Monica, Nadin Isna (2026) Analisis Sentimen Berbasis Aspek Papa Ulasan Aplikasi JConnect Mobile Menggunakan Fine-Tuning IndoBERT. Undergraduate thesis, UPN Veteran Jawa Timur.
![[img]](https://repository.upnjatim.ac.id/style/images/fileicons/text.png) |
Text (Cover)
22082010118.-cover.pdf Download (2MB) |
|
Text (Bab 1)
22082010118.-bab1.pdf Download (97kB) |
|
Text (Bab 2)
22082010118.-bab2.pdf Restricted to Repository staff only until 17 July 2029. Download (423kB) |
|
Text (Bab 3)
22082010118.-bab3.pdf Restricted to Repository staff only until 17 July 2029. Download (4MB) |
|
Text (Bab 4)
22082010118.-bab4.pdf Restricted to Repository staff only until 17 July 2029. Download (7MB) |
|
Text (Bab 5)
22082010118.-bab5.pdf Download (76kB) |
|
Text (Daftar Pustaka)
22082010118.-daftarpustaka.pdf Download (152kB) |
|
Text (Lampiran)
22082010118.-lampiran.pdf Restricted to Repository staff only Download (295kB) |
Abstract
Digital transformation in the banking sector drives banks to continuously improve the quality of digital services through mobile banking applications. Although the number of JConnect Mobile users continues to increase, challenges related to user experience remain, as reflected in user reviews and the application's relatively lower rating compared with other regional development bank mobile banking applications. This condition aligns with the Bank Jatim Information Technology Strategic Plan (RSTI) 2027, namely Customer Empowerment through AI/ML. Therefore, sentiment analysis of user reviews is needed to understand users' perceptions, needs, and concerns more comprehensively. This study aimed to perform aspect-based sentiment analysis on JConnect Mobile user reviews using a fine-tuned IndoBERT model and to compare its performance with a fine-tuned mBERT model as the baseline. The study used JConnect Mobile user reviews collected from Google Play Store and App Store between 2021 and 2025. The research stages included literature review, requirements analysis, data collection, data labeling, data preprocessing, data splitting, feature extraction, model development, model evaluation, and model implementation. The experiments were conducted using 5-fold cross-validation with 24 hyperparameter combination scenarios to obtain the optimal model configuration. The results showed that the fine-tuned IndoBERT model achieved the best performance, with mean Macro F1-Scores of 0.86 for the interface aspect, 0.83 for the features and performance aspect, and 0.74 for the service aspect. Furthermore, IndoBERT consistently outperformed mBERT across all aspects because it had been pre-trained on an Indonesian-language corpus, enabling it to better capture the linguistic characteristics, vocabulary, and contextual nuances of Indonesian user reviews. The aspect-based sentiment analysis model was successfully implemented in a web-based system using the Flask framework, which was able to perform both single and batch review analysis and provided an interactive and informative dashboard for visualizing the analysis results.
| Item Type: | Thesis (Undergraduate) | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Contributors: |
|
||||||||||||
| Subjects: | Q Science > QA Mathematics > QA76.6 Computer Programming T Technology > T Technology (General) T Technology > T Technology (General) > T58.6-58.62 Management Information Systems |
||||||||||||
| Divisions: | Faculty of Computer Science > Departemen of Information Systems | ||||||||||||
| Depositing User: | Nadin Isna Monica | ||||||||||||
| Date Deposited: | 17 Jul 2026 07:34 | ||||||||||||
| Last Modified: | 17 Jul 2026 07:34 | ||||||||||||
| URI: | https://repository.upnjatim.ac.id/id/eprint/55763 |
Actions (login required)
 |
View Item |