中文版

cn-math-taxonomy（GitHub：https://github.com/xiaoyang111222/cn-math-taxonomy ，2026-10-18 创建，今晨抓取时已 6,223★）是中小学·高中知识图谱，仓库简介给出的规模是 1644 个知识点、2403 条先修边、6723 道题，对齐课标，附自适应前置诊断；README 标题的覆盖面已扩展到数学 + 语文 + 科学 + 英语 + 物理 + 历史 + 地理。

按 README 的说法，它对齐《义务教育数学课程标准（2022 年版 2025 年修订）》和《普通高中数学课程标准（2017 年版 2025 年修订）》的必修与选择性必修，以及义务教育与高中英语课标；数学教材为人教版、英语按人教 PEP 与 Go for it!、初中物理按人教版新版。最有特色的是诊断工具：数学答错会沿先修边从高中、初中一直回溯到具体的小学知识点；英语单独成图，不和数学连边；物理也单独成图，用到的数学写在知识点的 mathLinks 字段里。数据以 topics + dependencies 的格式开放，在线试用见 https://math.pansgo.com/ 。许可证为 CC BY-NC 4.0：个人学习、教学、研究可免费使用，商业使用（付费课程、付费 App、机构收费服务等）需单独授权。

为什么值得了解

• 对做教育类产品、知识图谱或 AI 辅导的人，这是现成的中文数据骨架：知识点、先修关系、题库三件套一次给齐，不用从零爬教材目录
• “答错回溯到小学知识点”这个诊断思路值得借鉴：补习不该从错题本身开始，而该从断掉的那条先修链开始
• 但要按作者自己的提醒来用：先修关系（哪些是 hard、哪些是 soft，甚至某条边该不该存在）都需要一线教师审核后才能用于教学；标了“待核”的内容不要当成官方原文或页码

English Version

cn-math-taxonomy (GitHub: https://github.com/xiaoyang111222/cn-math-taxonomy , created 2026-10-18, 6,223 stars at this morning’s fetch) is a knowledge graph spanning primary school through high school; the repo description puts its scale at 1,644 knowledge points, 2,403 prerequisite edges and 6,723 questions, aligned to curriculum standards, with an adaptive prerequisite diagnosis tool. The README title already extends coverage to mathematics, Chinese, science, English, physics, history and geography.

Per the README, it aligns with the Compulsory Education Mathematics Curriculum Standards (2022 edition, 2025 revision) and the Senior High School Mathematics Curriculum Standards (2017 edition, 2025 revision) for compulsory and elective-compulsory content, plus the compulsory and senior-high English standards; mathematics follows the People’s Education Press textbooks, English follows PEP and Go for it!, and junior-high physics follows the new PEP edition. Its most distinctive feature is the diagnosis tool: a wrong mathematics answer traces back along prerequisite edges from high school and junior high all the way to a specific primary-school knowledge point; English forms a separate graph not linked to mathematics, and physics is likewise separate, with the mathematics it uses recorded in each point’s mathLinks field. The data is open in topics + dependencies format, with an online demo at https://math.pansgo.com/ . It is licensed CC BY-NC 4.0: free for personal study, teaching and research, while commercial use (paid courses, paid apps, paid institutional services) needs a separate licence.

Why it’s worth knowing

• For anyone building education products, knowledge graphs or AI tutoring, this is a ready-made Chinese data skeleton: knowledge points, prerequisite relations and a question bank in one package, without scraping textbook tables of contents from scratch
• The “trace a wrong answer back to the primary-school point” diagnosis idea is worth borrowing: remediation should start from the broken link in the prerequisite chain, not from the wrong question itself
• Use it with the author’s own warning attached: prerequisite relations (which are hard, which are soft, whether an edge should exist at all) still need review by front-line teachers before classroom use, and anything marked “pending verification” should not be treated as official text or page numbers