Dynamically Fused Graph Network for Multi-hop Reasoning

By Yunxuan Owen Xiao, Yanru Qu, Lin Qiu, Hao Zhou, Lei Li, Weinan Zhang, Yong Yu. 2019

213 citations recorded by the source corpus.

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.

Subfields: Artificial Intelligence, Natural Language Processing, Graph Learning

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 Natural Language Processing

In Graph Learning

Citations in the corpus

Reasoning Like Program Executors (2022, methodology): “(ii) Similarly, on HotpotQA (Distractor), specialized model baselines include DFGN (Qiu et al., 2019), SAE (Tu et al., 2020), C2F Reader (Shao et al., 2020) and the SOTA model HGN (Fang et al., 2020).”
Learning to Retrieve Reasoning Paths over Wikipedia Graph for Question Answering (2019, neutral): “0 – – – – DFGN (Xiao et al., 2019) – – – – 69.”
Text Modular Networks: Learning to Decompose Tasks in the Language of Existing Models (2021, background): “, 2020) have targetted smaller subsets to develop models with interpretable reasoning Similar approaches have been used to produce reasoning chains for multi-hop datasets (Chen et al., 2019; Kundu et al., 2019; Xiao et al., 2019) question relevant to the input context.”
Differentiable Reasoning over a Virtual Knowledge Base (2020, background): “Recent papers have also looked at complex QA using graph neural networks (Sun et al., 2018; Cao et al., 2019; Xiao et al., 2019) or by identifying paths of entities in text (Jiang et al.”
Hierarchical Graph Network for Multi-hop Question Answering (2019, methodology): “DFGN (Xiao et al., 2019) constructs a dynamic entity graph, where in each reasoning step irrelevant entities are softly masked out and a fusion module”
Multi-hop Question Answering via Reasoning Chains (2019, neutral): “DecompRC (Min et al., 2019b) 55.20 69.63 N QFE (Nishida et al., 2019) 53.86 68.06 Y DFGN (Qiu et al., 2019) 56.31 69.69 Y HGN (Fang et al., 2019) 66.07 79.36 Y SAE (Tu et al., 2019a) 66”
DDRQA: Dynamic Document Reranking for Open-domain Multi-hop Question Answering (2020, methodology): “Inspired by Qiu et al. (2019), we apply a dynamic soft mask on the entities, serving as the information “gatekeeper” which assigns more weights to entities pertaining to the question.”
Is Graph Structure Necessary for Multi-hop Question Answering? (2020, methodology): “61 DFGN (Qiu et al., 2019) 33.”
Teaching Pre-Trained Models to Systematically Reason Over Implicit Knowledge (2020, background): “, [30, 31, 32, 33, 34], where again model behavior is often opaque [35, 36, 37].”
DAGN: Discourse-Aware Graph Network for Logical Reasoning (2021, methodology): “Graph-based methods (Qiu et al., 2019; De Cao et al., 2019; Cao et al., 2019; Ran et al., 2019; Chen et al., 2020b; Xu et al., 2020b; Zhang et al., 2020) explicitly model the reasoning process with constructed graphs, then learn and update features through message passing based on graphs.”
Low-Resource Generation of Multi-hop Reasoning Questions (2020, methodology): “By referring to the leaderboard on the HotpotQA website, we chose an open-source model for MRC-QA, named Dynamically Fused Graph Network (DFGN) (Qiu et al., 2019), which achieved the state-of-the-art at the paper submission time.”
Graph Sequential Network for Reasoning over Sequences (2020, methodology): “Since the point of our paper is to show the efﬁcacy of the proposed GSN model, we only show results on development set; getting numbers on test set requires several other modules (Xiao et al., 2019; Tu et al., 2019a) that are unrelated to our proposed GSN.”
Language Generation with Multi-hop Reasoning on Commonsense Knowledge Graph (2020, background): “In multi-hop question answering tasks, the reasoning process is augmented with entity graph (Cao et al., 2019; Qiu et al., 2019) or concept paths (Bauer et al.”
Select, Answer and Explain: Interpretable Multi-hop Reading Comprehension over Multiple Documents (2019, methodology): “Third, current applications of GNN for QA tasks usually take entities (De Cao, Aziz, and Titov 2019; Tu et al. 2019; Xiao et al. 2019) as graph nodes and reasoning is achieved by conducting message passing over nodes with contextual information.”
Find or Classify? Dual Strategy for Slot-Value Predictions on Multi-Domain Dialog State Tracking (2019, background): “Recent work on question answering has shown that the joint encoding of query-context pairs is crucial to achieving high accuracy (Qiu et al., 2019; Asai et al., 2019) of our proposed DS-DST model.”
CogLTX: Applying BERT to Long Texts (2020, methodology): “Previous methods usually leverage the graph structure between key entities across the paragraphs [13, 36].”
Is Graph Structure Necessary for Multi-hop Reasoning? (2020, unknown): “We use the open-source implementation of DFGN1 (Qiu et al., 2019), the published state-of-the-art”
Leap-Of-Thought: Teaching Pre-Trained Models to Systematically Reason Over Implicit Knowledge (2020, background): “It similarly distinguishes our work from multi-hop QA, e.g., [25, 26, 27, 28, 29], where again model behavior is often opaque [30].”
A Simple Yet Strong Pipeline for HotpotQA (2020, methodology): “, 2020) and models (Min et al., 2019b; Xiao et al., 2019; Tu et al., 2020) with varying complexities have been proposed over the past few years.”