<![CDATA[Abstrak - Perkembangan teknologi terus mendukung produktivitas masyarakat, salah satunya melalui beralihnya kebiasaan mencatat manual ke aplikasi manajemen tugas seperti Notion. Namun, tingginya volume ulasan pengguna di Google Play Store menghasilkan sentimen yang sangat beragam, sehingga menyulitkan pengembang untuk mengidentifikasi aspek yang perlu dioptimasi secara cepat. Penelitian ini bertujuan untuk menganalisis sentimen pengguna aplikasi Notion pada Google Play Store guna mengklasifikasikan ulasan ke dalam dua kategori, yaitu sentimen positive dan negative. Penelitian ini menerapkan model IndoBERT menggunakan 718 data ulasan yang diperoleh dari Google Play Store, kemudian dipilih 456 data dengan kategori sentimen positive dan negative untuk analisis sentimen melalui tahap pengumpulan data, pelabelan data secara manual dengan label sentimen positive dan negative, preprocessing dengan cleaning data dan data transformation (case folding dan stopword removal), analisis sentimen dengan model IndoBERT, visualisasi Word Cloud dan evaluasi mode dengan confusion matriks dan metrik evaluasi akurasi, precision, recall dan F1-Score. Hasil evaluasi menunjukkan bahwa model mampu mengklasifikasikan kategori ulasan dengan tingkat akurasi sebesar 95%, precision pada data sentimen negative dan positive sebesar 92% dan 97% , recall untuk sentimen negative dan positive sebesar 89% dan 97%, dan F1-Score pada sentimen negative dan positive sebesar 90% dan 97%. Dengan demikian, IndoBERT disimpulkan dapat menjadi metode yang efektif dalam analisis sentimen ulasan aplikasi digital. Hasil ini juga dapat menjadi acuan bagi tim pengembang dalam melakukan optimasi aplikasi. Kata kunci: Analisis Sentimen; IndoBERT; Google Play Store; Notion; Abstract - Technological advancements continue to support public productivity, one of which is demonstrated by the shift from manual note-taking habits to task management applications like Notion. However, the high volume of user reviews on the Google Play Store generates highly diverse sentiments, making it challenging for developers to quickly identify areas that require optimization. This study aims to analyze user sentiment toward the Notion application on the Google Play Store to classify reviews into two categories: positive and negative sentiments. This study implements the IndoBERT model using 718 review data points obtained from the Google Play Store. From this dataset, 456 reviews categorized under positive and negative sentiments were selected for sentiment analysis. The methodology involves data collection, manual data labeling into positive and negative sentiment categories, preprocessing (including data cleaning and data transformation through case folding and stopword removal), sentiment analysis using the IndoBERT model, Word Cloud visualization, and model evaluation utilizing a confusion matrix alongside evaluation metrics such as accuracy, precision, recall, and F1-Score. The evaluation results demonstrate that the model is capable of classifying review categories with an accuracy rate of 95%. The precision for negative and positive sentiments is 92% and 97%, respectively; the recall for negative and positive sentiments is 89% and 97%, respectively; and the F1-Score for negative and positive sentiments is 90% and 97%, respectively. Consequently, it is concluded that IndoBERT can serve as an effective method for sentiment analysis of digital application reviews. These findings can also serve as a reference for development teams in optimizing the application. Keywords: sentiment analysis; IndoBERT; Google Play Store; Notion;]]>
