An artificial intelligence-based system for essay assessment at Nguyen Tat Thanh University

Abstract

In the context of modern higher education, manual essay grading has placed a tremendous burden on lecturers due to a huge volume of exams from large classes. This letter proposed an automated assessment system based on artificial intelligence (AI) intergrated with Google Drive for secure cloud storage and Streamlit for a friendly web interface to improve the efficiency and transparency of the learning process. To increase accuracy, pre-processing of words using TF-IDF algorithm was applied to extract core keywords from correct answers, followed by semantic comparison, cloud synchronization and intuitive interface. Results from real-world data testing proved that thanks to pre-processing, grading 30 papers was completed within 5 minutes, the deviation was reduced from (0,8 to 0,6) minutes and (300-450) minutes of grading time was saved compared to traditional grading method, thus achieving high reliability for large scale. Findings fron the present study confirmed that pre-processing method has increased accuracy by 25% by integrating Google Drive and Streamlit for comprehensive and scalable management, thereby solving Viet Nam's educational challenges.