Text Generation from Knowledge Graphs with Graph Transformers (2019-04-01T00:00:00.000000Z)

TL;DR

This work addresses the problem of generating coherent multi-sentence texts from the output of an information extraction system, and in particular a knowledge graph by introducing a novel graph transforming encoder which can leverage the relational structure of such knowledge graphs without imposing linearization or hierarchical constraints.

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.

Authors

Mirella Lapata

27 papers

Hannaneh Hajishirzi

32 papers

Yi Luan

5 papers

TL;DR

Abstract

Authors

References35 items

Deep Graph Convolutional Encoders for Structured Data to Text Generation

Data-to-Text Generation with Content Selection and Planning

Multi-Task Identification of Entities, Relations, and Coreference for Scientific Knowledge Graph Construction

Graph-to-Sequence Learning using Gated Graph Neural Networks

Paper Abstract Writing through Editing Mechanism

A Graph-to-Sequence Model for AMR-to-Text Generation

Construction of the Literature Graph in Semantic Scholar

Graph Attention Networks

Challenges in Data-to-Document Generation

Attention is All you Need

Neural AMR: Sequence-to-Sequence Models for Parsing and Generation

Get To The Point: Summarization with Pointer-Generator Networks

Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering

Capturing Reliable Fine-Grained Sentiment Associations by Crowdsourcing and Best–Worst Scaling

Neural Text Generation from Structured Data with Application to the Biography Domain

Modeling Coverage for Neural Machine Translation

Best-Worst Scaling: Theory, Methods and Applications

What to talk about and how? Selective Generation using LSTMs with Coarse-to-Fine Alignment

Effective Approaches to Attention-based Neural Machine Translation

Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification

Sequence to Sequence Learning with Neural Networks

Meteor Universal: Language Specific Translation Evaluation for Any Target Language

Inducing Document Plans for Concept-to-Text Generation

Sequence Transduction with Recurrent Neural Networks

Generative Alignment and Semantic Parsing for Learning from Ambiguous Supervision

Learning Semantic Correspondences with Less Supervision

Collective Content Selection for Concept-to-Text Generation

Bleu: a Method for Automatic Evaluation of Machine Translation

Long Short-Term Memory

Improving Language Understanding by Generative Pre-Training

Kipf and Max Welling . 2017 . Semi - Supervised Classification with Graph Convolutional Networks

Dropout: a simple way to prevent neural networks from overfitting

On the momentum term in gradient descent learning algorithms

Early Stopping-But When?

Best-Worst Scaling: A Model for the Largest Difference Judgments

Field of Study

Journal Information

Name

Volume

Venue Information

Name

Type

URL

Alternate Names