Semantic Graph Parsing with Recurrent Neural Network DAG Grammars

TL;DR

This paper introduces a graph-aware sequence model called recurrent neural network DAG grammars that predicts well-formed semantic graphs, improving multilingual semantic parsing performance across several languages.

Contribution

The paper proposes a novel recurrent neural network DAG grammar model that guarantees well-formed graph predictions, advancing semantic parsing techniques.

Findings

Achieved competitive results in English semantic parsing.

First results for German, Italian, and Dutch semantic parsing.

Model ensures only well-formed graphs during prediction.

Abstract

Semantic parses are directed acyclic graphs (DAGs), so semantic parsing should be modeled as graph prediction. But predicting graphs presents difficult technical challenges, so it is simpler and more common to predict the linearized graphs found in semantic parsing datasets using well-understood sequence models. The cost of this simplicity is that the predicted strings may not be well-formed graphs. We present recurrent neural network DAG grammars, a graph-aware sequence model that ensures only well-formed graphs while sidestepping many difficulties in graph prediction. We test our model on the Parallel Meaning Bank---a multilingual semantic graphbank. Our approach yields competitive results in English and establishes the first results for German, Italian and Dutch.