Multi Passage MRC

Models - Multi-Passage Machine Reading Comprehension

Multi-passage datasets:

• ComplexWebQuestions (CWebQ) - 2018
• QAngaroo - 2018
  • WikiHop
  • MedHop
• HotpotQA - 2018
• RC-QED - 2019

List of models/papers

Year	Title	Model	Datasets	Focus/Misc	Paper, Source Code
2019 (Nov 9)	Hierarchical Graph Network for Multi-hop Question Answering		HotpotQA		paper
2019 (Oct 7)	Multi-hop Question Answering via Reasoning Chains		HotpotQA, WikiHop		paper
2019	Token-level Dynamic Self-Attention Network for Multi-Passage Reading Comprehension	Dynamic Self-attention Network	SearchQA, Quasar-T, WikiHop		paper
2019	Retrieve, Read, Rerank: Towards End-to-End Multi-Document Reading Comprehension	RE^3QA	TriviaQA, SQuAD		paper, code
2019	Cognitive Graph for Multi-Hop Reading Comprehension at Scale	CogQA (BERT + graph neural network)	HotpotQA		paper, code
2019	Multi-hop reading comprehension across multiple documents by reasoning over heterogeneous graphs	Heterogeneous Document Entity graph	Qangaroo Wikihop		paper
2019	Explore, Propose, and Assemble: An Interpretable Model for Multi-Hop Reading Comprehension	Explore-Propose-Assemble reader	WikiHop, MedHop		paper, code
2019	Multi-hop Reading Comprehension through Question Decomposition and Rescoring	DECOMPRC	HotpotQA		paper, code
2019	Exploiting Explicit Paths for Multi-hop Reading Comprehension	PathNet	WikiHop		paper, code
2019	Answering while Summarizing: Multi-task Learning for Multi-hop QA with Evidence Extraction	Query Focused Extractor	HotpotQA, FEVER		paper
2019	Multi-Hop Paragraph Retrieval for Open-Domain Question Answering	MUPPET	HotpotQA, SquAD		paper, code
2019	Revealing the Importance of Semantic Retrieval for Machine Reading at Scale		HotpotQA, FEVER		paper, code
2019	Dynamically Fused Graph Network for Multi-hop Reasoning	Dynamically Fused Graph Network (DFGN)	HotpotQA		paper, code
2019	Answering Complex Open-domain Questions Through Iterative Query Generation	GOLDEN (Gold Entity) Retriever	HotpotQA		paper, code
2019	Multi-hop Reading Comprehension via Deep Reinforcement Learning based Document Traversal		QAngaroo		paper
2019	Select, Answer and Explain: Interpretable Multi-hop Reading Comprehension over Multiple Documents	SAE	HotpotQA		paper
2019	Question Answering by Reasoning Across Documentswith Graph Convolutional Networks		WikiHop		paper
2019	COARSE-GRAIN FINE-GRAIN COATTENTION NETWORK FOR MULTI-EVIDENCE QUESTION ANSWERING		WikiHop		paper
2018	Neural Models for Reasoning over Multiple Mentions using Coreference	Coref-GRU			paper
2018	Exploring Graph-structured Passage Representation for Multi-hop Reading Comprehension with Graph Neural Networks		WikiHop + CWebQ		paper

Analysis Multi-Passage datasets:

• Jifan Chen and Greg Durrett, Understanding Dataset Design Choices for Multi-hop Reasoning, NAACL 2019, paper.
• Sewon Min, Eric Wallace, et al., Compositional Questions Do Not Necessitate Multi-hop Reasoning, ACL 2019, paper, code.