Structured Neural Summarization

Abstract

Summarization of long sequences into a concise statement is a core problem in natural language processing, requiring non-trivial understanding of the input. Based on the promising results of graph neural networks on highly structured data, we develop a framework to extend existing sequence encoders with a graph component that can reason about long-distance relationships in weakly structured data such as text. In an extensive evaluation, we show that the resulting hybrid sequence-graph models outperform both pure sequence models as well as pure graph models on a range of summarization tasks.

Published in ICLR 2019 https://openreview.net/forum?id=H1ersoRqtm

References in corpus (9)

• Neural Message Passing for Quantum Chemistry
• Teaching Machines to Read and Comprehend
• Gated Graph Sequence Neural Networks
• The Natural Language Decathlon: Multitask Learning as Question Answering
• Fast Abstractive Summarization with Reinforce-Selected Sentence Rewriting
• A parallel corpus of Python functions and documentation strings for automated code documentation and code generation
• Graph Partition Neural Networks for Semi-Supervised Classification
• CodeSum: Translate Program Language to Natural Language
• Deep Learning to Detect Redundant Method Comments

Cited by in corpus (27)

• CodeXGLUE: A Machine Learning Benchmark Dataset for Code Understanding and Generation
• Natural Attack for Pre-trained Models of Code
• Reconstructive Sequence-Graph Network for Video Summarization
• Code Structure Guided Transformer for Source Code Summarization
• On the Generalizability of Neural Program Models with respect to Semantic-Preserving Program Transformations
• Deep Graph Matching and Searching for Semantic Code Retrieval
• Neural Reverse Engineering of Stripped Binaries using Augmented Control Flow Graphs
• Graph Neural Networks for Natural Language Processing: A Survey
• Retrieval-Augmented Generation for Code Summarization via Hybrid GNN
• Interpreting Graph Neural Networks for NLP With Differentiable Edge Masking
• Adversarial Robustness for Code
• Code Prediction by Feeding Trees to Transformers
• Discourse-Aware Neural Extractive Text Summarization
• Towards Demystifying Dimensions of Source Code Embeddings
• ATOM: Commit Message Generation Based on Abstract Syntax Tree and Hybrid Ranking
• Learning to Represent Programs with Heterogeneous Graphs
• CoCoSum: Contextual Code Summarization with Multi-Relational Graph Neural Network
• Enhancing Scientific Papers Summarization with Citation Graph
• TreeCaps: Tree-Based Capsule Networks for Source Code Processing
• Document Modeling with Graph Attention Networks for Multi-grained Machine Reading Comprehension
• Code Summarization with Structure-induced Transformer
• Enhancing Function Name Prediction using Votes-Based Name Tokenization and Multi-Task Learning
• Teaching Temporal Logics to Neural Networks
• GraphSearchNet: Enhancing GNNs via Capturing Global Dependencies for Semantic Code Search
• Pointing to Subwords for Generating Function Names in Source Code
• Keywords Guided Method Name Generation
• TAG : Type Auxiliary Guiding for Code Comment Generation