The most commonly used methods in text sentiment analysis are rule-based sentiment dictionary and machine learning, with the later referring to the use of vectors to represent text followed by the use of machine learning to classify the vectors. Both methods have their limitations, including inflexibility of rules, non-prominence of sentiment words. In this paper, we design a weight distributing method combining the two methods for text sentiment analysis, by which the sentence vectors obtained can both highlight words with sentiment meanings while retaining their text information. Empirical results show that based on this new method, the accuracy rate of text sentiment analysis can reach as high as 82.1%, which means 13.9% higher than rule-based sentiment dictionary method, and 7.7% higher than TF-IDF weighting method.
This paper constructs a predictive model of student reading literacy based on data from students who participated in the Program for International Student Assessment (PISA 2018) from four provinces/municipalities of China, i.e., Beijing, Shanghai, Jiangsu and Zhejiang. We calculated the contribution of influencing factors in the model by using eXtreme Gradient Boosting (XGBoost) algorithm and sHapley additive exPlanations (SHAP) values, and get the following findings: (1) Factors that have the greatest impact on students’ reading literacy are from individual and family levels, with school-level factors taking a relative back seat. (2) The most important influencing factors at individual level are reading metacognition and reading interest. (3) The most important factors at family level are ESCS (index of economic, social and cultural status) and language environment, and dialect is negative for reading literacy, whereas proficiency in both a dialect and Mandarin plays a positive role. (4) At the school level, the most important factors are time dedicated to learning and class discipline, and we found that there is an optimal value for learning time, which suggests that reasonable learning time is beneficial, but overextended learning time may make academic performance worse instead of improving it.
Collaborative problem solving (CPS) competency is critical in the twenty-first century. The Program for International Student Assessment (PISA) launched a large-scale assessment of CPS competency for the first time in 2015. Beijing, Shanghai, Jiangsu, and Guangdong provinces in China participated the assessment and scored an average of 496, which was slightly lower than the OECD average 500 and ranked 25th among the 51 countries and economies participating in the assessment. Therefore, this research was conducted to dig into the factors predicting students’ CPS competency, and help students improve it. Most research about CPS has fallen into the construction of the CPS framework and the effectiveness of CPS; research focusing on the factors predicting CPS competency is rare. Accordingly, a hierarchical linear model (HLM) was constructed to investigate the factors predicting students’ CPS competency in the current research. The model revealed that there was a large difference of students’ CPS competency among schools. In addition, among student-level variables, gender, grade, ESCS, ICT resources, students’ attitude toward CPS, and teacher unfairness were effective in predicting students’ CPS competency; among school-level variables, school location, schools’ ESCS and the proportion of all teachers fully certified predicted students’ CPS competency positively. The findings implied that in order to enhance students’ CPS competency, CPS competency training should be permeated through all the subjects; schools should employ teachers who are fully qualified; teachers should treat each student fairly; and students should be provided with more ICT resources and etc.
School adjustment is related to the educational achievements and future career development of rural left-behind children. School support is an important institutional measure to promote the school adjustment of rural left-behind children in China. Using Structural Equation Modeling (SEM) based on CEPS data, this paper analyzed the effects of school support on rural left-behind children’s school adjustment, and found that: (1) school support effectively improves rural left-behind children’s school adjustment ability; (2) school support enhances students’ school adjustment ability by improving their educational expectation and mental health, especially the effects of teachers’ relationship support and school soft environment support are significant; (3) teacher support has a greater effect on students’ school adjustment ability than environment support, and soft environment support enhances students’ school adjustment ability by improving their mental health, while the impact of hard environment support is weaker. This study provides theoretical support for the formulation of intervention policies for rural left-behind children and offers practical evidence for understanding the relationship between school education and family education.
This article explored the influencing factors of digital reading achievement based on the PISA 2018 assessment of students' reading achievement. An integrated Random Effect-Expectation Maximization (RE-EM) regression tree model was the first constructed to address the shortcomings of traditional machine learning methods for nested data estimation and the limitations of traditional linear models in handling complex data. Our study identified the key variables for the feature selection in the integrated RE-EM regression tree model include various aspects of Meta-cognition, as well as the affective element of Joy/Liking for Reading. Notably, this study found that Meta-cognition: Assess Credibility exhibits a ceiling effect on reading achievement, where the marginal effect on reading achievement significantly diminishes at the higher variable values. Additionally, Meta-cognition: Summarizing and Joy/Liking for Reading both demonstrate an approximately S-shaped curve influence on reading achievement. These findings were discussed in critical theoretical and policy implications.
With the rise of the internet wave in the twenty-first century, adolescents’ internet usage patterns have become increasingly widespread and diverse, drawing growing attention to how these preferences influence subject-specific literacy development. Based on data from the Programme for International Student Assessment (PISA) 2022 in Hong Kong, China, this study focuses on three typical usage patterns—learning-oriented, social-oriented, and entertainment-oriented internet preferences—to examine their effects on secondary students’ mathematical literacy and associated heterogeneous impacts. Employing propensity score matching and causal forest methods for empirical analysis, the findings reveal that: (1) Learning-oriented internet preferences have a small but statistically significant positive effect on mathematical literacy; (2) Social-oriented and entertainment-oriented internet preferences exhibit significant negative effects on mathematical literacy; (3) Heterogeneity analysis indicates that learning-oriented preferences provide stronger benefits for “disadvantaged” students (those with lower family socioeconomic status and insufficient family support); (4) Social-oriented and entertainment-oriented preferences demonstrate more pronounced negative effects on “advantaged” students (those with higher family socioeconomic status and stronger family support and teacher support). These results offer new empirical evidence for understanding the relationship between internet usage preferences and mathematical literacy, with important implications for designing targeted educational interventions.