Aspect Based Sentiment Analysis on Access by KAI Application

This dataset contains Indonesian-language user reviews of the Access by KAI mobile application collected from the Google Play Store. The dataset was developed for sentiment analysis research using the IndoBERT model. A total of 6,740 user reviews were collected between October 2024 and October 2025, with 787 manually labeled instances categorized into positive and negative sentiments. The reviews were obtained through a web scraping process using the “Most Relevant” and “Newest” sorting options to ensure a balanced mix of popular and recent feedback. The preprocessing pipeline includes text cleaning, normalization, and the removal of punctuation, emojis, and non-informative tokens. This dataset is suitable for Natural Language Processing (NLP) research in Indonesian, particularly for sentiment classification, text mining, and language model fine-tuning tasks.