Grounded Adaptation for Zero-shot Executable Semantic Parsing

2020/09/16 by Victor Zhong, Victor W. Zhong, Zhong, Victor +6 · 7 citations
Computer Science · #Artificial Intelligence (cs.AI) #Artificial intelligence #Computation and Language (cs.CL) #Computer science #Consistency (knowledge bases) #Databases (cs.DB) #Executable #FOS: Computer and information sciences #Logical form #Machine Learning (cs.LG) #Multimodal Machine Learning Applications #Natural Language Processing Techniques #Natural language processing #Parser combinator #Parsing #Programming language #Top-down parsing #Topic Modeling #cs.AI #cs.CL #cs.DB #cs.LG

paper · pdf · doi:10.48550/arxiv.2009.07396

published in arXiv (Cornell University) (Cornell University) · EMNLP 2020 long paper. 14 pages, 5 figures

openalex publication_date 2020/09/16 · arxiv created 2021/02/01 · arxiv updated 2021/02/03 · openalex created_date 2025/10/10 · openalex updated_date 2026/08/09

Abstract

We propose Grounded Adaptation for Zero-shot Executable Semantic Parsing (GAZP) to adapt an existing semantic parser to new environments (e.g. new database schemas). GAZP combines a forward semantic parser with a backward utterance generator to synthesize data (e.g. utterances and SQL queries) in the new environment, then selects cycle-consistent examples to adapt the parser. Unlike data-augmentation, which typically synthesizes unverified examples in the training environment, GAZP synthesizes examples in the new environment whose input-output consistency are verified. On the Spider, Sparc, and CoSQL zero-shot semantic parsing tasks, GAZP improves logical form and execution accuracy of the baseline parser. Our analyses show that GAZP outperforms data-augmentation in the training environment, performance increases with the amount of GAZP-synthesized data, and cycle-consistency is central to successful adaptation.

Citations

• CoSQL: A Conversational Text-to-SQL Challenge Towards Cross-Domain Natural Language Interfaces to Databases
• LLM-Supported Manufacturing Mapping Generation

Cited by

• Skeletons Matter: Dynamic Data Augmentation for Text-to-Query
• Text2SQL-Flow: A Robust SQL-Aware Data Augmentation Framework for Text-to-SQL
• MTSQL-R1: Towards Long-Horizon Multi-Turn Text-to-SQL via Agentic Training
• NL2GenSym: Natural Language to Generative Symbolic Rules for SOAR Cognitive Architecture via Large Language Models
• GradeSQL: Test-Time Inference with Outcome Reward Models for Text-to-SQL Generation from Large Language Models
• Text-to-SQL for Enterprise Data Analytics
• RingSQL: Schema-Independent Synthetic Data Generation for Text-to-SQL Reinforcement Learning

Related

• Decoupling Structure and Lexicon for Zero-Shot Semantic Parsing
• On The Ingredients of an Effective Zero-shot Semantic Parser
• Hierarchical Neural Data Synthesis for Semantic Parsing
• ZEROTOP: Zero-Shot Task-Oriented Semantic Parsing using Large Language Models
• Training Naturalized Semantic Parsers with Very Little Data