Generating Logical Forms from Graph Representations of Text and Entities

TL;DR

This paper introduces a GNN-based method for generating logical forms from text and entity graphs, effectively incorporating entity relations and achieving competitive results without pre-training, and superior performance with BERT.

Contribution

The paper presents a novel GNN architecture combined with a decoder copy mechanism for logical form generation from graph representations, improving over prior methods.

Findings

Competitive with state-of-the-art without pre-training

Outperforms existing approaches with BERT pre-training

Effective incorporation of entity relations in logical form generation

Abstract

Structured information about entities is critical for many semantic parsing tasks. We present an approach that uses a Graph Neural Network (GNN) architecture to incorporate information about relevant entities and their relations during parsing. Combined with a decoder copy mechanism, this approach provides a conceptually simple mechanism to generate logical forms with entities. We demonstrate that this approach is competitive with the state-of-the-art across several tasks without pre-training, and outperforms existing approaches when combined with BERT pre-training.