. Kaya, and T. Asuroglu, “OPEN A comprehensive evaluation of oversampling techniques for enhancing text classification performance,” pp. 1–20, 2025.
[6] A. A. Lubis, S. Iskandar, A. Idrus, Z. Indra, and K. S. S, “Klasifikasi Akun Buzzer Menggunakan Algoritma K-Nearest Neighbor pada Tagar # STYTanpaDiasporaNol di Media Sosial X,” 2025.
[7] A. D. Pramesti, K. Umam, and M. R. Handayani, “Identification of Buzzers in Skincare Reviews Using a Lexicon-Based Sentiment Analysis Method,” vol. 9, no. 5, pp. 2598–2606, 2025.
[8] O. Manullang, C. Prianto, and N. H. Harani, “Analisis Sentimen Untuk Memprediksi Hasil Calon Pemilu Presiden Menggunakan Lexicon Based Dan Random Forest,” J. Ilm. Inform., vol. 11, no. 02, pp. 159–169, 2023, doi: 10.33884/jif.v11i02.7987.
[9] N. Cahyono, S. Informatika, and U. A. Yogyakarta, “Analisis Sentimen Komentar Instagram Pada Program Kampus,” vol. 8, no. 2, pp. 2372–2381, 2024.
[10] Catur Arpal Perkasa, Amalia Andjani Arifiyanti, and Agus Salim, “Klasifikasi Akun Buzzer Pada Twitter Menggunakan Algoritma Naive Bayes,” J. Ilm. Tek. Inform. dan Komun., vol. 3, no. 1, pp. 01–12, 2023, doi: 10.55606/juitik.v3i1.363.
[11] R. Mustika, “Pergeseran Peran Buzzer Ke Dunia Politik Di Media Sosial,” Diakom J. Media dan Komun., vol. 2, no. 2, pp. 144–151, 2019, doi: 10.17933/diakom.v2i2.60.
[12] N. Nyoman, E. Perimawati, R. R. Huizen, and D. P. Hostiadi, “Analisa Pengaruh Pre-Procesing Data Untuk Model Deteksi Akun Palsu Pada Media Sosial,” vol. 2, no. 1, pp. 1117–1122, 2025.
[13] M. Wong, A. Alshehri, S. Kao, and H. He, “PolyNorm : Few-Shot LLM-Based Text Normalization for Text-to-Speech,” pp. 77–85, 2025.
[14] U. Khairani et al., “PENGARUH TAHAPAN PREPROCESSING TERHADAP MODEL INDOBERT DAN THE INFLUENCE OF PREPROCESSING STAGES ON INDOBERT AND INDOBERTWEET MODELS FOR EMOTION DETECTION IN INSTAGRAM NEWS,” vol. 11, no. 4, 2024, doi: 10.25126/jtiik.1148315.
[15] B. Liu, “Sentiment Analysis and Opinion Mining,” no. May, 2012.
[16] B. Wilie et al., “IndoNLU : Benchmark and Resources for Evaluating Indonesian Natural Language Understanding,” pp. 843–857, 2020.
[17] M. C. Monard, “A Study of the Behavior of Several Methods for Balancing Machine Learning Training Data”.
[18] R. Peranginangin, E. J. G. Harianja, I. K. Jaya, and B. Rumahorbo, “Penerapan Algoritma Safe-Level-Smote Untuk Peningkatan Nilai G-Mean Dalam Klasifikasi Data Tidak Seimbang,” METHOMIKA J. Manaj. Inform. dan Komputerisasi Akunt., vol. 4, no. 1, pp. 67–72, 2020, doi: 10.46880/jmika.vol4no1.pp67-72.
[19] E. F. Swana and W. Doorsamy, “Tomek Link and SMOTE Approaches for Machine Fault Classification with an Imbalanced Dataset,” 2022.
[20] C. Hsu, C. Chang, and C. Lin, “A Practical Guide to Support Vector Classification,” vol. 1, no. 1, pp. 1–20, 2025.
[21] L. Saitta, “Support-Vector Networks,” vol. 297, pp. 273–297, 1995.
[22] A. Roy and S. Chakraborty, “Support vector machine in structural reliability analysis: A review,” Reliab. Eng. Syst. Saf., vol. 233, no. April, 2023, doi: 10.1016/j.ress.2023.109126.
[23] F. A. Wicaksono, A. Romadhony, and Hasmawati, “Sentiment Analysis of University Social Media Using Support Vector Machine and Logistic Regression Methods,” Ind. J. Comput., vol. 7, no. 2, pp. 15–24, 2022, doi: 10.34818/indojc.2022.7.2.638.
[24] S. A. H. Bahtiar, C. K. Dewa, and A. Luthfi, “Comparison of Naïve Bayes and Logistic Regression in Sentiment Analysis on Marketplace Reviews Using Rating-Based Labeling,” J. Inf. Syst. Informatics, vol. 5, no. 3, pp. 915–927, 2023, doi: 10.51519/journalisi.v5i3.539.
[25] E. C. Zabor, C. A. Reddy, R. D. Tendulkar, and S. Patil, “Logistic Regression in Clinical Studies,” Int. J. Radiat. Oncol. Biol. Phys., vol. 112, no. 2, pp. 271–277, 2022, doi: 10.1016/j.ijrobp.2021.08.007.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Journal of Information System Management (JOISM)
This work is licensed under a Creative Commons Attribution 4.0 International License.
