Analisis Sentimen Berbasis Aspek pada Ulasan Pengguna Aplikasi Brimo di Google Play Store Menggunakan IndoRoBERTa

Afkar, Mukhairatil (2026) Analisis Sentimen Berbasis Aspek pada Ulasan Pengguna Aplikasi Brimo di Google Play Store Menggunakan IndoRoBERTa. Skripsi thesis, UIN Imam Bonjol Padang.

	Text (Cover) mukhairatil afkar_2217020025_cover.pdf - Published Version Download (801kB)
	Text (BAB I) mukhairatil afkar_2217020025_bab 1.pdf - Published Version Download (255kB)
	Text (BAB III) mukhairatil afkar_2217020025_bab 3.pdf - Published Version Download (340kB)
	Text (BAB V-DAPUS) mukhairatil afkar_2217020025_bab 5 + dapus.pdf - Published Version Download (270kB)
	Text (FULL TEXT) repositori_2217020025_mukhairatil afkar_skripsi_FULLTEXT.pdf - Published Version Restricted to Repository staff only Download (5MB)

Abstract

BRImo is a mobile banking application developed by PT Bank Rakyat Indonesia (Persero) Tbk that is widely used to support digital banking transactions. User reviews available on Google Play Store can be utilized to understand users' perceptions of the application. This study aims to analyze user sentiment toward the BRImo application based on specific aspects using the Aspect Based Sentiment Analysis (ABSA) approach and the IndoRoBERTa model. The research employed the Cross Industry Standard Process for Data Mining (CRISP-DM) methodology, which consists of the stages of Business Understanding, Data Understanding, Data Preparation, Modeling, and Evaluation. The dataset was collected through web scraping from Google Play Store and consisted of 5,000 Indonesian-language reviews. After attribute selection and aspect determination using the keyword-based aspect extraction method, 4,044 reviews were successfully mapped into three aspects: interface, features and performance, and service. Sentiment labeling was then conducted based on user ratings, followed by the fine-tuning process of the IndoRoBERTa model to classify sentiments into positive, negative, and neutral categories. The evaluation results on the testing dataset showed that the model achieved an accuracy of 87.41%, precision of 85.88%, recall of 87.41%, and F1-score of 86.54%. The sentiment analysis results indicated that the service aspect obtained the highest number of positive sentiments with 1,763 reviews, followed by the interface aspect with 1,006 reviews and the features and performance aspect with 526 reviews. Meanwhile, the features and performance aspect recorded the highest number of negative sentiments with 434 reviews, indicating that users still experienced issues related to application performance and stability. Overall, BRImo received predominantly positive sentiments across all aspects, demonstrating that the IndoRoBERTa model is capable of effectively classifying sentiments and providing deeper insights into user perceptions of the BRImo application.

Actions (login required)

View Item