TUMANGGOR, GAVRILA LOUISE (2024) Analisis sentimen berbasis aspek: klasifikasi ulasan produk kecantikan dengan indobert. Bachelor thesis, Universitas Pelita Harapan.
Preview
Title.pdf
Available under License Creative Commons Attribution Non-commercial Share Alike.
Download (166kB) | Preview
Preview
Abstract.pdf
Available under License Creative Commons Attribution Non-commercial Share Alike.
Download (222kB) | Preview
Preview
ToC.pdf
Available under License Creative Commons Attribution Non-commercial Share Alike.
Download (206kB) | Preview
Preview
Chapter1.pdf
Available under License Creative Commons Attribution Non-commercial Share Alike.
Download (645kB) | Preview
![Chapter2 [thumbnail of Chapter2]](http://repository.uph.edu/style/images/fileicons/text.png)
Chapter2.pdf
Restricted to Registered users only
Available under License Creative Commons Attribution Non-commercial Share Alike.
Download (438kB)
![Chapter3 [thumbnail of Chapter3]](http://repository.uph.edu/style/images/fileicons/text.png)
Chapter3.pdf
Restricted to Registered users only
Available under License Creative Commons Attribution Non-commercial Share Alike.
Download (634kB)
![Chapter4 [thumbnail of Chapter4]](http://repository.uph.edu/style/images/fileicons/text.png)
Chapter4.pdf
Restricted to Registered users only
Available under License Creative Commons Attribution Non-commercial Share Alike.
Download (1MB)
![Chapter5 [thumbnail of Chapter5]](http://repository.uph.edu/style/images/fileicons/text.png)
Chapter5.pdf
Restricted to Registered users only
Available under License Creative Commons Attribution Non-commercial Share Alike.
Download (158kB)
Preview
Bibliography.pdf
Available under License Creative Commons Attribution Non-commercial Share Alike.
Download (256kB) | Preview
![Appendices [thumbnail of Appendices]](http://repository.uph.edu/style/images/fileicons/text.png)
Appendices.pdf
Restricted to Repository staff only
Available under License Creative Commons Attribution Non-commercial Share Alike.
Download (1MB)
Abstract
In the digital era, consumer reviews of products on e-commerce platforms have
become a vital source of information for producers, sellers, and other consumers. However,
extracting meaningful insights from the vast and diverse array of reviews requires an indepth analytical approach. A key challenge in analyzing reviews is understanding user
sentiment toward specific aspects, such as product quality, seller service, and delivery
performance, rather than merely evaluating overall opinions. To address this, an AspectBased Sentiment Analysis (ABSA) method is essential, enabling the extraction of relevant
aspects from reviews and the accurate classification of sentiments associated with each
aspect.
This study proposes the implementation of the IndoBERT model to integrate the
tasks of Aspect Extraction (AE) and Aspect Sentiment Classification (ASC) for beauty
product reviews in Indonesian. IndoBERT was selected for its exceptional ability to
comprehend the context of the Indonesian language. The research dataset consists of
10,000 product reviews categorized into negative, neutral, and positive sentiments based
on review scores. The model is fine-tuned using hyperparameter configurations such as
batch size and the number of epochs to enhance accuracy. The research stages include
data preprocessing, aspect and sentiment labeling, model training, and performance
evaluation. The expected outcome is a model capable of accurately extracting review
aspects and classifying sentiments with a high level of confidence.
The model implementation achieved an accuracy of 86% and an F1-score of 0.857,
with a balanced accuracy of 88%. It demonstrated precise predictions for aspects such as
"product" (confidence score 0.98), "seller" (1.00), and "delivery" (0.86). Testing revealed
the model could process data at a rate of 33.173 samples per second and complete training
in 1.5072 seconds per iteration. These results confirm the model's effectiveness in
identifying aspects and sentiments in reviews, significantly contributing to aspect-based
sentiment analysis in the Indonesian language while supporting improvements in product
service quality and marketing strategies. / Dalam era digital, ulasan konsumen terhadap produk di platform e-commerce
menjadi sumber informasi penting bagi produsen, penjual, dan konsumen lainnya. Namun,
penggalian informasi dari ulasan yang sangat banyak dan beragam membutuhkan
pendekatan analisis yang mendalam. Tantangan utama dalam menganalisis ulasan adalah
memahami sentimen pengguna terhadap aspek spesifik, seperti kualitas produk, pelayanan
penjual, dan pengiriman, bukan hanya sekadar opini secara keseluruhan. Untuk itu,
dibutuhkan metode analisis sentimen berbasis aspek (Aspect-Based Sentiment
Analysis/ABSA) yang mampu mengekstraksi aspek-aspek relevan dalam ulasan dan
mengklasifikasikan sentimen yang terkandung pada setiap aspek secara akurat.
Penelitian ini mengusulkan penerapan model IndoBERT untuk menggabungkan
tugas Aspect Extraction (AE) dan Aspect Sentiment Classification (ASC) pada ulasan
produk kecantikan berbahasa Indonesia. IndoBERT dipilih karena keunggulannya dalam
memahami konteks bahasa Indonesia secara mendalam. Dataset penelitian mencakup
10.000 ulasan produk dengan kategori sentimen negatif, netral, dan positif berdasarkan skor
ulasan. Model dioptimalkan menggunakan teknik fine-tuning dengan konfigurasi
parameter seperti ukuran batch dan jumlah epoch yang disesuaikan untuk meningkatkan
akurasi. Tahapan penelitian meliputi praproses data, pelabelan aspek dan sentimen,
pelatihan model, dan evaluasi performa. Hasil akhir yang diharapkan adalah model yang
mampu mengekstraksi aspek-aspek ulasan secara akurat dan memberikan klasifikasi
sentimen dengan tingkat kepercayaan yang tinggi.
Implementasi model menghasilkan akurasi 86% dan F1-score 0.857, dengan
balanced accuracy mencapai 88%. Model juga menunjukkan kemampuan prediksi yang
akurat pada aspek seperti "produk" (confidence score 0.98), "penjual" (1.00), dan
"pengiriman" (0.86). Pengujian menunjukkan model dapat memproses data dengan
kecepatan 33.173 sampel per detik dan menyelesaikan pelatihan dalam waktu 1.5072 detik
per iterasi. Dengan hasil ini, model terbukti efektif dalam mengidentifikasi aspek dan
sentimen ulasan, memberikan kontribusi penting bagi analisis sentimen berbasis aspek
dalam bahasa Indonesia serta mendukung peningkatan kualitas layanan produk dan strategi
pemasaran
Item Type: | Thesis (Bachelor) |
---|---|
Creators: | Creators NIM Email ORCID TUMANGGOR, GAVRILA LOUISE NIM0108210011 nonik.gavrila13@gmail.com UNSPECIFIED |
Contributors: | Contribution Contributors NIDN/NIDK Email Contributor Samosir, Feliks V. P. NIDN0319049302 feliks.parningotan@uph.edu |
Subjects: | T Technology > T Technology (General) |
Divisions: | University Subject > Current > Faculty/School - UPH Karawaci > School of Information Science and Technology > Informatics Current > Faculty/School - UPH Karawaci > School of Information Science and Technology > Informatics |
Depositing User: | Magang Input |
Date Deposited: | 17 May 2025 02:14 |
Last Modified: | 17 May 2025 02:14 |
URI: | http://repository.uph.edu/id/eprint/68409 |