Natural Language Processing for Subjective Exam Grading in Programming Explanations in Thai Language

In the digital era, learning programming has become an essential skill, and assessing students' understanding in this area plays a crucial role in enhancing educational quality. However, evaluating subjective programming exam responses, especially those involving conceptual explanations or procedural descriptions, remains a challenging task due to the diversity and complexity of answers. Manual grading can lead to delays and inconsistencies in scoring. This research aims to develop an automated system for evaluating Thai subjective programming exams using Natural Language Processing (NLP) techniques to assess the correctness of responses. The system is implemented in Python, utilizing the ThaiNLP library for word tokenization, alongside Machine Learning algorithms such as Naive Bayes, Decision Tree, and Support Vector Machine (SVM) for classification and prediction of answers. The experimental dataset includes responses from 120 students who answered subjective questions about Python commands. Expert testing indicated high user satisfaction with an average score of 4.05. The findings show that SVM achieved the highest accuracy at 85.33%. This system can enhance efficiency and consistency in exam evaluation, reducing grading time and contributing to standardized scoring.