Track-SQL: Enhancing Generative Language Models with Dual-Extractive Modules for Schema and Context Tracking in Multi-turn Text-to-SQL

2026-03-06

North American Chapter of the Association for Computational Linguistics

AI Summary

This work addresses the challenge that generative language models struggle to effectively model dynamically evolving database schemas and contextual dependencies in multi-turn Text-to-SQL tasks. To this end, the authors propose a dual extraction-augmented architecture comprising a semantics-enhanced schema extractor and a schema-aware context extractor, which jointly track schema evolution and contextual information across dialogue turns with high precision. These components are integrated with a generative language model and co-optimized for SQL generation. The proposed approach achieves state-of-the-art performance, yielding absolute improvements of 7.1% and 9.55% in execution accuracy on the SparC and CoSQL benchmarks, respectively.

Technology Category

• Natural Language Processing: Language Grounding & Multi-modal NLP
• Planning, Routing, and Scheduling: Planning with Language Models
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Search and Retrieval-Augmented AI: Large language models for search
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Web Mining and Content Analysis: Models for Web evolution

Abstract

Generative language models have shown significant potential in single-turn Text-to-SQL. However, their performance does not extend equivalently to multi-turn Text-to-SQL. This is primarily due to generative language models'inadequacy in handling the complexities of context information and dynamic schema linking in multi-turn interactions. In this paper, we propose a framework named Track-SQL, which enhances generative language models with dual-extractive modules designed to track schema and contextual changes in multi-turn Text-to-SQL. Specifically, Track-SQL incorporates a \emph{Semantic-enhanced Schema Extractor} and a \emph{Schema-aware Context Extractor}. Experimental results demonstrate that Track-SQL achieves state-of-the-art performance on the SparC and CoSQL datasets. Furthermore, detailed ablation studies reveal that Track-SQL significantly improves execution accuracy in multi-turn interactions by 7.1\% and 9.55\% on these datasets, respectively. Our implementation will be open-sourced at https://github.com/DMIRLAB-Group/Track-SQL.

Problem

• multi-turn Text-to-SQL
• context tracking
• schema linking
• generative language models

Innovation

• Text-to-SQL
• multi-turn dialogue
• schema linking
• context tracking
• generative language models