TableReasoner: Advancing Table Reasoning Framework with Large Language Models

Sishi Xiong, Dakai Wang, Yu Zhao, Jie Zhang, Changzai Pan, Haowei He, Xiangyu Li, Wenhan Chang, Zhongjiang He, Shuangyong Song, Yongxiang Li · 2025

Read TableReasoner: Advancing Table Reasoning Framework with Large Language Models by Sishi Xiong, Dakai Wang, Yu Zhao, Jie Zhang, Changzai Pan, Haowei He, Xiangyu Li, Wenhan Chang, Zhongjiang He, Shuangyong Song, Yongxiang Li (2025) 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-2507-08046.txt.

Description

Sishi Xiong and co-authors build TableReasoner, a system that answers questions over messy real-world tables, in a 2025 paper. Real tables are huge, their columns are cryptically named, and entities clash. Asking a language model to eyeball the whole sheet invites hallucinated answers.

Their framework first builds a schema mixing a table's structure with its meaning. A linking plan then trims the schema to just what the question needs, clearing ambiguity before any reasoning starts. With that focused view, the model writes and runs programs to compute answers, looping through cycles of thinking and reflection. The system took first place in both subtasks of a shared table question answering challenge, showing the disciplined route beats raw prompting.

Catalog record

CREATORSishi Xiong, Dakai Wang, Yu Zhao, Jie Zhang, Changzai Pan, Haowei He, Xiangyu Li, Wenhan Chang, Zhongjiang He, Shuangyong Song, Yongxiang Li
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE43 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2507-08046

Browse the full library · About the Tower of Babel