<![CDATA[ChatGPT (Generative Pre-training Transformer) is an artificial intelligence technology designed to mimic human conversation in text form and has become an important tool in various fields, including education. This study aims to analyze public sentiment toward the use of ChatGPT, which can be categorized into positive and negative sentiments. The data for the study was obtained from 5,686 user reviews on the Twitter platform, collected through Google Colaboratory and processed with pre-processing steps. The data was labeled as positive and negative, then classified using fine-tuning on the XLNet model, a Transformer-based language model. The results show that the fine-tuned XLNet model achieved an accuracy of 88.45%, a precision of 89%, a recall of 88%, and an F1-score of 89%, as measured using the Confusion Matrix. This study demonstrates that fine-tuning XLNet is effective in classifying the sentiment of ChatGPT user reviews related to education.ChatGPT (Generative Pre-training Transformer) adalah teknologi kecerdasan buatan yang dirancang untuk menirukan percakapan manusia dalam bentuk teks dan telah menjadi alat penting di berbagai bidang, termasuk pendidikan. Penelitian ini bertujuan untuk mengalisis akurasi kinerja model fine tuning XL Net dari sentimen masyarakat terhadap penggunaan ChatGPT, yang dapat dikategorikan menjadi sentimen positif dan negatif. Data penelitian diperoleh dari 5.686 ulasan pengguna di platform Twitter, dikumpulkan melalui Google Colaboratory dan diproses dengan tahap pre-processing. Data diberi label positif dan negatif, lalu diklasifikasikan menggunakan metode fine-tuning pada model XLNet, model bahasa berbasis Transformer. Hasil penelitian menunjukkan bahwa model fine-tuning XLNet mencapai akurasi 88,45%, precision 89%, recall 88%, dan F1-score 89%, yang diukur menggunakan Confusion Matrix. Penelitian ini membuktikan bahwa fine-tuning XLNet efektif dalam mengklasifikasikan sentimen ulasan pengguna ChatGPT terkait pendidikan]]>
Copyrights © 2025
