SQLizer: query synthesis from natural language
Abstract
This paper presents a new technique for automatically synthesizing SQL queries from natural language (NL). At the core of our technique is a new NL-based program synthesis methodology that combines semantic parsing techniques from the NLP community with type-directed program synthesis and automated program repair. Starting with a program sketch obtained using standard parsing techniques, our approach involves an iterative refinement loop that alternates between probabilistic type inhabitation and automated sketch repair. We use the proposed idea to build an end-to-end system called SQLIZER that can synthesize SQL queries from natural language. Our method is fully automated, works for any database without requiring additional customization, and does not require users to know the underlying database schema. We evaluate our approach on over 450 natural language queries concerning three different databases, namely MAS, IMDB, and YELP. Our experiments show that the desired query is ranked within the top 5 candidates in close to 90% of the cases and that SQLIZER outperforms NALIR, a state-of-the-art tool that won a best paper award at VLDB'14.
Document Details
- Document Type
- Pub Defense Publication
- Publication Date
- Oct 12, 2017
- Source ID
- 10.1145/3133887
Entities
People
- Işıl Dillig
- Navid Yaghmazadeh
- Thomas Dillig
- Yuepeng Wang
Organizations
- Defense Advanced Research Projects Agency
- University of Texas at Austin