Rengga Prakoso Nugroho, Yerry Soepriyanto, Muhammad Tri Panunggal Aprianto, Aris Triwahyu Febriansah, Muhammad Syifa'Ul Qolbi, Khusnul Khuluq
Student feedback is a rich source of information. However, with so much feedback, a practical system is needed to provide a quick analysis of students' perceptions of their experiences. IndoBERT and Google NLP are models that can perform sentiment analysis on Indonesian language messages. There is a comparison between the two models with a dataset of feedback provided by higher education students to determine their ability to be used in a higher education context. There is a striking difference in neutral and negative sentiment between the two models, while positive sentiment produces consistent results. In general, the accuracy and precision of both models are very good, above 70 percent. The classification performance on each sentiment is also consistent but produces low recall and precision on negative and neutral sentiment. © 2024 IEEE.
Educational Technology, Teknologi Pendidikan ID, Sidoarjo, Indonesia; Educational Technology, State University of Malang, Malang, Indonesia
