Cognitive Graph for Multi-Hop Reading Comprehension at Scale

We propose a new CogQA framework for multi-hop question answering in web-scale documents. Inspired by the dual process theory in cognitive science, the framework gradually builds a \textit{cognitive graph} in an iterative process by coordinating an implicit extraction module (System 1) and an explicit reasoning module (System 2). While giving accurate answers, our framework further provides explainable reasoning paths. Specifically, our implementation based on BERT and graph neural network efficiently handles millions of documents for multi-hop reasoning questions in the HotpotQA fullwiki dataset, achieving a winning joint $F_1$ score of 34.9 on the leaderboard, compared to 23.6 of the best competitor.

[1]  Sebastian Riedel,et al.  Constructing Datasets for Multi-hop Reading Comprehension Across Documents , 2017, TACL.

[2]  Jonathan Evans.  Dual-processing accounts of reasoning, judgment, and social cognition. , 2008, Annual review of psychology.

[3]  Yelong Shen,et al.  ReasoNet: Learning to Stop Reading in Machine Comprehension , 2016, CoCo@NIPS.

[4]  Patrick R. Wheeler,et al.  Thinking, Fast and Slow. , 2013 .

[5]  Kevin Gimpel,et al.  Bridging Nonlinearities and Stochastic Regularizers with Gaussian Error Linear Units , 2016, ArXiv.

[6]  Wei Wang,et al.  Multi-Granularity Hierarchical Attention Fusion Networks for Reading Comprehension and Question Answering , 2018, ACL.

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

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

[9]  Gonzalo Navarro,et al.  A guided tour to approximate string matching , 2001, CSUR.

[10]  Jonathan Berant,et al.  The Web as a Knowledge-Base for Answering Complex Questions , 2018, NAACL.

[11]  Jonathan Evans.  In two minds: dual-process accounts of reasoning , 2003, Trends in Cognitive Sciences.

[12]  Ellen M. Voorhees,et al.  The TREC-8 Question Answering Track Report , 1999, TREC.

[13]  Jonathan Evans.  Heuristic and analytic processes in reasoning , 1984 .

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

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

[16]  M. D’Esposito.  Working memory. , 2008, Handbook of clinical neurology.

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

[18]  Jason Weston,et al.  The Goldilocks Principle: Reading Children's Books with Explicit Memory Representations , 2015, ICLR.

[19]  Nicholas J. Belkin,et al.  Interaction with Texts: Information Retrieval as Information-Seeking Behavior , 1993, Information Retrieval.

[20]  Sanda M. Harabagiu,et al.  The Structure and Performance of an Open-Domain Question Answering System , 2000, ACL.

[21]  Richard Socher,et al.  Ask Me Anything: Dynamic Memory Networks for Natural Language Processing , 2015, ICML.

[22]  Ali Farhadi,et al.  Query-Reduction Networks for Question Answering , 2016, ICLR.

[23]  Xavier Bresson,et al.  Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering , 2016, NIPS.

[24]  Max Welling,et al.  Semi-Supervised Classification with Graph Convolutional Networks , 2016, ICLR.

[25]  Razvan Pascanu,et al.  Relational inductive biases, deep learning, and graph networks , 2018, ArXiv.

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

[27]  S. Sloman.  The empirical case for two systems of reasoning. , 1996 .

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

[29]  Ming Zhou,et al.  Gated Self-Matching Networks for Reading Comprehension and Question Answering , 2017, ACL.

[30]  Kevin Gimpel,et al.  Gaussian Error Linear Units (GELUs) , 2016 .

[31]  Yingxu Wang,et al.  Discovering the Capacity of Human Memory , 2003 .

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

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

[34]  Yoshua Bengio,et al.  HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answering , 2018, EMNLP.

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