The field of intelligent tutoring systems is moving towards more personalized and adaptive learning experiences. Researchers are exploring new methods to improve student modeling and exercise recommendation, such as learning auxiliary concepts and preventing label leakage. These innovations have the potential to enhance student learning outcomes and provide more effective feedback.

Noteworthy papers include: Representation Learning of Auxiliary Concepts for Improved Student Modeling and Exercise Recommendation, which proposes a deep learning model to learn sparse binary representations of exercises. Enhancing Knowledge Tracing through Leakage-Free and Recency-Aware Embeddings, which introduces a solution to prevent label leakage and improve prediction accuracy. Beyond prior knowledge: The predictive role of knowledge-building in Tutor Learning, which investigates the role of knowledge-building in mediating the relationship between procedural and conceptual learning. Skill-based Explanations for Serendipitous Course Recommendation, which develops a deep learning-based concept extraction model to improve course recommendation.