Harnessing LLMs for Practical NL2SQL: Paradigms and Challenges

ABSTRACT

With the advent of large language models (LLMs), numerous NL2SQL approaches have been proposed, demonstrating exceptional performance across various benchmarks. However, effectively harnessing LLMs for practical NL2SQL applications remains a challenging question due to the different requirements among applications, such as training data, computational resources, etc. In this talk, I will explore the potential paradigms for developing NL2SQL models tailored for real-world scenarios and examine representative approaches within each paradigm. Additionally, I will discuss the research challenges and future directions in this field.

SPEAKER BIO

Ju FAN is a professor at Renmin University of China. He received his Ph.D. from Tsinghua University, and worked as a research fellow at National University of Singapore. His research interests are in general area of data management, and his current research focuses on building next-generation data preparation systems. He has published more than 60 papers at top conferences/journals, including SIGMOD, VLDB, ICDE and VLDB Journal. He is a publication chair for VLDB 2023/2024 and regularly serves as PC member for SIGMOD, VLDB and ICDE. He is also a recipient of ACM SIGMOD Research Highlight Award and ACM China Rising Star award.