Pemanfaatan Analisis Sentimen Ulasan Pelanggan dalam Peningkatan Kualitas Layanan E-Commerce XYZ Menggunakan IndoBERT dan Arsitektur Hybrid
Downloads
Ulasan pelanggan pada platform e-commerce mengandung informasi penting untuk mengevaluasi kualitas layanan. Namun, jumlah data yang besar dan penggunaan bahasa informal menyulitkan analisis secara manual. Penelitian ini bertujuan menganalisis distribusi sentimen pelanggan E-Commerce XYZ, membandingkan kinerja enam arsitektur model bahasa pra-latih, serta menyusun rekomendasi peningkatan layanan. Data mencakup 23.303 ulasan Google Play Store periode Januari 2025 hingga Februari 2026, dengan pelabelan oleh tiga anotator yang menghasilkan rata-rata Cohen’s Kappa sebesar 0,956. Analisis deskriptif menggunakan 23.191 ulasan setelah preprocessing, sedangkan pemodelan melibatkan 18.609 ulasan unik dengan pembagian data 80:10:10 secara stratified. Arsitektur yang diuji meliputi IndoBERT end-to-end, IndoBERT+CNN, IndoBERT+LSTM, IndoBERT+BiLSTM, IndoBERT+GAT, dan IndoRoBERTa. Setiap model dilatih menggunakan tiga seed dan dibandingkan berdasarkan rata-rata weighted F1-score validasi. Hasil menunjukkan bahwa sentimen positif mendominasi (52,50%), diikuti negatif (38,48%) dan netral (9,02%). Delivery dan Product memiliki proporsi sentimen negatif tertinggi, masing-masing 58,2% dan 55,0%, sementara Service mencakup 76,5% dari seluruh keluhan negatif. IndoBERT+CNN menghasilkan weighted F1-score validasi sebesar 0,9139 dan pengujian sebesar 0,8945. Berdasarkan volume dan intensitas keluhan, Service memerlukan perbaikan sistemik, Delivery dan Product memerlukan perbaikan terfokus, sedangkan Price cukup dipantau. Dengan ambang confidence 0,9, model mampu mengklasifikasikan 74,4% data secara otomatis dengan akurasi 95,81%. Temuan ini menunjukkan potensi analisis sentimen dalam mendukung pemantauan persepsi pelanggan dan penentuan langkah perbaikan layanan.
Ahmadian, H., Abidin, T. F., Riza, H., & Muchtar, K. (2024). Hybrid models for emotion classification and sentiment analysis in Indonesian language. Applied Computational Intelligence and Soft Computing, 2024(1), 2826773.
Christian, W., Adamlu, D., Yu, A., & Suhartono, D. (2025). Leveraging IndoBERT and DistilBERT for Indonesian emotion classification in e-commerce reviews. Procedia Computer Science, 269, 321–330.
Dzakwan Bintang Lazuardi, M., & Noor Fatyanosa, T. (2025). Klasifikasi Emosi Multikelas Berbasis Teks Bahasa Indonesia Menggunakan IndoRoBERTa. Jurnal Pengembangan Teknologi Informasi dan Ilmu Komputer, 9(9), 1205–1214.
Farizi, A. F. Al, & Sibaroni, Y. (2025). Implementation of BiLSTM and IndoBERT for Sentiment Analysis of TikTok Reviews. JIPI (Jurnal Ilmiah Penelitian dan Pembelajaran Informatika), 10(1), 96–106.
Gürbüz, M., & Kotan, M. (2025). Multi-Category E-Commerce Insights via Social Media Analysis using Machine Learning and BERT. Acta Infologica, 2025(1), 1–18.
Ihtada, F. K., Alfianita, R., & Aziz, O. Q. (2025). Aspect-based Multilabel Classification of E-commerce Reviews Using Fine-tuned IndoBERT. Kinetik: Game Technology, Information System, Computer Network, Computing, Electronics, and Control, 10(1), 45–56.
Imron, S., Setiawan, E. I., Santoso, J., & Purnomo, M. H. (2023). Aspect Based Sentiment Analysis Marketplace Product Reviews Using BERT, LSTM, and CNN. Jurnal RESTI (Rekayasa Sistem dan Teknologi Informasi), 7(3), 586–591.
Koto, F., Rahimi, A., Lau, J. H., & Baldwin, T. (2020). IndoLEM and IndoBERT: A Benchmark Dataset and Pre-trained Language Model for Indonesian NLP. Proceedings of the 28th International Conference on Computational Linguistics, 757–770.
Mandhasiya, D. G., Murfi, H., & Bustamam, A. (2024). The hybrid of BERT and deep learning models for Indonesian sentiment analysis. Indonesian Journal of Electrical Engineering and Computer Science, 33(1), 591–602.
Murfi, H., Theresia Gowandi, S., Ardaneswari, G., & Nurrohmah, S. (2024). BERT-based combination of convolutional and recurrent neural network for Indonesian sentiment analysis. Applied Soft Computing, 151, 111112.
Nur, M. A., Umar, N., Feng, Z., & Gani, H. (2025). Evaluation of IndoBERT and RoBERTa: Performance of Indonesian Language Transformer Models in Sentiment Classification. JIKO (Jurnal Informatika dan Komputer), 8(2), 121–127.
Olusegun, J., Chris, E., Favor, B., & Ramo, T. (2024). Aspect-Based Sentiment Analysis of Customer Reviews in E-Commerce. International Journal of Advanced Computer Science, 14(6), 112–120.
Statista. (2025). Retail Trade in Indonesia - Statistics & Facts. Statista Research Department.
Statista. (2026). Digital Commerce Transaction Value in Indonesia From 2018 to 2030. Statista Market Forecast.
Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, ?., & Polosukhin, I. (2023). Attention Is All You Need. Advances in Neural Information Processing Systems, 30, 1–11.
Xin, X., Wumaier, A., Kadeer, Z., & He, J. (2023). SSEMGAT: Syntactic and Semantic Enhanced Multi-Layer Graph Attention Network for Aspect-Level Sentiment Analysis. Applied Sciences, 13(8), 5085.
Yulianti, E., & Nissa, N. K. (2024). ABSA of Indonesian customer reviews using IndoBERT: single-sentence and sentence-pair classification approaches. Bulletin of Electrical Engineering and Informatics, 13(5), 3579–3589.
Copyright (c) 2026 Puspita Putri Nabilah, Ahmad Muklason
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-ShareAlike 4.0 International (CC-BY-SA). 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.
