cn-math-taxonomy: Turning K-9 math into an AI diagnostic graph
Developer open-sourced cn-math-taxonomy, a structured curriculum graph mapping Chinese K-9 math into 531 concepts, 909 prerequisites, and 2,271 diagnostic questions. Instead of generic chat answers, AI tools can now trace learning gaps across grades along prerequisite chains. Why it matters: It turns standard AI homework helpers into systematic learning planners. Free for non-commercial use.
cn-math-taxonomy is an open-source knowledge graph project designed to help AI tutoring tools trace learning gaps and map out personalized study paths for K-9 mathematics.
Most LLM-based homework helpers today can explain a problem step by step, but they have no idea why a student fails to grasp it in the first place. If a middle schooler struggles with quadratic equations, the root cause might trace back to elementary fraction division. cn-math-taxonomy fills this gap by turning the curriculum into a computable dependency network.
Mapping 9 years of math into 531 nodes
The repository aligns directly with China's 2022 national curriculum standards and standard PEP textbooks. It decomposes grades 1 to 9 mathematics into 531 distinct concept nodes, 909 prerequisite dependencies, and 2,271 diagnostic questions.
The core mechanism is its prerequisite chain. Every concept node defines what a student must master before learning it. When a student fails a diagnostic prompt, the algorithm follows the dependency edges backward, pinpointing exact knowledge gaps across grades.
The project formats its data around topics and dependencies inspired by Marble os-taxonomy, published as clean JSON arrays. Beyond K-9 math, the repo also drafts initial maps for high school math, physics, English, and biology. Diagnostic questions for elementary and middle school math have undergone programmatic verification.
How to access it and current limits
Developers can find the project on GitHub under xiaoyang111222/cn-math-taxonomy. The repository includes topics.json, dependencies.json, standard mapping documents, and a GitHub Pages web demo showing adaptive diagnostics in action.
It is released under the CC BY-NC 4.0 license. It is free for personal study, teaching, and research, while commercial use in paid apps or institutional services requires a separate license from the author.
Keep in mind that while math questions include automated check scripts, the author notes that concept granularity and prerequisite edges still require review by frontline teachers before production deployment. Some textbook alignments for recent editions also remain under ongoing verification.
Why it matters
For edtech developers, this project eliminates months of tedious curriculum digitization. Instead of hiring subject matter experts to handcraft prerequisite trees from scratch, you can plug this structured dataset directly into LLM agent workflows.
Generic prompt engineering cannot reliably diagnose foundational learning gaps. Combining deterministic prerequisite graphs with generative AI gives homework assistants a clear road map. It is a solid foundation for turning simple question-and-answer bots into genuine adaptive learning planners.
Sources (1)
- github.com(opens in a new tab)Originalgithub.com