SE-Eval: A Domain-Specific Rubric-Guided Dataset for Automated Essay Evaluation in Engineering Education

SE-Eval, a large-scale, rubric-based essay dataset comprising over 10,000 responses written by third-year engineering students to prompts covering core concepts in Software Engineering. Each essay is manually scored by three trained raters across five trait dimensions Content, Coherence, Clarity, Structure, and Language, along with an overall holistic score. Essays were collected under standardized academic conditions using a controlled online assessment platform, ensuring consistency and authenticity of submissions.SE-Eval是一款大规模、基于评分细则（rubric）的作文数据集，包含一万余份由工科三年级学生针对覆盖软件工程（Software Engineering）核心概念的提示任务撰写的答卷。每篇作文均由三名经过专业培训的评分员，从内容（Content）、连贯性（Coherence）、清晰度（Clarity）、结构（Structure）与语言（Language）五个特质维度开展人工评分，并同步给出整体综合得分（overall holistic score）。所有答卷均在标准化学术环境下，通过受控在线评估平台完成收集，确保了提交作品的一致性与真实性。