A Weighted Sentiment Dataset for Indonesian Telemedicine Text Classification

The data were collected from user reviews of three telemedicine applications in Indonesia, named Alodokter, Halodoc, and KlikDokter. The dataset contains 19,680 rows of user reviews that used Indonesia language and mixed more than one language.

Each user review is annotated manually by two human annotators following annotation guidelines created by psychological expertise. This dataset consists of 6 attributes: platform (application distributor platform), app_name (application name), rating (application user satisfaction score), review (Application user review), sentiment (data label), and weight (user review complexity label). The dataset supports experiments in advanced NLP technologies, including BERT, LSTM, and LLMs, such as a user review sentiment classification task.