Graph-to-Sequence Learning using Gated Graph Neural Networks

By Daniel Beck, Gholamreza Haffari, Trevor Cohn. Annual Meeting of the Association for Computational Linguistics, 2018

345 citations recorded by the source corpus.

Many NLP applications can be framed as a graph-to-sequence learning problem. Previous work proposing neural architectures on graph-to-sequence obtained promising results compared to grammar-based approaches but still rely on linearisation heuristics and/or standard recurrent networks to achieve the best performance. In this work propose a new model that encodes the full structural information contained in the graph. Our architecture couples the recently proposed Gated Graph Neural Networks with an input transformation that allows nodes and edges to have their own hidden representations, while tackling the parameter explosion problem present in previous work. Experimental results shows that our model outperforms strong baselines in generation from AMR graphs and syntax-based neural machine translation.

Subfields: Artificial Intelligence, Molecular Biology

View Paper: https://www.semanticscholar.org/paper/f11198dc277af1c5547ab3ae8d562fc8c16a7c70

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Molecular Biology

Citations in the corpus

Text Generation from Knowledge Graphs with Graph Transformers (2019, methodology): “Graph Preparation We convert each graph to an unlabeled connected bipartite graphs following a similar procedure to Beck et al. (2018).”
Graph Learning: A Survey (2021, background): “Gated Graph Neural Networks (GGNNs) can also be used to address neural machine translation and text generation [163].”
Encoding word order in complex embeddings (2019, background): “(2)Graph convolutional networks (GCNs) (Beck et al., 2018; Sahu et al., 2019) also encode positional information.”
Graph Neural Networks in Recommender Systems: A Survey (2020, background): “Recently, systems based on variants of GNN have demonstrated ground-breaking performance on many tasks related to graph data, such as physical systems [2, 45], protein structure [11], knowledge graph [15] and other areas [3, 4, 26].”
Graphspeech: Syntax-Aware Graph Attention Network for Neural Speech Synthesis (2020, background): “In graph-to-sequence learning [23, 24], we can frame the self-attention network in the framework of graph neural network, where the token sequence is considered as an unlabeled fully-connected graph (each token as a node), and the self-attention mechanism is considered as a speciﬁc message-passing…”
A Gentle Introduction to Deep Learning for Graphs (2019, background): “An example is neural machine translation, which can be formulated as a graph-to-sequence problem [8] to consider syntactic dependencies in the source and target sentence.”
Using Local Knowledge Graph Construction to Scale Seq2Seq Models to Multi-Document Inputs (2019, background): “Recently, many have explored neural architectures that can encode graph structured input (Bruna et al., 2013; Kipf and Welling, 2016; Beck et al., 2018; Zhou et al., 2018; Xu et al., 2018; Lai et al., 2019).”
Learning to Fix Build Errors with Graph2Diff Neural Networks (2019, methodology): “Beck et al. [7] develop a graph-to-sequence model with GNNs and an attention-based de-coder.”
A Comprehensive Survey on Graph Neural Networks (2019, background): “Beck et al. [156] apply a GGNN [17] to graph-to-sequence learning and neural machine translation.”
Composition-based Multi-Relational Graph Convolutional Networks (2019, background): “Initial attempts at learning representations for relations in graphs (Monti et al., 2018; Beck et al., 2018) have shown some performance gains on tasks like node classification and neural machine translation.”
Knowledge Graph-Augmented Abstractive Summarization with Semantic-Driven Cloze Reward (2020, methodology): “Both Gated Graph Neural Networks (GGNNs) (Beck et al., 2018) and Graph Convolutional Networks (GCNs) (Damonte and Cohen, 2019) are shown to be effective in generating sentences from AMR graphs.”
A Tree-to-Sequence Model for Neural NLG in Task-Oriented Dialog (2019, methodology): “Previous work on AMR and WebNLG (Beck et al., 2018; Song et al., 2018; Marcheggiani and Perez-Beltrachini, 2018) has demonstrated improvements over Seq2Seq models by using graph-to-sequence models; while similar in principle, these works do not explore the use of hierarchical content plans as…”
An Attentive Survey of Attention Models (2019, background): “2018], and (iv) Graph to Sequence Generation [Beck et al. 2018; Zheng et al. 2020].”
GPT-too: A Language-Model-First Approach for AMR-to-Text Generation (2020, methodology): “Beck et al. (2018) improve upon prior RNN graph encoding (Song et al., 2018) with Levi Graph Transformations.”
GraphTTS: Graph-to-Sequence Modelling in Neural Text-to-Speech (2020, methodology): “[11] shows GGNN has better performance than the sequence-to-sequence models on the abstract mean representation (AMR) and neural machine translation (NMT) tasks, which shows the possibility of applying graph-to-sequence models in other sequence-to-sequence tasks.”
Explainability Techniques for Graph Convolutional Networks (2019, background): “…Zitnik et al., 2018), physics (Battaglia et al., 2016; Chang et al., 2017; Gilmer et al., 2017; Watters et al., 2017; Sanchez-Gonzalez et al., 2018), visual recognition (Qi et al., 2017; 2018; Narasimhan et al., 2018), and natural language processing (Bastings et al., 2017; Beck et al., 2018).”