中国科学院行为科学重点实验室知识库

The problem of students' cognitive differences in reading comprehension in Chinese was modeled as a binary classification problem in Machine Learning. The average score of the student group in reading comprehension was normalized and binarized to obtain the difficulty label as the dependent variable for the study. Reading comprehension was affected by the genre of the article, the number of words, the number of sentences, the average number of Chinese characters contained in each sentence, the proportion of rare characters, and so on. These features were selected as independent variables for this study. Based on the Decision Tree (DT) method and the Deep Neural Network (DNN) method, two classification models were constructed. The F1 score was 82.3 and 86.7%, respectively, for the two methods with high accuracy. Although the accuracy of the DT model was slightly lower, the model was more interpretable. The result showed the pictorial features of the difficulty in reading comprehension in Chinese for teaching.