Deep Graph Convolutional Encoders for Structured Data to Text Generation

By Diego Marcheggiani, Laura Perez-Beltrachini. 2018

127 citations recorded by the source corpus.

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.

Subfields: Artificial Intelligence, Graph Learning, Natural Language Generation

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 Graph Learning

In Natural Language Generation

Citations in the corpus

GraDA: Graph Generative Data Augmentation for Commonsense Reasoning (2022, background): “Marcheggiani and Perez-Beltrachini (2018) encode input graphs using a graph convolutional encoder (Kipf and Welling, 2017).”
Text-to-Text Pre-Training for Data-to-Text Tasks (2020, methodology): “Liu et al. (2018) generate text by conditioning language models on tables, Puduppully et al. (2019) explicitly model entities and Marcheggiani and Perez-Beltrachini (2018) encode structured data using graph convolutional networks.”
Structural Information Preserving for Graph-to-Text Generation (2020, methodology): “GCNEC (Marcheggiani and Perez-Beltrachini, 2018) is a recent model using a graph convolution network (Kipf and Welling, 2017) for encoding KGs.”
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 interm”
Structural Neural Encoders for AMR-to-text Generation (2019, methodology): “al Language Processing, GCNs have been used for Semantic Role Labeling (Marcheggiani and Titov, 2017), NMT (Bastings et al., 2017), Named Entity Recognition (Cetoli et al., 2017) and text generation (Marcheggiani and Perez-Beltrachini, 2018). A graph-to-sequence neural network was ﬁrst introduced by”
Modeling Global and Local Node Contexts for Text Generation from Knowledge Graphs (2020, unknown): “Song et al. (2018) propose the graph recurrent network (GRN) to directly encode the AMR nodes, whereas Beck et al. (2018) develop a model based on GGNNs (Li et al., 2016).”
Improving Text-to-Text Pre-trained Models for the Graph-to-Text Task (2020, background): “or graph-based encoders have been proposed to better process the KG triple input (Vougiouklis et al., 2018; Distiawan et al., 2018; Marcheggiani and Perez-Beltrachini, 2018; Schmitt et al., 2020; Zhao et al., 2020).”
AggGen: Ordering and Aggregating while Generating (2021, background): “For applications where structured inputs are present, neural graph encoders (Marcheggiani and Perez-Beltrachini, 2018; Rao et al., 2019; Gao et al., 2020) or decoding of explicit graph references (Logan et al.”
Logic2Text: High-Fidelity Natural Language Generation from Logical Forms (2020, background): “Graph2seq+copy There is a line of research for graph neural network based encoders, such as (Marcheggiani and Perez-Beltrachini, 2018; Xu et al., 2018), etc.”
Line Graph Enhanced AMR-to-Text Generation with Mix-Order Graph Attention Networks (2020, unknown): “Moving to graph-to-sequence approaches, Marcheggiani and Perez-Beltrachini (2018) ﬁrst show that graph neural networks can signiﬁcantly improve the generation performance by explicitly encoding the structure of the graph.”
Enhancing AMR-to-Text Generation with Dual Graph Representations (2019, unknown): “Our approach markedly differs from existing graph-to-sequence models for MR-to-Text generation (Marcheggiani and Perez Beltrachini, 2018; Beck et al., 2018; Damonte and Cohen, 2019) in that these approaches aggregate all the immediate neighborhood information of a node in a single representation.”
Bridging the Structural Gap Between Encoding and Decoding for Data-To-Text Generation (2020, methodology): “To narrow the ensuing structural gap, we use another GCN-based neural planner to create a sequential content plan of this graph, which is represented as a re-ordered sequence of its nodes.”