学生多项选择题响应数据集
收藏资源简介:
该数据集包含3202个多项选择题,来源于三大教育评估平台,覆盖数学、生物、物理等六个核心学术领域。每个问题都有至少50名学生的回答,错误率超过5%,仅包含四个选项的问题。数据集旨在探究大型语言模型是否能捕捉到学生在选择题中常见的错误选择模式,为教育评估工具的设计提供实证基础。
This dataset comprises 3,202 multiple-choice questions sourced from three leading educational assessment platforms, covering six core academic fields including mathematics, biology, physics and other related disciplines. Each question is accompanied by responses from at least 50 students, with an incorrect answer rate exceeding 5%, and all questions feature exactly four answer options. This dataset is designed to investigate whether large language models (LLMs) can capture the common error selection patterns exhibited by students when solving multiple-choice questions, thereby providing an empirical foundation for the design of educational assessment tools.




