Improving Machine Reading Comprehension with General Reading Strategies

Sun, Kai; Yu, Dian; Yu, Dong; Cardie, Claire

doi:10.18653/v1/n19-1270

Cited by 107 publications

(84 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We demonstrate its effectiveness on two challenging multi-sentence reasoning datasets: MultiRC (Khashabi et al, 2018) and OpenBookQA (Mihaylov et al, 2018). Multee using ELMo contextual embeddings (Peters et al, 2018) matches state-of-the-art results achieved with large transfomer-based models (Radford et al, 2018) that were trained on a sequence of large scale tasks (Sun et al, 2019).…”

Section: Our Implementation Ofmentioning

confidence: 64%

See 1 more Smart Citation

Repurposing Entailment for Multi-Hop Question Answering Tasks

Trivedi¹,

Kwon²,

Khot³

et al. 2019

Proceedings of the 2019 Conference of the North

View full text Add to dashboard Cite

Question Answering (QA) naturally reduces to an entailment problem, namely, verifying whether some text entails the answer to a question. However, for multi-hop QA tasks, which require reasoning with multiple sentences, it remains unclear how best to utilize entailment models pre-trained on large scale datasets such as SNLI, which are based on sentence pairs. We introduceMultee, a general architecture that can effectively use entailment models for multi-hop QA tasks.Multee uses (i) a local module that helps locate important sentences, thereby avoiding distracting information, and (ii) a global module that aggregates information by effectively incorporating importance weights. Importantly, we show that both modules can use entailment functions pre-trained on a large scale NLI datasets. We evaluate performance on MultiRC and Open-BookQA, two multihop QA datasets. When using an entailment function pre-trained on NLI datasets, Multee outperforms QA models trained only on the target QA datasets and the OpenAI transformer models. The code is available at https://github. com/StonyBrookNLP/multee.

show abstract

Section: Our Implementation Ofmentioning

confidence: 64%

“…While ELMo contextual embeddings helped in MultiRC, it did not help OpenBookQA. We believe this is in part due to the mismatch between our ELMo training setup where all sentences are treated as a single sequence, which, while true in 8 Published on arXiv on Oct 31, 2018 (Sun et al, 2019). MultiRC, is not the case in OpenBookQA.…”

Section: Resultsmentioning

confidence: 99%

Repurposing Entailment for Multi-Hop Question Answering Tasks

Trivedi¹,

Kwon²,

Khot³

et al. 2019

Proceedings of the 2019 Conference of the North

View full text Add to dashboard Cite

show abstract

“…It is worth noting that recent large-scale language models (LMs) (Devlin et al, 2019;Radford et al, 2018) have now been applied on this task, leading to improved state-of-the-art results (Sun et al, 2018;Banerjee et al, 2019;Pan et al, 2019). However, our knowledge-gap guided approach to QA is orthogonal to the underlying model.…”

Section: Openbookqa Resultsmentioning

confidence: 99%

What’s Missing: A Knowledge Gap Guided Approach for Multi-hop Question Answering

Khot¹,

Sabharwal

Clark³

2019

Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conferen

View full text Add to dashboard Cite

Multi-hop textual question answering requires combining information from multiple sentences. We focus on a natural setting where, unlike typical reading comprehension, only partial information is provided with each question. The model must retrieve and use additional knowledge to correctly answer the question. To tackle this challenge, we develop a novel approach that explicitly identifies the knowledge gap between a key span in the provided knowledge and the answer choices. The model, GapQA, learns to fill this gap by determining the relationship between the span and an answer choice, based on retrieved knowledge targeting this gap. We propose jointly training a model to simultaneously fill this knowledge gap and compose it with the provided partial knowledge. On the Open-BookQA dataset, given partial knowledge, explicitly identifying what's missing substantially outperforms previous approaches.

show abstract

“…Multi-hop RC: There exist several different data sets that require reasoning in multiple steps in literature, for example bAbI (Weston et al, 2015), MultiRC (Khashabi et al, 2018) and Open-BookQA (Mihaylov et al, 2018). A lot of systems have been proposed to solve the multi-hop RC problem with these data sets (Sun et al, 2018;Wu et al, 2019). However, these data sets require multi-hop reasoning over multiple sentences or multiple common knowledge while the problem we want to solve in this paper requires collecting evidences across multiple documents.…”

Section: Related Workmentioning

confidence: 99%

Multi-hop Reading Comprehension across Multiple Documents by Reasoning over Heterogeneous Graphs

Tao¹,

Wang²,

Huang³

et al. 2019

Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics

136

View full text Add to dashboard Cite

Multi-hop reading comprehension (RC) across documents poses new challenge over singledocument RC because it requires reasoning over multiple documents to reach the final answer. In this paper, we propose a new model to tackle the multi-hop RC problem. We introduce a heterogeneous graph with different types of nodes and edges, which is named as Heterogeneous Document-Entity (HDE) graph. The advantage of HDE graph is that it contains different granularity levels of information including candidates, documents and entities in specific document contexts. Our proposed model can do reasoning over the HDE graph with nodes representation initialized with co-attention and self-attention based context encoders. We employ Graph Neural Networks (GNN) based message passing algorithms to accumulate evidences on the proposed HDE graph. Evaluated on the blind test set of the Qangaroo WIKIHOP data set, our HDE graph based single model delivers competitive result, and the ensemble model achieves the state-of-the-art performance.

show abstract

Improving Machine Reading Comprehension with General Reading Strategies

Abstract: This work was done when K. S. was an intern at the Tencent AI Lab, Bellevue, WA. our fine-tuned models that incorporate these strategies. Core code is available at https: //github.com/nlpdata/strategy/.

Cited by 107 publications

References 42 publications

Repurposing Entailment for Multi-Hop Question Answering Tasks

Repurposing Entailment for Multi-Hop Question Answering Tasks

What’s Missing: A Knowledge Gap Guided Approach for Multi-hop Question Answering

Multi-hop Reading Comprehension across Multiple Documents by Reasoning over Heterogeneous Graphs

Contact Info

Product

Resources

About