Dynamically Fused Graph Network for Multi-hop Reasoning

Abstract

Text-based question answering (TBQA) has been studied extensively in recent years. Most existing approaches focus on finding the answer to a question within a single paragraph. However, many difficult questions require multiple supporting evidence from scattered text among two or more documents. In this paper, we propose Dynamically Fused Graph Network(DFGN), a novel method to answer those questions requiring multiple scattered evidence and reasoning over them. Inspired by human's step-by-step reasoning behavior, DFGN includes a dynamic fusion layer that starts from the entities mentioned in the given query, explores along the entity graph dynamically built from the text, and gradually finds relevant supporting entities from the given documents. We evaluate DFGN on HotpotQA, a public TBQA dataset requiring multi-hop reasoning. DFGN achieves competitive results on the public board. Furthermore, our analysis shows DFGN produces interpretable reasoning chains.

Accepted by ACL 19

References in corpus (5)

• Teaching Machines to Read and Comprehend
• Bidirectional Attention Flow for Machine Comprehension
• Large-scale Simple Question Answering with Memory Networks
• Stochastic Answer Networks for Machine Reading Comprehension
• Reasoning with Memory Augmented Neural Networks for Language Comprehension

Cited by in corpus (16)

• Graph Neural Networks: A Review of Methods and Applications
• Graph Neural Networks: Taxonomy, Advances and Trends
• Learning to Retrieve Reasoning Paths over Wikipedia Graph for Question Answering
• Machine Reading Comprehension: The Role of Contextualized Language Models and Beyond
• Multi-hop Question Generation with Graph Convolutional Network
• A Survey on Explainability in Machine Reading Comprehension
• Hierarchical Graph Network for Multi-hop Question Answering
• Adapting Meta Knowledge Graph Information for Multi-Hop Reasoning over Few-Shot Relations
• xERTE: Explainable Reasoning on Temporal Knowledge Graphs for Forecasting Future Links
• MMM: Multi-stage Multi-task Learning for Multi-choice Reading Comprehension
• Graph Sequential Network for Reasoning over Sequences
• Answering Any-hop Open-domain Questions with Iterative Document Reranking
• Learning to Recover Reasoning Chains for Multi-Hop Question Answering via Cooperative Games
• Multi-hop Reading Comprehension across Documents with Path-based Graph Convolutional Network
• Parameter-Efficient Neural Question Answering Models via Graph-Enriched Document Representations
• Decomposing Complex Questions Makes Multi-Hop QA Easier and More Interpretable