Elamur Research⌕
☰
home / papers / 2026 / FACTS: Table Summarization via Offline Template Generation with Agentic Workflows

FACTS: Table Summarization via Offline Template Generation with Agentic Workflows

Paper · 2026 · Ye Yuan, Mohammad Amin Shabani and Siqi Liu

Summary

FACTS addresses query-focused table summarization, where a system writes a narrative answer to a user's request over one or more tables. Existing approaches each have drawbacks. Fine-tuned table-to-text models handle numerical logic poorly. Prompting an LLM with the full table is slow, runs into context limits and exposes confidential records. Earlier agentic pipelines depend on fragile plans or hand-crafted templates.

The authors instead use an agentic workflow that builds a reusable "offline template" from the table schema alone. It works in three steps:

  • derive guided questions and filtering rules that clarify the query;
  • write SQL to pull the needed values;
  • produce a Jinja2 template that turns the SQL output into prose.

An ensemble of models, the LLM Council, reviews and revises each step. Because execution happens locally, cell values never reach the LLM. One template also serves every table that shares the schema.

On FeTaQA, QTSumm and QFMTS, FACTS outperforms prompt-based and agentic baselines. This makes it attractive for regulated domains such as finance or healthcare, which need recurring, verifiable summaries.

Key points

  • The final artifact combines SQL queries with a Jinja2 template. Only the schema and the user query are sent to the LLM; raw rows stay local. Code is published at github.com/BorealisAI/FACTS.
  • Error analysis puts SQL execution accuracy near 94% and rendering accuracy near 98%. Together these give roughly 92% factual correctness of the generated summaries.
  • In a blind side-by-side human study on QFMTS against SPaGe, evaluators preferred FACTS 55% of the time on completeness, with preference rising to 59% on correctness and 60% on freedom from hallucination.
  • On 100 QTSumm samples, templates applied unchanged to GPT-5-paraphrased queries kept similar quality (BLEU 21.8, ROUGE-L 43.5, METEOR 50.8). Runtime stayed flat as tables grew from 100 to 1000 rows, while Reason-then-Summ and SPaGe grew slower.
  • Stated limitations:
    • Templates break if the schema changes.
    • The schema and query themselves are not obfuscated, which may be unacceptable in zero-trust settings.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓McGill UniversityMontreal · Canada91 works · 11,178 citations🔬Mila - Quebec Artificial Intelligence InstituteMontreal · Canada65 works · 14,425 citations

 Fields, topics and concepts

📚 Shelves 5The field of science this work belongs to (OpenAlex fields).Computer Science · 18,614 worksArtificial Intelligence · 17,965 worksArts and Humanities · 17,450 worksLanguage and Linguistics · 17,296 worksComputer Networks and Communications · 100 works

🧭 Topics 1Research subjects and languages the work is about.Summarization · 243 works

Cite

Cite as Ye Yuan, Mohammad Amin Shabani and Siqi Liu. 2026. FACTS: Table Summarization via Offline Template Generation with Agentic Workflows. Findings of the Association for Computational Linguistics: ACL 2026.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.findings-acl.453

@inproceedings{yuan2026facts,
  title = {{FACTS: Table Summarization via Offline Template Generation with Agentic Workflows}},
  author = {Ye Yuan, Mohammad Amin Shabani and Siqi Liu},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
  doi = {10.18653/v1/2026.findings-acl.453},
  url = {https://doi.org/10.18653/v1/2026.findings-acl.453}
}

References

28 references, 15 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Satanjeev Banerjee and Alon Lavie. 2005. METEOR: An automatic metric for MT evaluation with improved correlation with human judgments. In Proceedings of the ACL Workshop on Intrinsic and Extrinsic Evaluation Measures for Machine Translation and/or Summarization, pages 65–72, Ann Arbor, Michigan. Ass…
  2. Tom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, Sandhini Agarwal, Ariel Herbert-Voss, Gretchen Krueger, Tom Henighan, Rewon Child, Aditya Ramesh, Daniel M. Ziegler, Jeffrey Wu, Clemens Winter,… — Language Models are Few-Shot Learners
  3. Wenhu Chen, Jianshu Chen, Yu Su, Zhiyu Chen, and William Yang Wang. 2020. Logical natural language generation from open-domain tables. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 7929–7942, Online. Association for Computational Linguistics.
  4. Zhoujun Cheng, Haoyu Dong, Ran Jia, Pengfei Wu, Shi Han, Fan Cheng, and Dongmei Zhang. 2022a. FORTAP: Using formulas for numerical-reasoning-aware table pretraining. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 1150– 1166,… find
  5. Zhoujun Cheng, Haoyu Dong, Zhiruo Wang, Ran Jia, Jiaqi Guo, Yan Gao, Shi Han, Jian-Guang Lou, and Dongmei Zhang. 2022b. HiTab: A hierarchical table dataset for question answering and natural language generation. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguisti…
  6. Zhoujun Cheng, Tianbao Xie, Peng Shi, Chengzu Li, Rahul Nadkarni, Yushi Hu, Caiming Xiong, Dragomir Radev, Mari Ostendorf, Luke Zettlemoyer, Noah A. Smith, and Tao Yu. 2023. Binding language models in symbolic languages. In The Eleventh International Conference on Learning Representations, ICLR 2023…
  7. Mohit Iyyer, Wen-tau Yih, and Ming-Wei Chang. 2017. Search-based neural structured learning for sequential question answering. In Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 1821– 1831, Vancouver, Canada. Association for Comp…
  8. Zhengbao Jiang, Yi Mao, Pengcheng He, Graham Neubig, and Weizhu Chen. 2022. OmniTab: Pretraining with natural and synthetic data for few-shot table-based question answering. In Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human La… find
  9. Rémi Lebret, David Grangier, and Michael Auli. 2016. Neural text generation from structured data with application to the biography domain. In Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, pages 1203–1213, Austin, Texas. Association for Computational Linguist…
  10. Chin-Yew Lin and Eduard Hovy. 2003. Automatic evaluation of summaries using n-gram co-occurrence statistics. In Proceedings of the 2003 Human Language Technology Conference of the North American Chapter of the Association for Computational Linguistics, pages 150–157.
  11. Ao Liu, Haoyu Dong, Naoaki Okazaki, Shi Han, and Dongmei Zhang. 2022a. PLOG: Table-to-logic pre-training for logical table-to-text generation. In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, pages 5531– 5546, Abu Dhabi, United Arab Emirates. Association for… find
  12. Qian Liu, Bei Chen, Jiaqi Guo, Morteza Ziyadi, Zeqi Lin, Weizhu Chen, and Jian-Guang Lou. 2022b. TAPEX: table pre-training via learning a neural SQL executor. In The Tenth International Conference on Learning Representations, ICLR 2022, Virtual Event, April 25-29, 2022. OpenReview.net.
  13. Nafise Sadat Moosavi, Andreas Rücklé, Dan Roth, and Iryna Gurevych. 2021. Scigen: a dataset for reasoning-aware text generation from scientific tables. In Thirty-fifth Conference on Neural Information Processing Systems Datasets and Benchmarks Track (Round 2). find
  14. Linyong Nan, Chiachun Hsieh, Ziming Mao, Xi Victoria Lin, Neha Verma, Rui Zhang, Wojciech Kryściński, Hailey Schoelkopf, Riley Kong, Xiangru Tang, Mutethia Mutuma, Ben Rosand, Isabel Trindade, Renusree Bandaru, Jacob Cunningham, Caiming Xiong, Dragomir Radev, and Dragomir Radev. 2022. FeTaQA: Free…
  15. Giang Nguyen, Ivan Brugere, Shubham Sharma, Sanjay Kariyappa, Anh Totti Nguyen, and Freddy Lecue. 2025. Interpretable LLM-based table question answering. Transactions on Machine Learning Research. find
  16. Kishore Papineni, Salim Roukos, Todd Ward, and Wei-Jing Zhu. 2002. Bleu: a method for automatic evaluation of machine translation. In Proceedings of the 40th Annual Meeting of the Association for Computational Linguistics, pages 311–318, Philadelphia, Pennsylvania, USA. Association for Computational…
  17. Ankur Parikh, Xuezhi Wang, Sebastian Gehrmann, Manaal Faruqui, Bhuwan Dhingra, Diyi Yang, and Dipanjan Das. 2020. ToTTo: A controlled table-to-text generation dataset. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 1173–1186, Online. Associat…
  18. Panupong Pasupat and Percy Liang. 2015. Compositional semantic parsing on semi-structured tables. In Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing (Volume 1: Long Papers), pages 1470–…
  19. Mark Raasveldt and Hannes Mühleisen. 2019. Duckdb: an embeddable analytical database. In Proceedings of the 2019 International Conference on Management of Data, SIGMOD Conference 2019, Amsterdam, The Netherlands, June 30 - July 5, 2019, pages 1981– 1984. Acm. find
  20. Lya Hulliyyatus Suadaa, Hidetaka Kamigaito, Kotaro Funakoshi, Manabu Okumura, and Hiroya Takamura. 2021. Towards table-to-text generation with numerical reasoning. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference o… find
  21. Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Brian Ichter, Fei Xia, Ed H. Chi, Quoc V. Le, and Denny Zhou. 2022. Chain-of-thought prompting elicits reasoning in large language models. In Advances in Neural Information Processing Systems 35: Annual Conference on Neural Information Processi…
  22. Yunhu Ye, Binyuan Hui, Min Yang, Binhua Li, Fei Huang, and Yongbin Li. 2023. Large language models are versatile decomposers: Decomposing evidence and questions for table-based reasoning. In Proceedings of the 46th International ACM SIGIR Conference on Research and Development in Information Retriev… find
  23. Tao Yu, Rui Zhang, Kai Yang, Michihiro Yasunaga, Dongxu Wang, Zifan Li, James Ma, Irene Li, Qingning Yao, Shanelle Roman, Zilin Zhang, and Dragomir Radev. 2018. Spider: A large-scale human-labeled dataset for complex and cross-domain semantic parsing and text-to-SQL task. In Proceedings of the 2018…
  24. Weijia Zhang, Songgaojun Deng, and Evangelos Kanoulas. 2025. Beyond natural language plans: Structure-aware planning for query-focused table summarization. find
  25. Weijia Zhang, Vaishali Pal, Jia-Hong Huang, Evangelos Kanoulas, and Maarten de Rijke. 2024. Qfmts: Generating query-focused summaries over multi-table inputs. find
  26. Yilun Zhao, Lyuhao Chen, Arman Cohan, and Chen Zhao. 2024. TaPERA: Enhancing faithfulness and interpretability in long-form table QA by content planning and execution-based reasoning. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers),… find
  27. Yilun Zhao, Linyong Nan, Zhenting Qi, Rui Zhang, and Dragomir Radev. 2022. ReasTAP: Injecting table reasoning skills during pre-training via synthetic reasoning examples. In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, pages 9006–9018, Abu Dhabi, United Ara… find
  28. Yilun Zhao, Zhenting Qi, Linyong Nan, Boyu Mi, Yixin Liu, Weijin Zou, Simeng Han, Ruizhe Chen, Xiangru Tang, Yumo Xu, Dragomir Radev, and Arman Cohan. 2023. QTSumm: Query-focused summarization over tabular data. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processin… find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/yuan-2026-facts-table-summarization-offline-template/