Text Generation from Knowledge Graphs with Graph Transformers

TL;DR

This paper presents a novel graph transformer encoder that effectively generates coherent scientific texts from knowledge graphs, overcoming structural challenges and outperforming existing methods in informativeness and document structure.

Contribution

The work introduces a new graph transforming encoder for graph-to-text generation that leverages relational structures without linearization, enhancing coherence and informativeness.

Findings

Produces more informative texts than competitors

Generates texts with better document structure

Outperforms existing encoder-decoder methods

Abstract

Generating texts which express complex ideas spanning multiple sentences requires a structured representation of their content (document plan), but these representations are prohibitively expensive to manually produce. In this work, we address the problem of generating coherent multi-sentence texts from the output of an information extraction system, and in particular a knowledge graph. Graphical knowledge representations are ubiquitous in computing, but pose a significant challenge for text generation techniques due to their non-hierarchical nature, collapsing of long-distance dependencies, and structural variety. We introduce a novel graph transforming encoder which can leverage the relational structure of such knowledge graphs without imposing linearization or hierarchical constraints. Incorporated into an encoder-decoder setup, we provide an end-to-end trainable system for graph-to-text generation that we apply to the domain of scientific text. Automatic and human evaluations show that our technique produces more informative texts which exhibit better document structure than competitive encoder-decoder methods.