Coarse-to-Fine Decoding for Neural Semantic Parsing

TL;DR

The paper introduces a coarse-to-fine decoding framework for neural semantic parsing that first generates a high-level meaning sketch and then completes the details conditioned on the input and the sketch. By disentangling structure from specifics and sharing coarse templates, the model achieves competitive results across logical form, code, and SQL tasks with relatively simple decoders. Empirical results demonstrate consistent improvements over single-stage baselines, with ablations confirming the value of the sketch and table-aware components. The approach offers a flexible, broadly applicable strategy for structured meaning representations and points to future work in weak supervision and alternative sketch definitions.

Abstract

Semantic parsing aims at mapping natural language utterances into structured meaning representations. In this work, we propose a structure-aware neural architecture which decomposes the semantic parsing process into two stages. Given an input utterance, we first generate a rough sketch of its meaning, where low-level information (such as variable names and arguments) is glossed over. Then, we fill in missing details by taking into account the natural language input and the sketch itself. Experimental results on four datasets characteristic of different domains and meaning representations show that our approach consistently improves performance, achieving competitive results despite the use of relatively simple decoders.

Figures (3)

• Figure 1: We first generate the meaning sketch $a$ for natural language input $x$. Then, a fine meaning decoder fills in the missing details (shown in red) of meaning representation $y$. The coarse structure $a$ is used to guide and constrain the output decoding.
• Figure 2: Table-aware input encoder (left) and table column encoder (right) used for WikiSQL.
• Figure 3: Fine meaning decoder of the WHERE clause used for WikiSQL.