Performance Comparison of IndoBERT and Bi-LSTM Models for Sentiment Analysis of Shopee App Users in Indonesia
DOI:
https://doi.org/10.30871/jaic.v10i2.10996Keywords:
BERT, IndoBERT, Sentiment Analysis, Natural Language Processing (NLP), Ecommerce ReviewsAbstract
This study compares the performance of IndoBERT and Bidirectional Long Short-Term Memory (Bi-LSTM) models for sentiment classification of Indonesian-language product reviews from the Shopee e-commerce platform. The original dataset employed a 1–5 rating scale, which was reduced to two sentiment categories: negative (ratings 1–2) and positive (ratings 4–5), while reviews with a rating of 3 were excluded due to their ambiguous nature. IndoBERT (indobenchmark/indobert-base-p1) was applied through a fine-tuning process, whereas the Bi-LSTM model was trained from scratch using comprehensive text preprocessing, including case folding, stopword removal, stemming, tokenization, and padding. The dataset consisted of 7,225 reviews, divided into 5,780 training samples and 1,445 testing samples. Model performance was evaluated using accuracy, precision, recall, and F1-score metrics. Experimental results demonstrate that IndoBERT outperforms Bi-LSTM, achieving an accuracy and F1-score of 85.61%, compared to 78.27% obtained by the Bi-LSTM model. These findings indicate that transformer-based models are more effective in capturing contextual semantics in Indonesian text than recurrent neural network-based approaches.
Downloads
References
[1] A. H. Ruger, M. Suyanto, and M. P. Kurniawan, “Sentimen Analisis Pelanggan Shopee di Twitter dengan Algoritma Naive Bayes,” J. Inf. Technol., vol. 1, no. 2, pp. 26–29, 2021, doi: 10.46229/jifotech.v1i2.282.
[2] R. C. Rivaldi, T. D. Wismarini, J. T. Lomba, and J. Semarang, “Analisis Sentimen Pada Ulasan Produk Dengan Metode Natural Language Processing (NLP) (Studi Kasus Zalika Store 88 Shopee),” vol. 17, no. 1, pp. 120–128, 2024.
[3] N. Firdausy, I. Yuadi, and I. Puspitasari, “Analisis Sentimen Evaluasi Reaksi E-Learning Menggunakan Algorima Naïve Bayes, Support Vector Machine Dan Deep Learning,” Techno.Com, vol. 22, no. 3, pp. 677–689, 2023, doi: 10.33633/tc.v22i3.8160.
[4] O. Boas, J. Putro, A. Jacobus, and F. D. Kambey, “Aspect-Based Sentiment Analysis Product Review Using CNN and Bidirectional LSTM,” vol. 20, no. 2, pp. 117–124, 2025.
[5] S. W. Nadya Sikana, “Analisis Sentimen untuk Ulasan Produk E-Commerce,” vol. 26, no. 2, pp. 223–238, 2025.
[6] M. M. Shivaji Alaparthi, “Bidirectional Encoder Representations from Transformers (BERT): A sentiment analysis odyssey Shivaji,” arXiv Prepr., no. 1, pp. 1–15, 2020, doi: https://doi.org/10.48550/arXiv.2007.01127.
[7] D. Yuliana, A. Ningrum, E. Daniati, M. N. Muzaki, and S. Informasi, “Perbandingan Model BERT dan RNN-LSTM pada Analisis Sentimen Aplikasi BRI Mobile,” vol. 4, no. 2, pp. 75–85, 2025.
[8] J. Saquer, “A Comparative Analysis of Transformer and LSTM Models for Detecting Suicidal Ideation on Reddit”.
[9] S. Alfaris and Kusnawi, “Komparasi Metode KNN dan Naive Bayes Terhadap Analisis Sentimen Pengguna Aplikasi Shopee,” Indones. J. Comput. Sci., vol. 12, no. 5, pp. 2766–2776, 2023, doi: 10.33022/ijcs.v12i5.3304.
[10] R. Illahi, S. Agustian, S. K. Riau, S. Baru, and K. Pekanbaru, “Bidirectional Lstm Dan Indobert Dengan Dataset Terbatas,” vol. 7, no. 1, pp. 74–84.
[11] N. Wayan, A. Sekar, and A. R. Isnain, “Analisis Sentimen Terhadap Media Sosial Twitter dengan Kasus Kampanye Anti-Korupsi di Indonesia Menggunakan Naive Bayes,” J. Media Inform. Budidarma, vol. 8, no. April, pp. 695–703, 2024, doi: 10.30865/mib.v8i2.7582.
[12] G. Darmawan, S. Alam, and M. I. Sulistyo, “Analisis Sentimen Berdasarkan Ulasan Pengguna Aplikasi Mypertamina Pada Google Playstore Menggunakan Metode Naïve Bayes,” STORAGE – J. Ilm. Tek. dan Ilmu Komput., vol. 2, no. 3, pp. 100–108, 2023.
[13] M. R. R. Lillah, D. S. Maylawati, W. B. Zulfikar, W. Uriawan, and A. Wahana, “Implementasi Algoritma K-Nearest Neighbor (KNN) untuk analisis sentimen pengguna aplikasi Tokopedia,” Intellect Indones. J. Learn. Teachnological Innov., vol. 02, no. 02, pp. 171–184, 2023, [Online]. Available: https://digilib.uinsgd.ac.id/77056/%0Ahttps://digilib.uinsgd.ac.id/77056/6/4_BAB I.pdf
[14] N. Z. B. Jannah and K. Kusnawi, “Comparison of Naïve Bayes and SVM in Sentiment Analysis of Product Reviews on Marketplaces,” Sinkron, vol. 8, no. 2, pp. 727–733, 2024, doi: 10.33395/sinkron.v8i2.13559.
[15] A. Ananta Firdaus, A. Id Hadiana, and A. Kania Ningsih, “Klasifikasi Sentimen pada Aplikasi Shopee Menggunakan Fitur Bag of Word dan Algoritma Random Forest,” Ranah Res. J. Multidiscip. Res. Dev., vol. 6, no. 5, pp. 1678–1683, 2024, doi: 10.38035/rrj.v6i5.994.
[16] N. Ray, A. Tambunan, D. Retno, and S. Saputro, “Hybrid Integration Of Bert And Bilstm Models For,” vol. 20, no. 2, pp. 1719–1730, 2026.
[17] M. Mustafa and S. Kumar, “BERT-Enhanced Bi-LSTM with weighted cross-entropy for multilingual sentiment classification,” vol. 11, no. 3, pp. 396–416, 2025.
[18] Y. Huang, “Sentiment Analysis of News Content Based on,” pp. 616–620, 2025, doi: 10.1145/3783669.3783764.
[19] M. F. Cahyadi and T. H. Rochadiani, “Implementasi Ensemble Deep Learning Untuk Analisis Sentimen Terhadap Genre Game Mobile,” J. Media Inform. Budidarma, vol. 8, no. 3, p. 1512, 2024, doi: 10.30865/mib.v8i3.7832.
[20] J. Kecerdasan, A. Rekayasa, and K. Wau, “Application of Fine-Tuned IndoBERT for Sentiment Classification Local Product Reviews on Tokopedia Marketplace with Limited Dataset,” vol. 5, no. 1, pp. 1–5, 2025.
[21] S. Aras, R. Ruimassa, E. Agustinus, B. Wambrauw, and E. B. Palalangan, “Sentiment Analysis on Shopee Product Reviews Using IndoBERT,” vol. 6, no. 3, pp. 1616–1627, 2024, doi: 10.51519/journalisi.v6i3.814.
[22] A. C. Adamuthe, “Improved Text Classification using Long Short-Term Memory and Word Embedding Technique,” vol. 13, no. 1, pp. 19–32, 2020.
[23] Y. Wang, X. Cheng, and X. Meng, “Sentiment Analysis with An Integrated Model of BERT and Bi-LSTM Based on Multi-Head Attention Mechanism,” vol. 50, no. 1, 2023.
[24] A. Sami, S. Buyrukoğlu, and M. Rashad, “Advanced deep learning techniques for sentiment analysis : combining Bi-LSTM , CNN , and attention layers,” vol. 11, no. 1, pp. 55–71, 2025.
[25] C. Fri, R. Elouahbi, Y. Taki, and A. Remaida, “Enhanced Bidirectional LSTM for Sentiment Analy sis of Learners ’ Posts in MOOCs,” vol. 16, no. 5, pp. 163–172, 2025.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2026 Taufik Ramadhan, Kusnawi Kusnawi
This work is licensed under a Creative Commons Attribution-ShareAlike 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 a Creative Commons Attribution License (Attribution-ShareAlike 4.0 International (CC BY-SA 4.0) ) that allows others to share the work with an acknowledgement 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 acknowledgement 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 (See The Effect of Open Access).
