Sentence Rewriting for Semantic Parsing

TL;DR

This paper introduces a sentence rewriting approach to semantic parsing that addresses vocabulary mismatch issues by transforming natural language sentences into forms aligned with target logical forms, improving parsing accuracy.

Contribution

It proposes two novel sentence rewriting methods for different mismatch types, enhancing semantic parser performance on benchmark datasets.

Findings

Outperforms baseline with 3.4% F1 improvement

Generates more accurate logical forms

Provides robust sentence parsing

Abstract

A major challenge of semantic parsing is the vocabulary mismatch problem between natural language and target ontology. In this paper, we propose a sentence rewriting based semantic parsing method, which can effectively resolve the mismatch problem by rewriting a sentence into a new form which has the same structure with its target logical form. Specifically, we propose two sentence-rewriting methods for two common types of mismatch: a dictionary-based method for 1-N mismatch and a template-based method for N-1 mismatch. We evaluate our entence rewriting based semantic parser on the benchmark semantic parsing dataset -- WEBQUESTIONS. Experimental results show that our system outperforms the base system with a 3.4% gain in F1, and generates logical forms more accurately and parses sentences more robustly.