Structural Neural Encoders for AMR-to-text Generation (2019-03-27T00:00:00.000000Z)

TL;DR

The extent to which reentrancies (nodes with multiple parents) have an impact on AMR-to-text generation is investigated by comparing graph encoder to tree encoders, where reENTrancies are not preserved.

Abstract

AMR-to-text generation is a problem recently introduced to the NLP community, in which the goal is to generate sentences from Abstract Meaning Representation (AMR) graphs. Sequence-to-sequence models can be used to this end by converting the AMR graphs to strings. Approaching the problem while working directly with graphs requires the use of graph-to-sequence models that encode the AMR graph into a vector representation. Such encoding has been shown to be beneficial in the past, and unlike sequential encoding, it allows us to explicitly capture reentrant structures in the AMR graphs. We investigate the extent to which reentrancies (nodes with multiple parents) have an impact on AMR-to-text generation by comparing graph encoders to tree encoders, where reentrancies are not preserved. We show that improvements in the treatment of reentrancies and long-range dependencies contribute to higher overall scores for graph encoders. Our best model achieves 24.40 BLEU on LDC2015E86, outperforming the state of the art by 1.1 points and 24.54 BLEU on LDC2017T10, outperforming the state of the art by 1.24 points.

Authors

Shay B. Cohen

5 papers

Marco Damonte

2 papers

TL;DR

Abstract

Authors

References34 items

Deep Graph Convolutional Encoders for Structured Data to Text Generation

Graph-to-Sequence Learning using Gated Graph Neural Networks

Toward Abstractive Summarization Using Semantic Representations

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

Graph2Seq: Graph to Sequence Learning with Attention-based Neural Networks

Automatic differentiation in PyTorch

Dealing with Co-reference in Neural Semantic Parsing

Graph Convolutional Networks for Named Entity Recognition

Improved Neural Machine Translation with a Syntax-Aware Encoder and Decoder

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

Graph Convolutional Encoders for Syntax-aware Neural Machine Translation

Encoding Sentences with Graph Convolutional Networks for Semantic Role Labeling

OpenNMT: Open-Source Toolkit for Neural Machine Translation

How Grammatical is Character-level Neural Machine Translation? Assessing MT Quality with Contrastive Translation Pairs

Assessing the Ability of LSTMs to Learn Syntax-Sensitive Dependencies

Neural Headline Generation on Abstract Meaning Representation

An Incremental Parser for Abstract Meaning Representation

Pointing the Unknown Words

Tree-to-Sequence Attentional Neural Machine Translation

Gated Graph Sequence Neural Networks

Convolutional Networks on Graphs for Learning Molecular Fingerprints

Highway Networks

Improved Semantic Representations From Tree-Structured Long Short-Term Memory Networks

Neural Machine Translation by Jointly Learning to Align and Translate

The Stanford CoreNLP Natural Language Processing Toolkit

Abstract Meaning Representation for Sembanking

Speech recognition with deep recurrent neural networks

METEOR: An Automatic Metric for MT Evaluation with Improved Correlation with Human Judgments

Bleu: a Method for Automatic Evaluation of Machine Translation

Learning long-term dependencies with gradient descent is difficult

Semi-supervised classiﬁcation with graph convolutional networks

Dropout: a simple way to prevent neural networks from overfitting

Proceedings of the ACL Workshop on Intrinsic and Extrinsic Evaluation Measures for Machine Translation and/or Summarization

Finite geometrical systems

Field of Study

Venue Information

Name

Type

URL

Alternate Names