Development of an Automated Credit Transfer Prototype System Using OCR and Natural Language Processing Technologies: A Case Study of Thonburi University

Abstract

Credit transfer is a crucial process that facilitates student mobility, especially in assisting students as they transition from the high vocational certificate level to the bachelor's degree level. Currently, the verification of academic transcripts is mainly conducted manually, resulting in operational delays and a higher risk of human error. This study aims to develop a prototype for an automated web application designed to simplify the credit transfer process. The system utilizes the Tesseract Engine, which is part of optical character recognition (OCR) technology, to extract information from images of academic transcripts. Additionally, to assess the similarity between various courses, the application incorporates Natural Language Processing (NLP) techniques. Developed using Python and the Streamlit framework, the application effectively presents the analysis results alongside the initial approval statuses. Experimental findings from a sample of 100 academic transcripts of students in the Digital Media Technology program, Faculty of Science and Technology, show that the system achieves 100% accuracy in extracting course codes and grade point averages (GPA). Moreover, it accurately analyzes course similarity to recommend credit transfers; for example, courses with substantial content overlap, such as "Digital Technology for Management," received a similarity score of 92.5%, resulting in a valid recommendation for transfer approval. This system significantly decreases the workload of registrar staff and standardizes the credit transfer process, marking an important step toward the future development of a highly secure system based on blockchain technology.