Elamur Research⌕
☰
home / papers / 2026 / UniDataBench: Evaluating Data Analytics Agents Across Structured and Unstructured Data

UniDataBench: Evaluating Data Analytics Agents Across Structured and Unstructured Data

Paper · 2026 · Han Weng, Zhou Liu, Yuanfeng Song et al.

Summary

In business analytics, useful conclusions often require combining several kinds of storage: relational tables, NoSQL collections, and plain-text files. Existing agent benchmarks mostly use one CSV-style source, or they test narrow questions instead of open, goal-driven analysis. The authors address this with UniDataBench, which contains 100 high-level analysis tasks across five business scenarios. Expert annotators took conclusions from real industry reports and turned them into broad analytical goals. Because the original data is usually confidential, the authors then generated synthetic data in CSV, SQLite, NoSQL and TXT formats that reproduces the reported trends. Visual checks by humans confirmed that each target insight can actually be recovered.

The authors also present ReActInsight, an agent with four main parts:

  • it builds a unified metadata graph across sources;
  • it proposes candidate join keys (a Join-Hint);
  • it breaks the goal into sub-questions;
  • it writes code that debugs itself and visualizes results adaptively.

ReActInsight beats the AgentPoirot baseline in insight quality while using fewer API calls. The benchmark matters because it is one of the first evaluations of agents that must reason across mixed data formats, as real enterprise work requires.

Key points

  • Benchmark scale: UniDataBench has 100 tasks, a file count of 223 and about 17,945 average rows. By comparison, InsightBench has 112 files and 545 rows and uses only CSV.
  • Insight taxonomy: insights fall into four types: Trend, Comparison, Extreme Value and Attribution.
  • Quality and efficiency: with GPT-4o, ReActInsight scores 0.4737 on insight quality, against 0.4361 for AgentPoirot. It needs about 6 API calls per task instead of about 20. A cascade of GPT-4o-mini and GPT-4o keeps quality at 0.4510 and lowers cost from roughly $0.15 to $0.08 per task, about 47% cheaper.
  • Ablation: removing adaptive visualization hurt most, cutting the Summary-level score from 0.5792 to 0.5434. Removing Join-Hint had the smallest effect.
  • Limitations: the authors note that building the benchmark took heavy manual effort and that 100 tasks is a small sample. It also does not cover multimodal data such as images or audio. Finally, it tests rediscovery of known patterns rather than fully open-ended exploration.

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

🎓Peking UniversityBeijing · China315 works · 43,268 citations🏢ByteDanceBeijing · China47 works · 4,339 citations

 Fields, topics and concepts

📚 Shelves 7The 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 worksBusiness, Management and Accounting · 40 worksManagement Information Systems · 9 works

Cite

Cite as Han Weng, Zhou Liu, Yuanfeng Song, Xiaoming Yin, Xing Chen and Wentao Zhang. 2026. UniDataBench: Evaluating Data Analytics Agents Across Structured and Unstructured Data. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.acl-long.1556

@inproceedings{weng2026unidatabench,
  title = {{UniDataBench: Evaluating Data Analytics Agents Across Structured and Unstructured Data}},
  author = {Han Weng, Zhou Liu, Yuanfeng Song, Xiaoming Yin, Xing Chen and Wentao Zhang},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2026},
  doi = {10.18653/v1/2026.acl-long.1556},
  url = {https://doi.org/10.18653/v1/2026.acl-long.1556}
}

References

33 references, 5 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Jincheng Bai, Zhenyu Zhang, Jennifer Zhang, and Jason Zhu. 2025. Insight agents: An llm-based multi-agent system for data insights. In Proceedings of the 48th International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 4335–4339. find
  2. Ken Black. 2023. Business statistics: for contemporary decision making. John Wiley & Sons. find
  3. Rui Ding, Shi Han, Yong Xu, Haidong Zhang, and Dongmei Zhang. 2019. Quickinsights: Quick and automatic discovery of insights from multi-dimensional data. In Proceedings of the 2019 International Conference on Management of Data, SIGMOD ’19, page 317–332, New York, NY, USA. Association for Computing… find
  4. Alex Egg, Martin Iglesias Goyanes, Friso Kingma, Andreu Mora, Leandro von Werra, and Thomas Wolf. 2025. Dabstep: Data agent benchmark for multi-step reasoning. Preprint, arXiv:2506.23719. — [arXiv:2506.23719](https://arxiv.org/abs/2506.23719) find
  5. Ahmad Ghazal, Tilmann Rabl, Minqing Hu, Francois Raab, Meikel Poess, Alain Crolotte, and Hans-Arno Jacobsen. 2013. Bigbench: Towards an industry standard benchmark for big data analytics. In Proceedings of the 2013 ACM SIGMOD international conference on Management of data, pages 1197–1208. find
  6. Xinyi He, Mengyu Zhou, Xinrun Xu, Xiaojun Ma, Rui Ding, Lun Du, Yan Gao, Ran Jia, Xu Chen, Shi Han, Zejian Yuan, and Dongmei Zhang. 2024. Text2analysis: a benchmark of table question answering with advanced data analysis and unclear queries. In Proceedings of the Thirty-Eighth AAAI Conference on Art… find
  7. Sirui Hong, Yizhang Lin, Bang Liu, Bangbang Liu, Binhao Wu, Ceyao Zhang, Danyang Li, Jiaqi Chen, Jiayi Zhang, Jinlin Wang, Li Zhang, Lingyao Zhang, Min Yang, Mingchen Zhuge, Taicheng Guo, Tuo Zhou, Wei Tao, Robert Tang, Xiangtao Lu, and 9 others. 2025. Data interpreter: An LLM agent for data science…
  8. Xueyu Hu, Ziyu Zhao, Shuang Wei, Ziwei Chai, Qianli Ma, Guoyin Wang, Xuwu Wang, Jing Su, Jingjing Xu, Ming Zhu, Yao Cheng, Jianbo Yuan, Jiwei Li, Kun Kuang, Yang Yang, Hongxia Yang, and Fei Wu. 2024. Infiagent-dabench: Evaluating agents on data analysis tasks. In The Twelfth International Conference… find
  9. Yiming Huang, Jianwen Luo, Yan Yu, Yitong Zhang, Fangyu Lei, Yifan Wei, Shizhu He, Lifu Huang, Xiao Liu, Jun Zhao, and 1 others. 2024. Da-code: Agent data science code generation benchmark for large language models. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Proce… find
  10. Pingchuan Ma, Rui Ding, Shuai Wang, Shi Han, and Dongmei Zhang. 2023. InsightPilot: An LLM-empowered automated data exploration system. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing: System Demonstrations, pages 346–352, Singapore. Association for Computat… find
  11. Veit Jahns. 2013. Data insights: new ways to visualize and make sense of data by hunter whitney. SIGSOFT Softw. Eng. Notes, 38(6):45–46. find
  12. Bodhisattwa Prasad Majumder, Harshit Surana, Dhruv Agarwal, Bhavana Dalvi Mishra, Abhijeetsingh Meena, Aryan Prakhar, Tirth Vora, Tushar Khot, Ashish Sabharwal, and Peter Clark. 2025. Discoverybench: Towards data-driven discovery with large language models. In International Conference on Learning Re…
  13. Liqiang Jing, Zhehui Huang, Xiaoyang Wang, Wenlin Yao, Wenhao Yu, Kaixin Ma, Hongming Zhang, Xinya Du, and Dong Yu. 2025. DSBench: How far are data science agents from becoming data science experts? In The Thirteenth International Conference on Learning Representations. find
  14. Yuhang Lai, Chengxi Li, Yiming Wang, Tianyi Zhang, Ruiqi Zhong, Luke Zettlemoyer, Wen-tau Yih, Daniel Fried, Sida Wang, and Tao Yu. 2023. Ds-1000: a natural and reliable benchmark for data science code generation. In Proceedings of the 40th International Conference on Machine Learning, ICML’23. JMLR… find
  15. LangChain. 2024. Pandas dataframe. https://python.langchain.com/v0.2/docs/ integrations/toolkits/pandas/. Accessed: 2024-06-03. ↗
  16. Fangyu Lei, Jixuan Chen, Yuxiao Ye, Ruisheng Cao, Dongchan Shin, Hongjin SU, Zhaoqing Suo, Hongcheng Gao, Wenjing Hu, Pengcheng Yin, Victor Zhong, Caiming Xiong, Ruoxi Sun, Qian Liu, Sida Wang, and Tao Yu. 2025. Spider 2.0: Evaluating language models on real-world enterprise text-to-SQL workflows. I… find
  17. Hongxin Li, Jingran Su, Yuntao Chen, Qing Li, and Zhaoxiang Zhang. 2023. Sheetcopilot: Bringing software productivity to the next level through large language models. In Thirty-seventh Conference on Neural Information Processing Systems. find
  18. Yang Liu, Dan Iter, Yichong Xu, Shuohang Wang, Ruochen Xu, and Chenguang Zhu. 2023. G-eval: NLG evaluation using gpt-4 with better human alignment. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 2511–2522, Singapore. Association for Computational Lin…
  19. Jinwei Lu, Yuanfeng Song, Zhiqian Qin, Haodi Zhang, Chen Zhang, and Raymond Chi-Wing Wong. 2025. Bridging the gap: Enabling natural language queries for nosql databases through text-to-nosql translation. arXiv preprint arXiv:2502.11201. — [arXiv:2502.11201](https://arxiv.org/abs/2502.11201) find
  20. Jinwei Lu, Yuanfeng Song, Chen Zhang, and Raymond Chi-Wing Wong. 2026. Multivis-agent: A multi-agent framework with logic rules for reliable and comprehensive cross-modal data visualization. Proc. ACM Manag. Data, 4(1). find
  21. OpenAI. 2024. Code interpreter. https://platform.openai.com/docs/assistants/ tools/code-interpreter. Accessed: 2024-06-03. ↗
  22. Alberto Sánchez Pérez, Alaa Boukhary, Paolo Papotti, Luis Castejón Lozano, and Adam Elwood. 2025. An LLM-based approach for insight generation in data analysis. In Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Langua… find
  23. Zhiqian Qin, Yuanfeng Song, Jinwei Lu, Yuanwei Song, Shuaimin Li, and Chen Jason Zhang. 2025. Mul-tiTEND: A multilingual benchmark for natural language to NoSQL query translation. In Findings of the Association for Computational Linguistics: ACL 2025, pages 24632–24657, Vienna, Austria. Association… find
  24. Gaurav Sahu, Abhay Puri, Juan A. Rodriguez, Amirhossein Abaskohi, Mohammad Chegini, Alexandre Drouin, Perouz Taslakian, Valentina Zantedeschi, Alexandre Lacoste, David Vázquez, Nicolas Chapados, Christopher Pal, Sai Rajeswar, and Issam H. Laradji. 2025. Insightbench: Evaluating business analytics ag… find
  25. Rajeev Sharma, Sunil Mithas, and Atreyi Kankanhalli. 2014. Transforming decision-making processes: a research agenda for understanding the impact of business analytics on organisations. European Journal of Information Systems, 23(4):433–441. find
  26. Yuanfeng Song, Xuefang Zhao, and Raymond Chi-Wing Wong. 2024. Marrying dialogue systems with data visualization: Interactive data visualization generation from natural language conversations. In Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pages 2733– 2744. find
  27. Xiangru Tang, Yiming Zong, Jason Phang, Yilun Zhao, Wangchunshu Zhou, Arman Cohan, and Mark Gerstein. 2024. Struc-bench: Are large language models good at generating complex structured tabular data? find
  28. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 2: Short Papers), pages 12–34, Mexico City, Mexico. Association for Computational Linguistics. find
  29. Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik R Narasimhan, and Yuan Cao. 2023. React: Synergizing reasoning and acting in language models. In The Eleventh International Conference on Learning Representations.
  30. Tao Yu, Rui Zhang, Kai Yang, Michihiro Yasunaga, Dongxu Wang, Zifan Li, James Ma, Irene Li, Qingning Yao, Shanelle Roman, and 1 others. 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 Conference on Empir…
  31. Ran Zhang and Mohannad Elhamod. 2025. Data-to-dashboard: Multi-agent llm framework for insightful visualization in enterprise analytics. arXiv preprint arXiv:2505.23695. — [arXiv:2505.23695](https://arxiv.org/abs/2505.23695) find
  32. Wenqi Zhang, Yongliang Shen, Weiming Lu, and Yueting Zhuang. 2024a. Data-copilot: Bridging billions of data and humans with autonomous workflow. In ICLR 2024 Workshop on Large Language Model (LLM) Agents. find
  33. Yuge Zhang, Qiyang Jiang, XingyuHan XingyuHan, Nan Chen, Yuqing Yang, and Kan Ren. 2024b. Benchmarking data science agents. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 5677–5700, Bangkok, Thailand. Association for Computat… find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/weng-2026-unidatabench-evaluating-data-analytics-agents/