Cahyani, Astri (2026) ANALISIS SENTIMEN KEPUASAN PENGGUNA APLIKASI E-WALLET DI INDONESIA MENGGUNAKAN INDOBERT DAN PEMODELAN TOPIK BERDASARKAN ULASAN GOOGLE PLAY STORE. Undergraduate thesis, UPN Veteran Jawa Timur.
![[img]](https://repository.upnjatim.ac.id/style/images/fileicons/text.png) |
Text
COVER.pdf Download (2MB) |
|
Text
BAB I.pdf Download (358kB) |
|
Text
BAB II.pdf Restricted to Repository staff only until 26 January 2028. Download (623kB) |
|
Text
BAB III.pdf Restricted to Repository staff only until 26 January 2028. Download (262kB) |
|
Text
BAB IV.pdf Restricted to Repository staff only until 26 January 2028. Download (5MB) |
|
Text
BAB V.pdf Download (228kB) |
|
Text
DAFTAR PUSTAKA.pdf Download (4MB) |
|
Text
LAMPIRAN.pdf Restricted to Repository staff only Download (2MB) |
Abstract
The growth of digital wallet services in Indonesia has generated a large volume of user reviews on Google Play Store, thus requiring an automated analysis method to understand user satisfaction levels. This study aims to assess user perceptions of three popular e-wallets, namely DANA, GoPay, and ShopeePay, using sentiment analysis based onIndoBERT and BERTopic topic modeling. A total of 42,306 reviews were collected through web scraping during the period of May 2025. After going through the pre-processing stage, the data was labeled based on rating scores and divided into 80% training data and 20% test data. The IndoBERT-base-p2 model was fine-tunedand produced an accuracy of 90.3%, with the highest F1-score in the negative(0.948), positive (0.836), and neutral (0.751) classes. The results of the analysis show that ShopeePay has the highest positive sentiment (87.82%),followed by DANA (75.91%) and GoPay (69.73%). GoPay has the highest negative sentiment (28.22%), mainly related to balances not being credited and loan feature issues. Topic modeling revealed that positive reviews were dominated by themes of ease and speed of transactions, while negative reviews were related to balance issues, account verification, and system disruptions. This study shows that IndoBERT is effective for Indonesian sentiment analysis and that BERTopic is capable of accurately identifying main topics.
| Item Type: | Thesis (Undergraduate) | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Contributors: |
|
||||||||
| Subjects: | T Technology > T Technology (General) | ||||||||
| Divisions: | Faculty of Engineering > Departement of Industrial Engineering | ||||||||
| Depositing User: | ASTRI CAHYANI | ||||||||
| Date Deposited: | 26 Jan 2026 07:28 | ||||||||
| Last Modified: | 26 Jan 2026 07:56 | ||||||||
| URI: | https://repository.upnjatim.ac.id/id/eprint/49148 |
Actions (login required)
 |
View Item |