Towards Robust Real-World Spreadsheet Understanding with Multi-Agent Multi-Format Reasoning

Houxing Ren, Mingjie Zhan, Zimu Lu, Ke Wang, Yunqiao Yang, Haotian Hou, Hongsheng Li · 2026

Read Towards Robust Real-World Spreadsheet Understanding with Multi-Agent Multi-Format Reasoning by Houxing Ren, Mingjie Zhan, Zimu Lu, Ke Wang, Yunqiao Yang, Haotian Hou, Hongsheng Li (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-2604-12282.txt.

Description

Ren and six colleagues present SpreadsheetAgent, a multi-agent system in two stages for understanding real-world spreadsheets that treats tables as more than plain text. It reads localized regions step by step across code results, images, and LaTeX tables. It first sketches structure and row and column summaries, then reasons over that sketch with a verification module checking extracted structures. With GPT-OSS-120B it scores 38.16 percent on Spreadsheet Bench, beating the ChatGPT Agent baseline by 2.89 points.

Catalog record

CREATORHouxing Ren, Mingjie Zhan, Zimu Lu, Ke Wang, Yunqiao Yang, Haotian Hou, Hongsheng Li
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE60787
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.12282

Browse the full library · About the Tower of Babel