An effective hybrid automated Chinese scoring system for medical education

Examination is an effective method for evaluating educational achievements but manually grading is time-consuming and error-prone. According to the characteristics of Chinese, this paper designs an effective and reliable hybrid automatic grading system for medical term interpretation questions, which integrates keyword similarity, semantic similarity and sentence length similarity. This model therefore provides a feasible framework in the sense that automated scoring system can be used reliably in real practice for subjective questions in medical education. In practice, the parameters can be pre-trained for different courses and different levels of students, so as to achieve the best performance of the model. From 15 professional noun explanation questions of 300 medical students in the course of medical psychology, statistics show that the difference between the manual and automatic assessment of all students' answers for each question is less than or equal to 2. Moreover, a comparative study of the automated scoring system is conducted to assess the overall testing performance. The experimental verification shows that the model posed in this paper is effective and robust. To enhance user convenience, a GUI software has been created using the Python “Tkinter” library. This software can be directly applied to different scenarios to demonstrate the expansive application potential of the system.