Cognitive Graph for Multi-Hop Reading Comprehension at Scale

By Ming Ding, Chang Zhou, Qibin Chen, Hongxia Yang, Jie Tang. 2019

246 citations recorded by the source corpus.

We propose a new CogQA framework for multi-hop reading comprehension question answering in web-scale documents. Founded on the dual process theory in cognitive science, the framework gradually builds a 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 23.1 of the best competitor.

Subfields: Artificial Intelligence, Natural Language Processing

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

Citations in the corpus

Reasoning Like Program Executors (2022, background): “Reasoning via Specialized Models Early works typically design specialized models and augment them into LMs for different types of questions (Dua et al., 2019; Andor et al., 2019; Hu et al., 2019; Ding et al., 2019).”
Learning to Retrieve Reasoning Paths over Wikipedia Graph for Question Answering (2019, methodology): “Cognitive Graph (Ding et al., 2019) incorporates entity links between documents for multi-hop QA to extend the list of retrieved documents.”
Knowledge Guided Text Retrieval and Reading for Open Domain Question Answering (2019, methodology): “Other graph retrieval methods have been developed, either using entity name matching (Ding et al., 2019; Xiong et al., 2019b; Godbole et al., 2019) or hyperlinks (Asai et al.”
Revealing the Importance of Semantic Retrieval for Machine Reading at Scale (2019, methodology): “Previous systems (Ding et al., 2019; Yang et al., 2018) on H OTPOT QA treat supporting fact retrieval (sentence-level retrieval) just as an auxiliary task for providing extra model explainability.”
Adaptive Information Seeking for Open-Domain Question Answering (2021, methodology): “…strategies, employing the same retrieval function multiple times (Das et al., 2019a; Feldman and El-Yaniv, 2019; Xiong et al., 2021; Ding et al., 2019; Qi et al., 2019; Zhang et al., 2020; Qi et al., 2020) or pre-designated sequence of applying retrieval functions (Asai et al.,…”
Pre-Trained Models: Past, Present and Future (2021, methodology): “The success of CogQA and CogLTX may provide some thoughts on this challenge.”
MultiQA: An Empirical Investigation of Generalization and Transfer in Reading Comprehension (2019, background): “…in (Tay et al., 2018) for N EWS QA, in Lin et al. (2018), for S EARCH QA, in Das et al. (2019) for TQA-U, in (Talmor and Berant, 2018b) for CWQ, in Ding et al. (2019) for H OTPOT QA, in (Abujabal et al., 2018) for C OM QA, and in Bao et al. (2016) for CQ. train on all examples in the large…”
Self-Supervised Learning: Generative or Contrastive (2020, background): “DEEP neural networks [71] have shown outstanding performance on various machine learning tasks, especially on supervised learning in computer vision (image classification [26], [48], [54], semantic segmentation [39], [76]), natural language processing (pre-trained language models [27], [68], [75], [”
Differentiable Reasoning over a Virtual Knowledge Base (2020, background): “Hence, Sun et al. (2019) and Ding et al. (2019), incorporate a dynamic retrieval process to add text about entities identiﬁed as relevant in the previous layer of the model.”
Logic-Guided Data Augmentation and Regularization for Consistent Question Answering (2020, neutral): “As we only consider the questions whose answers can be extracted from p , we do not add any modiﬁ-cations to the model unlike some previous studies in HotpotQA (Min et al., 2019b; Ding et al., 2019).”
Neural, symbolic and neural-symbolic reasoning on knowledge graphs (2021, background): “Recently for question answering upon text corpus, neural-symbolic models incorporating human cognition have shown their superiority on reasoning capacity (Ding et al., 2019), as they consistently perform well with the increase of the hops in questions.”
Accelerating Training of Transformer-Based Language Models with Progressive Layer Dropping (2020, unknown): “It has been demonstrated recently that Transformer-based networks have obtained superior performance in many NLP tasks (e.g., the GLUE benchmark [6] and the challenging multi-hop reasoning task [7]) than recurrent neural networks or convolutional neural networks.”
Simple yet Effective Bridge Reasoning for Open-Domain Multi-Hop Question Answering (2019, methodology): “Note that the CogQA paper does not provide details of the entity linker, so the results with our entity linker may not be the same to the one used by CogQA.”
On the Advance of Making Language Models Better Reasoners (2022, background): “…symbolic systems, including knowledge graphs (Mihaylov and Frank, 2018; Bauer et al., 2018; Kundu et al., 2019; Wang et al., 2019; Lin et al., 2019; Ding et al., 2019; Feng et al., 2020; Wang et al., 2022b), or question taxonomies (Dua et al., 2019; Andor et al., 2019; Hu et al., 2019b; Wang et…”
Transformer-XH: Multi-Evidence Reasoning with eXtra Hop Attention (2020, methodology): “Similar to previous work (Ding et al., 2019), the textual representation for each node in the graph is the [SEP]-delimited concatenation of the question, anchor text (the text in the hyperlink in parent nodes pointing to the child node), and the paragraph itself.”
Answer Complex Questions: Path Ranker Is All You Need (2021, methodology): “Cognitive Graph [6] trains a model to predict the spans for next-hop which helps to extract the most evidential entity.”
Towards Knowledge-Based Recommender Dialog System (2019, unknown): “, 2018), question answering (Rajpurkar et al., 2016; Yang et al., 2018; Ding et al., 2019) and natural language generation (Liu et al.”