@inproceedings{guo-etal-2026-rethinking,
    title = "Rethinking Table Pruning in {T}able{QA}: From Sequential Revisions to Gold Trajectory-Supervised Parallel Search",
    author = "Guo, Yu  and
      Ye, Shenghao  and
      Chen, Shuangwu  and
      Wen, Zijian  and
      Zhang, Tao  and
      Qirui, Bai  and
      Jin, Dong  and
      Hou, Yunpeng  and
      He, Huasen  and
      Jianyang  and
      Tan, Xiaobin",
    editor = "Liakata, Maria  and
      Moreira, Viviane P.  and
      Zhang, Jiajun  and
      Jurgens, David",
    booktitle = "Proceedings of the 64th Annual Meeting of the {A}ssociation for {C}omputational {L}inguistics (Volume 1: Long Papers)",
    month = jul,
    year = "2026",
    address = "San Diego, California, United States",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingest-acl/2026.acl-long.591/",
    pages = "12960--12976",
    ISBN = "979-8-89176-390-6",
    abstract = "Table Question Answering (TableQA) benefits significantly from table pruning, which extracts compact sub-tables by eliminating redundant cells to streamline downstream reasoning. However, existing pruning methods typically rely on sequential revisions driven by unreliable critique signals, often failing to detect the loss of answer-critical data. To address this limitation, we propose TabTrim, a novel table pruning framework which transforms table pruning from sequential revisions to gold trajectory-supervised parallel search. TabTrim derives a gold pruning trajectory using the intermediate sub-tables in the execution process of gold SQL queries, and trains a pruner and a verifier to make the step-wise pruning result align with the gold pruning trajectory. During inference, TabTrim performs parallel search to explore multiple candidate pruning trajectories and identify the optimal sub-table. Extensive experiments demonstrate that TabTrim achieves state-of-the-art performance across diverse tabular reasoning tasks: TabTrim-8B reaches 73.5{\%} average accuracy, outperforming the strongest baseline by 3.2{\%}, including 79.4{\%} on WikiTQ and 61.2{\%} on TableBench."
}