Personalized education, which focuses on individual student needs, preferences, and abilities, aligns with contemporary educational trends and is a key component of learning environments in the digital age. The rapid progress and widespread application of AI technology has greatly enhanced the effectiveness of personalized education. Supported by massive educational data and advanced technologies such as data mining, we can deeply understand and optimize students’ learning trajectories, goals, and competencies. This paper provides a comprehensive review of recent advancements in personalized educational data mining, focusing on four key areas: educational recommendation, cognitive diagnosis, knowledge tracing, and learning analysis.
Educational recommendation systems utilize historical interaction data between students and educational objects to integrate rich content such as course information, knowledge concepts, and student profiles into recommendation systems, thereby recommending courses, knowledge concepts, and educational resources that align with students’ learning needs.
Cognitive diagnosis aims at identifying students’ strengths and weaknesses to guide specific teaching interventions. It can be divided into macro and micro levels. Macro-level diagnosis assesses students’ potential skills using classical test theory (CTT) and item response theory (IRT), while micro-level diagnosis provides detailed diagnosis of students’ mastery of each knowledge unit through comprehensive analysis and the design of diagnostic functions or models.
Knowledge tracing utilizes sequence-centered techniques to establish models of student-practice interaction sequences using deep neural networks, aiming at accurately diagnosing students’ strengths and weaknesses. This process integrates learning engagement support methods to analyze students’ immediate performance and long-term learning trajectories in detail, designing indicators such as learning speed, number of attempts, and forgetting patterns. Additionally, knowledge enhancement strategies utilize the multi-dimensional information embedded in exercises to enhance comprehensive tracking and accurate prediction of learners’ knowledge states.
Learning analytics aims at comprehensively analyzing students’ behavior patterns, interaction methods, and learning habits. Behavioral analysis explores students’ behavior patterns and habits during the learning process, such as sequence mining, linear segmentation, and machine learning. Predictive analysis aims at predicting future behavior and outcomes based on students’ past and current learning engagement, such as dropout prediction and learning outcome prediction.
In the future, the development of personalized educational data mining will depend on emphasizing a deeper understanding of students’ psychological states and integrating recent advancements in explainability and multimodal learning. Explainability aims at enhancing the transparency of educational systems, ensuring that students, teachers, and decision-makers can understand the reasons for recommendations. Meanwhile, multimodal learning, by integrating different forms of data input (such as text, audio, and video), provides a more comprehensive perspective on students’ learning behaviors and cognitive processes, allowing for a deeper and more nuanced insight into learners’ psychological states.
Additionally, the continuous exploration of knowledge enhancement methods is indispensable. This includes utilizing exercise-concept graphs, problem difficulty assessment, and exercise content text analysis to enrich the understanding of learning interactions and further refine the tracking and prediction of learners’ knowledge states. Through these comprehensive methods, personalized educational data mining can not only more accurately assess students’ cognitive levels but also promote more personalized and psychologically adaptive teaching strategies, thereby maintaining students’ learning motivation and improving teaching quality and learning outcomes. Therefore, future personalized education will pay more attention to utilizing data-driven insights to build educational environments that meet students’ individual needs while effectively promoting their mental health and cognitive development.

The work titled “A Review of Data Mining in Personalized Education: Current Trends and Future Prospects”, was published on Frontiers of Digital Education (published on July 2, 2024).
DOI: 10.1007/s44366-024-0019-6