Iterative XGBoost Algorithm and Application on Modeling the Cognitive Laws of Chinese Characters Based on Psychological Cognitive Feature Extraction

• Published in: 2022 International Symposium on Advances in Informatics, Electronics and Education (ISAIEE) pp. 587 - 591
• Main Authors: Wang, Yihan, Ren, Xiaopeng
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.12.2022
• Subjects: Chinese character; Data models; Education; Feature extraction; Iterative algorithms; Machine learning; Machine learning algorithms; personalized teaching; Psychology; XGboost
• DOI: 10.1109/ISAIEE57420.2022.00124

Chinese character learning is an important subject in primary school teaching. The survey shows that students have different degrees of difficulty in recognizing different Chinese characters. Firstly, according to the laws of pedagogy and students' cognitive psychology, this paper extracted some characteristics of Chinese characters as the independent variables of the study. Secondly, students were tested for Chinese character cognition, and the average score of each Chinese character was calculated to identify the difficulty of Chinese characters, which was used as the dependent variable of the study. Then, the cognitive difficulty classification problem of Chinese characters was modeled as a binary classification problem in machine learning, and an iterative XGBoost algorithm is proposed to complete the modeling task. When the classification model was applied to the test data set, the accuracy rate, recall rate and F1 score are 84.5%, 90.0% and 87.2% respectively. The model has good classification accuracy, and has certain reference significance for teachers to carry out personalized teaching.