Learning patterns are crucial for predicting student dropout in educational settings, providing insights into students’ behaviors and motivations. However, existing mainstream dropout prediction models have limitations in effectively mining these learning patterns and cannot mine these learning patterns in large-scale, distributed educational datasets. In this study, we analyze the representations of mainstream models and identify their inability to capture students’ distinct learning patterns and personalized variations across courses. Addressing these challenges, our study adopts a federated learning approach, tailoring the analysis to leverage distributed data while maintaining privacy and decentralization. We introduce the Federated Learning Pattern Aware Dropout Prediction Model (FLPADPM), which utilizes a one-dimensional convolutional neural network (CNN) and a bidirectional long short-term memory (LSTM) layer within a federated learning framework. This model is designed to effectively capture nuanced learning patterns and adapt to variations across diverse educational settings. To evaluate the performance of LPADPM, we conduct an empirical evaluation using the KDD Cup 2015 and XuetangX datasets. Our results demonstrate that LPADPM outperforms state-of-the-art models in accurately predicting student dropout behavior. Furthermore, we visualize the representations generated by LPADPM, which confirm its ability to effectively mine learning patterns in different courses. Our results showcase the model’s ability to capture and analyze learning patterns across various courses and institutions within a federated learning context.

In online intelligent education systems, to offer proactive studying services to students (e.g., learning path recommendation), a crucial demand is to track students’ knowledge mastery levels over time. However, existing works ignore the impact of learning transfer on knowledge tracing and only track knowledge proficiency. Knowledge proficiency alone cannot fully reflect students’ knowledge mastery levels. A student’s knowledge structure (the similarities and differences within knowledge concepts) and abstract principle mastery level (common attributes among knowledge concepts, such as learning methods) also need to be tracked. To this end, we propose a novel multilevel Knowledge Tracing model with Learning Transfer (KTLT) to track students’ knowledge mastery levels. First, we clarify the relationships among abstract principles, knowledge structure, and knowledge proficiency by utilizing the learning transfer theory in educational psychology. Then, we associate each problem with a knowledge vector in which each element represents an explicit knowledge concept by leveraging educational priors (i.e., a Qmatrix). Correspondingly, each student is represented as a knowledge vector (knowledge proficiency) and an abstract principle vector (abstract principle mastery). Given a student’s knowledge and abstract principle vector over time, we use the learning and forgetting curve as priors to capture the student’s knowledge proficiency and abstract principle mastery level over time. Furthermore, we embed the knowledge concept by using a student’s abstract principle mastery level and obtain a personalized knowledge relevance matrix (a student’s knowledge structure) by calculating the cosine similarity among the knowledge embedding results. Finally, we design a probabilistic matrix factorization framework by combining student and problem priors for tracking a student’s knowledge mastery. Extensive experiments on two real-world datasets demonstrate both the effectiveness and explanatory ability of the KTLT.

Knowledge tracing, which is used to predict students’ performance based on their previous practices, has attracted many researchers’ attention. Especially in this rising period of intelligent education, many knowledge tracing methods have been developed. However, most of the existing knowledge tracing methods focus on the personality of practices and knowledge concepts but ignore the contexts related to the studying process. In this paper, we propose a context-based knowledge tracing model, which combines students’ historical performance and their studying contexts during knowledge mastery. To be specific, we first define five studying contexts for performance prediction. The basic context is the current knowledge state of a student, which is described by their practice sequences. Then, a QR-matrix is defined to represent the relationship among questions, knowledge concepts, and responses, which describes the contexts of questions and knowledge. Furthermore, an improved LSTM model is proposed to capture the context of students’ memory and forgetness, and a multi-head attention mechanism is designed to capture the context of students’ behaviors. Finally, based on the captured contexts, the prediction model ContextKT is established. Our prediction model is evaluated on two real educational datasets. The experimental results show our model is effective and efficient in student performance prediction, and it outperforms the other existing methods.

In order to find solutions to math word problems, some modules have been designed to check the generated expressions, but they neither take into account the symmetry between math word problems and their corresponding mathematical expressions, nor do they utilize the efficiency of pretrained language models in natural language understanding tasks. Anyway, designing fine-tuning tasks for pretrained language models that encourage cooperation with other modules to improve the performance of math word problem solvers is an unaddressed problem. To solve these problems, in this paper we propose a BERT-based model for solving math word problems with a supervision module. Based on pretrained language models, we present a fine-tuning task to predict the number of different operators in the expressions to learn the potential relationships between the problems and the expressions. Meanwhile, a supervision module is designed to check the incorrect expressions generated and improve the model’s performance by optimizing the encoder. A series of experiments are conducted on three datasets, and the experimental results demonstrate the effectiveness of our model and its component’s designs.