作文智能评阅与过程性成长追踪数据集

本数据集基于学生作文原稿及教师详细批改痕迹（圈点、评语、打分、最终得分），融合专家标注的维度标签（立意、结构、素材、语言、文采），经过系统化深度加工形成。每条记录不仅包含作文文本与教师批注，更衍生出多维错误类型谱系、亮点分层分类、交叉维度贡献值（如“立意实现度-语言贡献值”）以及可执行的修改规则，实现了从离散作文样本到可训练AI评阅模型的高质量数据资产的创造性转变。数据集直接支撑作文自动评分、病句识别、亮点分析及个性化修改建议生成，为智慧教育提供了可直接调用的核心数据引擎。This dataset is developed based on original student compositions, detailed teacher feedback traces (including annotations, comments, scoring marks and final grades), and expert-annotated dimension tags (central theme, structure, supporting materials, language expression and literary grace), through systematic in-depth processing. Each record not only includes the composition text and teacher's annotations, but also generates a multi-dimensional error type taxonomy, hierarchical classification of highlights, cross-dimensional contribution values (e.g., "theme realization degree - language contribution value"), and executable revision guidelines, realizing a creative transformation from discrete composition samples to high-quality data assets suitable for training AI essay grading models. This dataset directly supports automatic essay scoring, grammatical error identification, highlight analysis and personalized revision suggestion generation, serving as a directly callable core data engine for smart education.

搜集汇总

背景与挑战

背景概述

该数据集基于学生作文原稿和教师详细批改痕迹，融合专家标注的立意、结构、素材、语言、文采等维度标签，经过系统化深度加工形成。它不仅包含作文文本与批注，还衍生出多维错误类型谱系、亮点分层分类和交叉维度贡献值，支持作文自动评分、病句识别、亮点分析及个性化修改建议生成。数据集为智慧教育提供了可直接调用的核心数据引擎，实现了从离散样本到可训练AI评阅模型的高质量数据资产转变。