Falcon generates complex synthetic data for natural language SQL queries
FALCON: A Model and Dataset Agnostic Framework for Synthetic Data Generation for NL2SQL Pairs
Computation and LanguageDatabasesMachine Learning
Summary
Accessing data stored in databases using everyday language can be tricky because people phrase questions in very different ways. The authors created a method called FALCON that makes lots of realistic, challenging examples of natural language questions paired with the correct database queries. Their method includes making complex SQL queries and adding ambiguous language to better prepare AI models. They show that training with these examples helps AI perform better on tough questions, while still doing well on simpler ones. FALCON works with any database or AI model without needing external services.
What this means in practice
- •For database engineering teams: Generate diverse, realistic training data for natural language interfaces to databases without external APIs or large models.
- •For chatbot developers: Create NL-to-SQL training data that improves understanding of ambiguous and complex user queries in conversational AI.
Authors
Darian Lee, Shannon Rumsey, Jack St. Clair, Xinyi Tang, Aditya Bansal, Yuanming Shi
Abstract
Relational databases are among the most widely deployed forms of structured knowledge, and natural language access to them requires grounding language onto schema entities and relations while handling the ambiguity inherent in how people phrase requests. Existing synthetic NL-to-SQL data generation methods largely ignore this ambiguity and produce oversimplified queries that fail to prepare models for the complexity of real-world structured knowledge access. We present FALCON, a framework that generates realistic, ambiguity-aware NL-to-SQL data matching the complexity of challenging real-world benchmarks, at low cost using compact open models. Our approach combines reserved-word SQL seeding and persona-based prompting to generate structurally complex queries, while alignment-based filtering preserves difficulty by distinguishing genuinely incorrect examples from complex but valid queries. Human evaluation confirms consistent high quality across model sizes, and our generated data exceeds existing benchmarks in both SQL complexity and natural language richness. Difficulty-stratified analysis shows models trained on FALCON data increasingly outperform baseline-trained models as query complexity increases, validating our pipeline's success in generating challenging training data. When combined with a small proportion of existing benchmark data, mixed training recovers performance on simpler queries while preserving these advantages on complex ones. The model- and database-agnostic design enables organizations to generate high-complexity NL-to-SQL training data locally without external APIs.