MathDoc: Benchmarking Structured Extraction and Active Refusal on Noisy Mathematics Exam Papers
Chenyue Zhou, Jiayi Tuo, Shitong Qin, Wei Dai, Mingxuan Wang, Ziwei Zhao Duoyang Li, Shiyang Su, Yanxi Lu, Yanbiao Ma · 2026
Read MathDoc: Benchmarking Structured Extraction and Active Refusal on Noisy Mathematics Exam Papers by Chenyue Zhou, Jiayi Tuo, Shitong Qin, Wei Dai, Mingxuan Wang, Ziwei Zhao Duoyang Li, Shiyang Su, Yanxi Lu, Yanbiao Ma (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-10104.txt.
Description
Chenyue Zhou and co authors test how well AI reads math exam sheets. Exams on paper hold the base of smart school tools. A tool must pull each item out in a clean form. Real sheets are worn, marked, and blurred. That noise breaks many tools.
The team built MathDoc, a test set of noisy exam sheets. It checks two skills at once. One is clean pull out of whole items. The next is to say no to a sheet that lacks parts. A tool that fills gaps with guesses can harm grades. The set thus checks care as well as skill.
Catalog record
| CREATOR | Chenyue Zhou, Jiayi Tuo, Shitong Qin, Wei Dai, Mingxuan Wang, Ziwei Zhao Duoyang Li, Shiyang Su, Yanxi Lu, Yanbiao Ma |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 36 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-10104