Text-to-SQL Systems: A Comprehensive Study of Evolution, Architectural Advancements, Challenges, and the Role of Large Language Models in Intelligent Query Generation

Year : 2026 | Volume : 13 | Issue : 02 | Page : 10 19
By

Yogesh Adhana,

Shefali Madan,

  1. Student, Department of Computer Science and Engineering J.C. Bose University of Science and Technology, Haryana, India
  2. Associate Professor, Department of Computer Science and Engineering J.C. Bose University of Science and Technology, Haryana, India

Abstract

Interacting with relational databases traditionally requires knowledge of Structured Query Language (SQL), which creates a significant barrier for users who do not have technical expertise. To address this challenge, Text-to-SQL systems have emerged as an important area of research, enabling users to express their information needs in natural language while automatically generating executable SQL queries. These systems aim to make database access more intuitive, efficient, and accessible across a wide range of domains. This paper provides a structured review of the development of Text-to-SQL systems, tracing their evolution from early rule-based and template-driven methods to more advanced machine learning, deep learning, and Large Language Model (LLM)-based approaches. It analyzes the major techniques used in these systems and highlights how recent advances in Natural Language Processing (NLP) have improved contextual understanding, semantic parsing, and the accuracy of SQL query generation. The paper also examines the benefits of modern LLM-driven architectures, particularly their ability to generalize across diverse schemas and handle more complex user requests. At the same time, several persistent challenges remain, including schema linking, ambiguity in natural language queries, domain adaptation, interpretability, and the high computational cost of large models. Emerging directions such as Retrieval-Augmented Generation (RAG), prompt-based learning, and hybrid architectures are also discussed as promising solutions. Overall, the study concludes that LLM-based Text-to-SQL systems represent a major advancement, but further research is necessary to improve their efficiency, reliability, and adaptability in real-world applications.

Keywords: Text-to-SQL, natural language processing, large language models, query generation, database systems, RAG

[This article belongs to Journal of Advanced Database Management & Systems ]

How to cite this article: Yogesh Adhana, Shefali Madan. Text-to-SQL Systems: A Comprehensive Study of Evolution, Architectural Advancements, Challenges, and the Role of Large Language Models in Intelligent Query Generation. Journal of Advanced Database Management & Systems. 2026; 13(02):10-19.
How to cite this URL: Yogesh Adhana, Shefali Madan. Text-to-SQL Systems: A Comprehensive Study of Evolution, Architectural Advancements, Challenges, and the Role of Large Language Models in Intelligent Query Generation. Journal of Advanced Database Management & Systems. 2026; 13(02):10-19. Available from: https://journals.stmjournals.com/joadms/article=2026/view=259031

References

  1. Zhong V, Xiong C, Socher R. Seq2SQL: generating structured queries from natural language using reinforcement learning [Preprint]. arXiv:1709.00103; 2017.
  2. Xu X, Liu C, Song D. SQLNet: generating structured queries from natural language without reinforcement learning [Preprint]. arXiv:1711.04436; 2017.
  3. Yu T, Zhang R, Yang K, Yasunaga M, Wang D, Li Z, et al. 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 Empirical Methods in Natural Language Processing; 2018 Oct-Nov; Brussels, Belgium. Stroudsburg (PA): Association for Computational Linguistics; 2018. p. 3911-3921. doi:10.18653/v1/D18-1425.
  4. Guo J, Zhan Z, Gao Y, Xiao Y, Lou JG, Liu T, et al. Towards complex text-to-SQL in cross-domain database with intermediate representation. In: Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics; 2019 Jul; Florence, Italy. Stroudsburg (PA): Association for Computational Linguistics; 2019. p. 4524-4535. doi:10.18653/v1/P19-1444.
  5. Wang B, Shin R, Liu X, Polozov O, Richardson M. RAT-SQL: relation-aware schema encoding and linking for text-to-SQL parsers. In: Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics; 2020 Jul; Online. Stroudsburg (PA): Association for Computational Linguistics; 2020. p. 7567-7578. doi:10.18653/v1/2020.acl-main.677.
  6. Scholak T, Schucher N, Bahdanau D. PICARD: parsing incrementally for constrained auto-regressive decoding from language models. In: Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing; 2021 Nov; Online and Punta Cana, Dominican Republic. Stroudsburg (PA): Association for Computational Linguistics; 2021. p. 9895-9901. doi:10.18653/v1/2021.emnlp-main.779.
  7. Raffel C, Shazeer N, Roberts A, Lee K, Narang S, Matena M, et al. Exploring the limits of transfer learning with a unified text-to-text transformer. J Mach Learn Res. 2020;21(140):1-67.
  8. Brown T, Mann B, Ryder N, Subbiah M, Kaplan JD, Dhariwal P, et al. Language models are few-shot learners. In: Advances in Neural Information Processing Systems. 2020;33:1877-1901.
  9. Li H, Zhang J, Li C, Chen H. RESDSQL: decoupling schema linking and skeleton parsing for text-to-SQL. Proc AAAI Conf Artif Intell. 2023;37(11):13067-13075. doi:10.1609/aaai.v37i11.26535.
  10. Guo C, Tian Z, Tang J, Li S, Wang T. Multi-pattern retrieval-augmented framework for Text-to-SQL with Poincaré-Skeleton retrieval and meta-instruction reasoning. Inf Process Manag. 2025;62(3):103978. doi:10.1016/j.ipm.2024.103978.
Regular Issue Subscription Review Article
Volume 13
Issue 02
Received 25/04/2026
Accepted 25/05/2026
Published 30/06/2026
Publication Time 66 Days

Login

My IP

PlumX Metrics

Support