UNITQA: A Unified Automated Tabular Question Answering System with Multi-Agent Large Language Models

Jun Peng Zhu; Peng Cai; Kai Xu; Li Li; Yishen Sun; Shuai Zhou; Haihuang Su; Liu Tang; Qi Liu

doi:10.1145/3722212.3725140

UNITQA: A Unified Automated Tabular Question Answering System with Multi-Agent Large Language Models

Jun Peng Zhu
, Peng Cai^*
, Kai Xu^*
, Li Li
, Yishen Sun
, Shuai Zhou
, Haihuang Su
, Liu Tang
, Qi Liu

^*Corresponding author for this work

School of Data Science and Engineering

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

1 Scopus citations

Abstract

Automated tabular question answering (TQA) has attracted significant attention in data analysis and natural language processing communities due to its powerful capabilities. The emergence of large language models (LLMs) has initiated a paradigm shift in this field. However, existing state-of-the-art approaches cannot generally operate on multiple tables from multiple heterogeneous systems, and the answer accuracy is insufficient to meet the demands of the industrial field. This paper presents UNITQA, a unified automated tabular question-answering system through multi-agent LLMs. First, UNITQA offers a user-friendly GUI interface that enables users to use natural language questions to execute TQA tasks. Second, UNITQA consists of five agents who collaborate to complete user-specified tasks. To efficiently orchestrate different agents, UNITQA utilizes a dynamic agent scheduling algorithm based on a finite-state machine. Third, UNITQA integrates a series of data connectors that allow UNITQA to access various tables from multiple heterogeneous systems. We have implemented and deployed UNITQA in numerous production environments and have demonstrated its usability and efficiency in representative real-world scenarios.

Original language	English
Title of host publication	SIGMOD-Companion 2025 - Companion of the 2025 International Conference on Management of Data
Editors	Amol Deshpande, Ashraf Aboulnaga, Babak Salimi, Badrish Chandramouli, Bill Howe, Boon Thau Loo, Boris Glavic, Carlo Curino, Daisy Zhe Wang, Dan Suciu, Daniel Abadi, Divesh Srivastava, Eugene Wu, Faisal Nawab, Ihab Ilyas, Jeffrey Naughton, Jennie Rogers, Jignesh Patel, Joy Arulraj, Jun Yang, Karima Echihabi, Kenneth Ross, Khuzaima Daudjee, Laks Lakshmanan, Minos Garofalakis, Mirek Riedewald, Mohamed Mokbel, Mourad Ouzzani, Oliver Kennedy, Oliver Kennedy, Paolo Papotti, Peter Alvaro, Peter Bailis, Renee Miller, Senjuti Basu Roy, Sergey Melnik, Stratos Idreos, Sudeepa Roy, Theodoros Rekatsinas, Viktor Leis, Wenchao Zhou, Wolfgang Gatterbauer, Zack Ives
Publisher	Association for Computing Machinery
Pages	279-282
Number of pages	4
ISBN (Electronic)	9798400715648
DOIs	https://doi.org/10.1145/3722212.3725140
State	Published - 22 Jun 2025
Event	2025 ACM SIGMOD/PODS International Conference on Management of Data, SIGMOD-Companion 2025 - Berlin, Germany Duration: 22 Jun 2025 → 27 Jun 2025

Publication series

Name	Proceedings of the ACM SIGMOD International Conference on Management of Data
ISSN (Print)	0730-8078

Conference

Conference	2025 ACM SIGMOD/PODS International Conference on Management of Data, SIGMOD-Companion 2025
Country/Territory	Germany
City	Berlin
Period	22/06/25 → 27/06/25

Keywords

large language models
tabular question answering

Access to Document

10.1145/3722212.3725140

Cite this

Zhu, J. P., Cai, P., Xu, K., Li, L., Sun, Y., Zhou, S., Su, H., Tang, L., & Liu, Q. (2025). UNITQA: A Unified Automated Tabular Question Answering System with Multi-Agent Large Language Models. In A. Deshpande, A. Aboulnaga, B. Salimi, B. Chandramouli, B. Howe, B. T. Loo, B. Glavic, C. Curino, D. Zhe Wang, D. Suciu, D. Abadi, D. Srivastava, E. Wu, F. Nawab, I. Ilyas, J. Naughton, J. Rogers, J. Patel, J. Arulraj, J. Yang, K. Echihabi, K. Ross, K. Daudjee, L. Lakshmanan, M. Garofalakis, M. Riedewald, M. Mokbel, M. Ouzzani, O. Kennedy, O. Kennedy, P. Papotti, P. Alvaro, P. Bailis, R. Miller, S. B. Roy, S. Melnik, S. Idreos, S. Roy, T. Rekatsinas, V. Leis, W. Zhou, W. Gatterbauer, ... Z. Ives (Eds.), SIGMOD-Companion 2025 - Companion of the 2025 International Conference on Management of Data (pp. 279-282). (Proceedings of the ACM SIGMOD International Conference on Management of Data). Association for Computing Machinery. https://doi.org/10.1145/3722212.3725140

Zhu, Jun Peng ; Cai, Peng ; Xu, Kai et al. / UNITQA : A Unified Automated Tabular Question Answering System with Multi-Agent Large Language Models. SIGMOD-Companion 2025 - Companion of the 2025 International Conference on Management of Data. editor / Amol Deshpande ; Ashraf Aboulnaga ; Babak Salimi ; Badrish Chandramouli ; Bill Howe ; Boon Thau Loo ; Boris Glavic ; Carlo Curino ; Daisy Zhe Wang ; Dan Suciu ; Daniel Abadi ; Divesh Srivastava ; Eugene Wu ; Faisal Nawab ; Ihab Ilyas ; Jeffrey Naughton ; Jennie Rogers ; Jignesh Patel ; Joy Arulraj ; Jun Yang ; Karima Echihabi ; Kenneth Ross ; Khuzaima Daudjee ; Laks Lakshmanan ; Minos Garofalakis ; Mirek Riedewald ; Mohamed Mokbel ; Mourad Ouzzani ; Oliver Kennedy ; Oliver Kennedy ; Paolo Papotti ; Peter Alvaro ; Peter Bailis ; Renee Miller ; Senjuti Basu Roy ; Sergey Melnik ; Stratos Idreos ; Sudeepa Roy ; Theodoros Rekatsinas ; Viktor Leis ; Wenchao Zhou ; Wolfgang Gatterbauer ; Zack Ives. Association for Computing Machinery, 2025. pp. 279-282 (Proceedings of the ACM SIGMOD International Conference on Management of Data).

@inproceedings{476b5dd9178d48b29b7f38d6f470e0b4,

title = "UNITQA: A Unified Automated Tabular Question Answering System with Multi-Agent Large Language Models",

abstract = "Automated tabular question answering (TQA) has attracted significant attention in data analysis and natural language processing communities due to its powerful capabilities. The emergence of large language models (LLMs) has initiated a paradigm shift in this field. However, existing state-of-the-art approaches cannot generally operate on multiple tables from multiple heterogeneous systems, and the answer accuracy is insufficient to meet the demands of the industrial field. This paper presents UNITQA, a unified automated tabular question-answering system through multi-agent LLMs. First, UNITQA offers a user-friendly GUI interface that enables users to use natural language questions to execute TQA tasks. Second, UNITQA consists of five agents who collaborate to complete user-specified tasks. To efficiently orchestrate different agents, UNITQA utilizes a dynamic agent scheduling algorithm based on a finite-state machine. Third, UNITQA integrates a series of data connectors that allow UNITQA to access various tables from multiple heterogeneous systems. We have implemented and deployed UNITQA in numerous production environments and have demonstrated its usability and efficiency in representative real-world scenarios.",

keywords = "large language models, tabular question answering",

author = "Zhu, \{Jun Peng\} and Peng Cai and Kai Xu and Li Li and Yishen Sun and Shuai Zhou and Haihuang Su and Liu Tang and Qi Liu",

note = "Publisher Copyright: {\textcopyright} 2025 ACM.; 2025 ACM SIGMOD/PODS International Conference on Management of Data, SIGMOD-Companion 2025 ; Conference date: 22-06-2025 Through 27-06-2025",

year = "2025",

month = jun,

day = "22",

doi = "10.1145/3722212.3725140",

language = "英语",

series = "Proceedings of the ACM SIGMOD International Conference on Management of Data",

publisher = "Association for Computing Machinery",

pages = "279--282",

editor = "Amol Deshpande and Ashraf Aboulnaga and Babak Salimi and Badrish Chandramouli and Bill Howe and Loo, \{Boon Thau\} and Boris Glavic and Carlo Curino and \{Zhe Wang\}, Daisy and Dan Suciu and Daniel Abadi and Divesh Srivastava and Eugene Wu and Faisal Nawab and Ihab Ilyas and Jeffrey Naughton and Jennie Rogers and Jignesh Patel and Joy Arulraj and Jun Yang and Karima Echihabi and Kenneth Ross and Khuzaima Daudjee and Laks Lakshmanan and Minos Garofalakis and Mirek Riedewald and Mohamed Mokbel and Mourad Ouzzani and Oliver Kennedy and Oliver Kennedy and Paolo Papotti and Peter Alvaro and Peter Bailis and Renee Miller and Roy, \{Senjuti Basu\} and Sergey Melnik and Stratos Idreos and Sudeepa Roy and Theodoros Rekatsinas and Viktor Leis and Wenchao Zhou and Wolfgang Gatterbauer and Zack Ives",

booktitle = "SIGMOD-Companion 2025 - Companion of the 2025 International Conference on Management of Data",

address = "美国",

}

Zhu, JP, Cai, P, Xu, K, Li, L, Sun, Y, Zhou, S, Su, H, Tang, L & Liu, Q 2025, UNITQA: A Unified Automated Tabular Question Answering System with Multi-Agent Large Language Models. in A Deshpande, A Aboulnaga, B Salimi, B Chandramouli, B Howe, BT Loo, B Glavic, C Curino, D Zhe Wang, D Suciu, D Abadi, D Srivastava, E Wu, F Nawab, I Ilyas, J Naughton, J Rogers, J Patel, J Arulraj, J Yang, K Echihabi, K Ross, K Daudjee, L Lakshmanan, M Garofalakis, M Riedewald, M Mokbel, M Ouzzani, O Kennedy, O Kennedy, P Papotti, P Alvaro, P Bailis, R Miller, SB Roy, S Melnik, S Idreos, S Roy, T Rekatsinas, V Leis, W Zhou, W Gatterbauer & Z Ives (eds), SIGMOD-Companion 2025 - Companion of the 2025 International Conference on Management of Data. Proceedings of the ACM SIGMOD International Conference on Management of Data, Association for Computing Machinery, pp. 279-282, 2025 ACM SIGMOD/PODS International Conference on Management of Data, SIGMOD-Companion 2025, Berlin, Germany, 22/06/25. https://doi.org/10.1145/3722212.3725140

UNITQA: A Unified Automated Tabular Question Answering System with Multi-Agent Large Language Models. / Zhu, Jun Peng; Cai, Peng; Xu, Kai et al.
SIGMOD-Companion 2025 - Companion of the 2025 International Conference on Management of Data. ed. / Amol Deshpande; Ashraf Aboulnaga; Babak Salimi; Badrish Chandramouli; Bill Howe; Boon Thau Loo; Boris Glavic; Carlo Curino; Daisy Zhe Wang; Dan Suciu; Daniel Abadi; Divesh Srivastava; Eugene Wu; Faisal Nawab; Ihab Ilyas; Jeffrey Naughton; Jennie Rogers; Jignesh Patel; Joy Arulraj; Jun Yang; Karima Echihabi; Kenneth Ross; Khuzaima Daudjee; Laks Lakshmanan; Minos Garofalakis; Mirek Riedewald; Mohamed Mokbel; Mourad Ouzzani; Oliver Kennedy; Oliver Kennedy; Paolo Papotti; Peter Alvaro; Peter Bailis; Renee Miller; Senjuti Basu Roy; Sergey Melnik; Stratos Idreos; Sudeepa Roy; Theodoros Rekatsinas; Viktor Leis; Wenchao Zhou; Wolfgang Gatterbauer; Zack Ives. Association for Computing Machinery, 2025. p. 279-282 (Proceedings of the ACM SIGMOD International Conference on Management of Data).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - UNITQA

T2 - 2025 ACM SIGMOD/PODS International Conference on Management of Data, SIGMOD-Companion 2025

AU - Zhu, Jun Peng

AU - Cai, Peng

AU - Xu, Kai

AU - Li, Li

AU - Sun, Yishen

AU - Zhou, Shuai

AU - Su, Haihuang

AU - Tang, Liu

AU - Liu, Qi

PY - 2025/6/22

Y1 - 2025/6/22

N2 - Automated tabular question answering (TQA) has attracted significant attention in data analysis and natural language processing communities due to its powerful capabilities. The emergence of large language models (LLMs) has initiated a paradigm shift in this field. However, existing state-of-the-art approaches cannot generally operate on multiple tables from multiple heterogeneous systems, and the answer accuracy is insufficient to meet the demands of the industrial field. This paper presents UNITQA, a unified automated tabular question-answering system through multi-agent LLMs. First, UNITQA offers a user-friendly GUI interface that enables users to use natural language questions to execute TQA tasks. Second, UNITQA consists of five agents who collaborate to complete user-specified tasks. To efficiently orchestrate different agents, UNITQA utilizes a dynamic agent scheduling algorithm based on a finite-state machine. Third, UNITQA integrates a series of data connectors that allow UNITQA to access various tables from multiple heterogeneous systems. We have implemented and deployed UNITQA in numerous production environments and have demonstrated its usability and efficiency in representative real-world scenarios.

AB - Automated tabular question answering (TQA) has attracted significant attention in data analysis and natural language processing communities due to its powerful capabilities. The emergence of large language models (LLMs) has initiated a paradigm shift in this field. However, existing state-of-the-art approaches cannot generally operate on multiple tables from multiple heterogeneous systems, and the answer accuracy is insufficient to meet the demands of the industrial field. This paper presents UNITQA, a unified automated tabular question-answering system through multi-agent LLMs. First, UNITQA offers a user-friendly GUI interface that enables users to use natural language questions to execute TQA tasks. Second, UNITQA consists of five agents who collaborate to complete user-specified tasks. To efficiently orchestrate different agents, UNITQA utilizes a dynamic agent scheduling algorithm based on a finite-state machine. Third, UNITQA integrates a series of data connectors that allow UNITQA to access various tables from multiple heterogeneous systems. We have implemented and deployed UNITQA in numerous production environments and have demonstrated its usability and efficiency in representative real-world scenarios.

KW - large language models

KW - tabular question answering

UR - https://www.scopus.com/pages/publications/105010216581

U2 - 10.1145/3722212.3725140

DO - 10.1145/3722212.3725140

M3 - 会议稿件

AN - SCOPUS:105010216581

T3 - Proceedings of the ACM SIGMOD International Conference on Management of Data

SP - 279

EP - 282

BT - SIGMOD-Companion 2025 - Companion of the 2025 International Conference on Management of Data

A2 - Deshpande, Amol

A2 - Aboulnaga, Ashraf

A2 - Salimi, Babak

A2 - Chandramouli, Badrish

A2 - Howe, Bill

A2 - Loo, Boon Thau

A2 - Glavic, Boris

A2 - Curino, Carlo

A2 - Zhe Wang, Daisy

A2 - Suciu, Dan

A2 - Abadi, Daniel

A2 - Srivastava, Divesh

A2 - Wu, Eugene

A2 - Nawab, Faisal

A2 - Ilyas, Ihab

A2 - Naughton, Jeffrey

A2 - Rogers, Jennie

A2 - Patel, Jignesh

A2 - Arulraj, Joy

A2 - Yang, Jun

A2 - Echihabi, Karima

A2 - Ross, Kenneth

A2 - Daudjee, Khuzaima

A2 - Lakshmanan, Laks

A2 - Garofalakis, Minos

A2 - Riedewald, Mirek

A2 - Mokbel, Mohamed

A2 - Ouzzani, Mourad

A2 - Kennedy, Oliver

A2 - Papotti, Paolo

A2 - Alvaro, Peter

A2 - Bailis, Peter

A2 - Miller, Renee

A2 - Roy, Senjuti Basu

A2 - Melnik, Sergey

A2 - Idreos, Stratos

A2 - Roy, Sudeepa

A2 - Rekatsinas, Theodoros

A2 - Leis, Viktor

A2 - Zhou, Wenchao

A2 - Gatterbauer, Wolfgang

A2 - Ives, Zack

PB - Association for Computing Machinery

Y2 - 22 June 2025 through 27 June 2025

ER -

Zhu JP, Cai P, Xu K, Li L, Sun Y, Zhou S et al. UNITQA: A Unified Automated Tabular Question Answering System with Multi-Agent Large Language Models. In Deshpande A, Aboulnaga A, Salimi B, Chandramouli B, Howe B, Loo BT, Glavic B, Curino C, Zhe Wang D, Suciu D, Abadi D, Srivastava D, Wu E, Nawab F, Ilyas I, Naughton J, Rogers J, Patel J, Arulraj J, Yang J, Echihabi K, Ross K, Daudjee K, Lakshmanan L, Garofalakis M, Riedewald M, Mokbel M, Ouzzani M, Kennedy O, Kennedy O, Papotti P, Alvaro P, Bailis P, Miller R, Roy SB, Melnik S, Idreos S, Roy S, Rekatsinas T, Leis V, Zhou W, Gatterbauer W, Ives Z, editors, SIGMOD-Companion 2025 - Companion of the 2025 International Conference on Management of Data. Association for Computing Machinery. 2025. p. 279-282. (Proceedings of the ACM SIGMOD International Conference on Management of Data). doi: 10.1145/3722212.3725140

UNITQA: A Unified Automated Tabular Question Answering System with Multi-Agent Large Language Models

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this