Deep Graph Convolutional Encoders for Structured Data to Text Generation

TL;DR

This paper introduces a graph convolutional encoder for structured data to text generation, directly leveraging graph structures instead of linearized sequences, leading to improved performance on graph-to-sequence tasks.

Contribution

It presents a novel graph convolutional encoder that explicitly encodes input graph structures for text generation, outperforming traditional sequence-based methods.

Findings

Graph convolutional encoder improves generation quality.

Explicit graph structure encoding benefits performance.

Empirical results on two datasets confirm advantages.

Abstract

Most previous work on neural text generation from graph-structured data relies on standard sequence-to-sequence methods. These approaches linearise the input graph to be fed to a recurrent neural network. In this paper, we propose an alternative encoder based on graph convolutional networks that directly exploits the input structure. We report results on two graph-to-sequence datasets that empirically show the benefits of explicitly encoding the input graph structure.