理科解题数据集

理科解题数据集是一个包含54182个常见类型理科解题数据的庞大集合，涵盖了初高中数学题和物理题。其中，数学题数量为 43588 个，包括代数题、应用题和几何题等；物理题数量为10594 个。该数据集旨在为自动解题和评分算法的研究提供丰富的训练和测试样本，以推动理科题目自动解题和评分技术的发展，特别是在复杂多变的初高中应用题和几何题领域，突破现有算法在多模态信息联合建模和多步推理方面的局限性。The Science Problem-Solving Dataset is a large collection consisting of 54,182 common-type science problem-solving samples, covering middle and high school mathematics and physics problems. Of these, 43,588 are mathematics problems, covering algebraic problems, word problems, geometry problems and other types; the other 10,594 are physics problems. This dataset is designed to provide rich training and test samples for research on automatic problem-solving and scoring algorithms, so as to advance the development of automatic problem-solving and scoring technologies for science subject problems, particularly in the domain of complex and variable middle and high school word problems and geometry problems, and break through the limitations of existing algorithms in joint modeling of multi-modal information and multi-step reasoning.

搜集汇总

背景与挑战

背景概述

理科解题数据集是一个包含54182个初高中数学和物理题目的集合，其中数学题43588个、物理题10594个。该数据集旨在为自动解题和评分算法的研究提供训练和测试样本，以推动相关技术在多模态信息联合建模和多步推理方面的发展。