Multi-paragraph Reading Comprehension with Token-level Dynamic Reader and Hybrid Verifier

Multi-paragraph reading comprehension requires the model to infer answers of arbitrary user-generated questions by reasoning cross-passage information. Previous work usually generates answer by directly employing a pointer network to predict the start and end position of the answer. However, span-level reading is insufficient since intermediate words may matter more. In this paper, we propose a novel unified network that includes a selector, a Token-level dynamic reader, and a Hybrid verifier (TH-Net). The core of token-level dynamic reader is a gate mechanism which dynamically selects important intermediate words according to boundary words. We decide the reader score from each token being both the boundary and the content. Moreover, we adopt a hybrid network verifier considering semantic answer-answer and entailment question-answer relationships to robust the model in case of being fooled by adversarial answers. Our experiments on SQuAD-document, SQuAD-open, and Trivia-wiki datasets show significant and consistent improvement as compared to other baselines and achieve the state-of-the-art performance on two of them.

[1]  Wei Zhang,et al.  Evidence Aggregation for Answer Re-Ranking in Open-Domain Question Answering , 2017, ICLR.

[2]  Hai Zhao,et al.  Retrospective Reader for Machine Reading Comprehension , 2020, ArXiv.

[3]  Ming-Wei Chang,et al.  BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding , 2019, NAACL.

[4]  Yoon Kim,et al.  Convolutional Neural Networks for Sentence Classification , 2014, EMNLP.

[5]  Richard Socher,et al.  Efficient and Robust Question Answering from Minimal Context over Documents , 2018, ACL.

[6]  Alexandre Lacoste,et al.  Accurate Supervised and Semi-Supervised Machine Reading for Long Documents , 2017, EMNLP.

[7]  Yiming Yang,et al.  XLNet: Generalized Autoregressive Pretraining for Language Understanding , 2019, NeurIPS.

[8]  Wei Zhang,et al.  R3: Reinforced Ranker-Reader for Open-Domain Question Answering , 2018, AAAI.

[9]  Zhen Huang,et al.  Retrieve, Read, Rerank: Towards End-to-End Multi-Document Reading Comprehension , 2019, ACL.

[10]  Percy Liang,et al.  Adversarial Examples for Evaluating Reading Comprehension Systems , 2017, EMNLP.

[11]  Percy Liang,et al.  Know What You Don’t Know: Unanswerable Questions for SQuAD , 2018, ACL.

[12]  Deng Cai,et al.  Smarnet: Teaching Machines to Read and Comprehend Like Human , 2017, ArXiv.

[13]  Kai Liu,et al.  Multi-Passage Machine Reading Comprehension with Cross-Passage Answer Verification , 2018, ACL.

[14]  Jun Xu,et al.  HAS-QA: Hierarchical Answer Spans Model for Open-domain Question Answering , 2019, AAAI.

[15]  Zhiyuan Liu,et al.  Denoising Distantly Supervised Open-Domain Question Answering , 2018, ACL.

[16]  Luke S. Zettlemoyer,et al.  End-to-end Neural Coreference Resolution , 2017, EMNLP.

[17]  Luo Si,et al.  A Deep Cascade Model for Multi-Document Reading Comprehension , 2018, AAAI.

[18]  Eunsol Choi,et al.  TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehension , 2017, ACL.

[19]  Christopher Clark,et al.  Simple and Effective Multi-Paragraph Reading Comprehension , 2017, ACL.

[20]  An Yang,et al.  Enhancing Pre-Trained Language Representations with Rich Knowledge for Machine Reading Comprehension , 2019, ACL.

[21]  Ali Farhadi,et al.  Bidirectional Attention Flow for Machine Comprehension , 2016, ICLR.

[22]  Jian Zhang,et al.  SQuAD: 100,000+ Questions for Machine Comprehension of Text , 2016, EMNLP.

[23]  Zhen Wang,et al.  Joint Training of Candidate Extraction and Answer Selection for Reading Comprehension , 2018, ACL.

[24]  Jimmy J. Lin,et al.  End-to-End Open-Domain Question Answering with BERTserini , 2019, NAACL.

[25]  Zhuosheng Zhang,et al.  SG-Net: Syntax-Guided Machine Reading Comprehension , 2019, AAAI.

[26]  Furu Wei,et al.  Read + Verify: Machine Reading Comprehension with Unanswerable Questions , 2018, AAAI.

[27]  David Berthelot,et al.  WikiReading: A Novel Large-scale Language Understanding Task over Wikipedia , 2016, ACL.

[28]  Phil Blunsom,et al.  Teaching Machines to Read and Comprehend , 2015, NIPS.

[29]  Ming Zhou,et al.  Reinforced Mnemonic Reader for Machine Reading Comprehension , 2017, IJCAI.

[30]  Jason Weston,et al.  Reading Wikipedia to Answer Open-Domain Questions , 2017, ACL.

[31]  Hai Zhao,et al.  Dual Co-Matching Network for Multi-choice Reading Comprehension , 2020, AAAI.

[32]  Lukasz Kaiser,et al.  Attention is All you Need , 2017, NIPS.

[33]  Omer Levy,et al.  Jointly Predicting Predicates and Arguments in Neural Semantic Role Labeling , 2018, ACL.

[34]  Rajarshi Das,et al.  Multi-step Retriever-Reader Interaction for Scalable Open-domain Question Answering , 2019, ICLR.

[35]  Navdeep Jaitly,et al.  Pointer Networks , 2015, NIPS.