CLMay 14, 2023

Learning to Simulate Natural Language Feedback for Interactive Semantic Parsing

arXiv:2305.08195v2229 citations
Originality Incremental advance
AI Analysis

This work addresses the scalability issue in interactive semantic parsing for applications like text-to-SQL by reducing reliance on expensive human feedback, though it is incremental as it builds on prior interactive parsing methods.

The paper tackles the high cost of human-annotated natural language feedback for training interactive semantic parsers by proposing a task to simulate such feedback, showing that their simulator generates high-quality feedback that boosts parser error correction, achieving comparable performance to full human annotations in low-data settings.

Interactive semantic parsing based on natural language (NL) feedback, where users provide feedback to correct the parser mistakes, has emerged as a more practical scenario than the traditional one-shot semantic parsing. However, prior work has heavily relied on human-annotated feedback data to train the interactive semantic parser, which is prohibitively expensive and not scalable. In this work, we propose a new task of simulating NL feedback for interactive semantic parsing. We accompany the task with a novel feedback evaluator. The evaluator is specifically designed to assess the quality of the simulated feedback, based on which we decide the best feedback simulator from our proposed variants. On a text-to-SQL dataset, we show that our feedback simulator can generate high-quality NL feedback to boost the error correction ability of a specific parser. In low-data settings, our feedback simulator can help achieve comparable error correction performance as trained using the costly, full set of human annotations.

Code Implementations2 repos
Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes