Tag
论文研究通过聚类数学题目的语义嵌入来自动构建教育分类体系，并在 ASSISTments 数据集上评估其在深度知识追踪任务中的表现，结果显示预测性能提升约 3-5 个 AUC 百分点。
CSTutorBench is a benchmark for evaluating small language models as tutors for block-based programming, focusing on pedagogical behaviors. Preliminary results show models struggle with deeper tutoring skills like avoiding answer leakage, and prompt revision improves scores.
Researchers from Arizona State University present a framework for evaluating adaptive personalization of educational reading materials using theory-grounded simulated learners, incorporating memory models, misconception revision, and Bayesian Knowledge Tracing. Experiments across three subjects show adaptive reading significantly improved outcomes in computer science but had mixed results in chemistry and biology.