Hierarchical Graph Network for Multi-hop Question Answering

Abstract

In this paper, we present Hierarchical Graph Network (HGN) for multi-hop question answering. To aggregate clues from scattered texts across multiple paragraphs, a hierarchical graph is created by constructing nodes on different levels of granularity (questions, paragraphs, sentences, entities), the representations of which are initialized with pre-trained contextual encoders. Given this hierarchical graph, the initial node representations are updated through graph propagation, and multi-hop reasoning is performed via traversing through the graph edges for each subsequent sub-task (e.g., paragraph selection, supporting facts extraction, answer prediction). By weaving heterogeneous nodes into an integral unified graph, this hierarchical differentiation of node granularity enables HGN to support different question answering sub-tasks simultaneously. Experiments on the HotpotQA benchmark demonstrate that the proposed model achieves new state of the art, outperforming existing multi-hop QA approaches.

arXiv commentAccepted to EMNLP 2020

References in corpus (13)

• Longformer: The Long-Document Transformer
• Bidirectional Attention Flow for Machine Comprehension
• Machine Comprehension Using Match-LSTM and Answer Pointer
• Big Bird: Transformers for Longer Sequences
• NewsQA: A Machine Comprehension Dataset
• Multi-hop Question Answering via Reasoning Chains
• Coarse-grain Fine-grain Coattention Network for Multi-evidence Question Answering
• Dynamically Fused Graph Network for Multi-hop Reasoning
• Is Graph Structure Necessary for Multi-hop Question Answering?
• Multi-step Entity-centric Information Retrieval for Multi-Hop Question Answering
• Cluster-Former: Clustering-based Sparse Transformer for Long-Range Dependency Encoding
• Multi-Hop Paragraph Retrieval for Open-Domain Question Answering
• Simple yet Effective Bridge Reasoning for Open-Domain Multi-Hop Question Answering

Cited by in corpus (17)

• Big Bird: Transformers for Longer Sequences
• Multi-hop Question Answering via Reasoning Chains
• Is Graph Structure Necessary for Multi-hop Question Answering?
• A Survey on Explainability in Machine Reading Comprehension
• Answering Complex Open-Domain Questions with Multi-Hop Dense Retrieval
• Learning Representations of Bi-level Knowledge Graphs for Reasoning beyond Link Prediction
• ExplanationLP: Abductive Reasoning for Explainable Science Question Answering
• Natural Language QA Approaches using Reasoning with External Knowledge
• Iterative Hierarchical Attention for Answering Complex Questions over Long Documents
• Graph Sequential Network for Reasoning over Sequences
• Generating Followup Questions for Interpretable Multi-hop Question Answering
• Deep Neural Networks for Relation Extraction
• Coarse-grained decomposition and fine-grained interaction for multi-hop question answering
• A Graph Representation of Semi-structured Data for Web Question Answering
• Fact-level Extractive Summarization with Hierarchical Graph Mask on BERT
• Fine-tuning Multi-hop Question Answering with Hierarchical Graph Network
• Parameter-Efficient Neural Question Answering Models via Graph-Enriched Document Representations