Reading Comprehension has received significant attention in recent years as high quality Question Answering (QA) datasets have become available. Despite state-of-the-art methods achieving strong overall accuracy, Multi-Hop (MH) reasoning remains particularly challenging. To address MH-QA specifically, we propose a Deep Reinforcement Learning based method capable of learning sequential reasoning across large collections of documents so as to pass a query-aware, fixed-size context subset to existing models for answer extraction. Our method is comprised of two stages: a linker, which decomposes the provided support documents into a graph of sentences, and an extractor, which learns where to look based on the current question and already-visited sentences. The result of the linker is a novel graph structure at the sentence level that preserves logical flow while still allowing rapid movement between documents. Importantly, we demonstrate that the sparsity of the resultant graph is invariant to context size. This translates to fewer decisions required from the Deep-RL trained extractor, allowing the system to scale effectively to large collections of documents.
The importance of sequential decision making in the document traversal step is demonstrated by comparison to standard IE methods, and we additionally introduce a BM25-based IR baseline that retrieves documents relevant to the query only. We examine the integration of our method with existing models on the recently proposed QAngaroo benchmark and achieve consistent increases in accuracy across the board, as well as a 2-3x reduction in training time.
A Navigational Approach for Comprehensive RAG via Traversal over Proposition Graphs
Idiap Research Institute（日内瓦研究所）；École Polytechnique Fédérale de Lausanne（联邦理工学院洛桑分校）；University of Manchester（曼彻斯特大学）；CRUK Manchester Institute（CRUK曼彻斯特研究所）
Breaking the Static Graph: Context-Aware Traversal for Robust Retrieval-Augmented Generation
Huawei Hong Kong Research Center（华为香港研究中心）；The Hong Kong University of Science and Technology（香港科学与技术大学）；The Chinese University of Hong Kong, Shenzhen（香港中文大学（深圳））
Medical Knowledge Graph QA for Drug-Drug Interaction Prediction based on Multi-hop Machine Reading Comprehension
School of Cyber Science and Engineering, Qufu Normal University（曲阜师范大学网络空间安全学院）；School of Computer Science and Technology, East China Normal University（华东师范大学计算机科学与技术学院）；Network and Information Center, Qufu Normal University（曲阜师范大学网络与信息中心）；School of Electronics and Information Engineering, Harbin Institute of Technology, Shenzhen（哈尔滨工业大学（深圳）电子与信息工程学院）
ClueReader: Heterogeneous Graph Attention Network for Multi-hop Machine Reading Comprehension
Qufu Normal University（曲阜师范大学）；East China Normal University（华东师范大学）；Shanghai Institute of AI for Education（上海人工智能教育研究院）；Harbin Institute of Technology（哈尔滨工业大学）；HUTECH University（胡志明市工业大学）；Andalusian Research Institute in Data Science and Computational Intelligence (DaSCI)（安达卢西亚数据科学与计算智能研究院（DaSCI））；University of Granada（格拉纳达大学）；Iwate Prefectural University（岩手县立大学）
我们的目标是跨语言机器阅读理解（MRC）的任务，在直接zero-shot设置，通过合并的句法特征，从普遍的相似性（UD），我们使用的关键特征是每个句子内的句法关系。虽然以前的工作已经证明了有效的语法引导的MRC模型，我们建议采用句间的句法关系，除了基本的句内关系，进一步利用在多句输入的MRC任务的句法依赖。在我们的方法中，我们建立了句子间依赖图（ISDG）连接依赖树，以形成跨句子的全局句法关系。然后，我们提出了ISDG编码器，编码的全球依赖图，通过一跳和多跳依赖路径明确地解决句子间的关系。在三个多语言MRC数据集（XQuAD，MLQA，TyDiQA-GoldP）上的实验表明，我们的编码器只在英语上训练，能够提高所有14个测试集的zero-shot性能，覆盖8种语言，平均提高高达3.8 F1 / 5.2 EM，某些语言的5.2 F1 / 11.2 EM。进一步的分析表明，这种改进可以归因于对跨语言一致性句法路径的关注。
ViMMRC 2.0 -- Enhancing Machine Reading Comprehension on Vietnamese Literature Text
Faculty of Information Science and Engineering, University of Information Technology, Ho Chi Minh City, Vietnam（信息科学与工程学院，信息科技大学，胡志明市，越南）；Vietnam National University, Ho Chi Minh City, Vietnam（越南国家大学，胡志明市，越南）
Coarse-grained decomposition and fine-grained interaction for multi-hop question answering
School of Electronic and Information, Beijing Jiaotong University（北京交通大学电子信息学院）；Key Laboratory of Communication and Information Systems, Beijing Municipal Commission of Education（北京市教育委员会通信与信息系统重点实验室）
StepChain GraphRAG: Reasoning Over Knowledge Graphs for Multi-Hop Question Answering
University of Technology Sydney, Australia（澳大利亚技术大学）；Data61, CSIRO, Australia（数据61，CSIRO）；School of Engineering, Edith Cowan University, Australia（埃迪斯·科温大学工程学院）；University of New South Wales, Australia（新南威尔士大学）
Is Multihop QA in DiRe Condition? Measuring and Reducing Disconnected Reasoning
Stony Brook University（石溪大学）；Allen Institute for AI（艾伦人工智能研究所）
在多跳问答方面有真正的进展吗？模型通常利用数据集工件来产生正确的答案，而不将多个支持事实的信息连接起来。这限制了我们衡量真实进展的能力，并违背了构建多跳QA数据集的目的。我们为解决这一问题作出三项贡献。首先，我们将这种不受欢迎的行为形式化为跨支持事实子集的断开推理。这允许开发一个模型不可知的探针，用于测量任何模型可以通过断开推理欺骗多少。其次，使用\emph{contrastive support sufficiency}的概念，我们引入了现有数据集的自动转换，减少了断开推理的数量。第三，我们的实验表明，在阅读理解环境中，多跳问答并没有取得太大的进展。对于最近的一个大规模模型（XLNet），我们发现，在HotpotQA上的F1得分72中，只有18分是通过多事实推理获得的，与简单的RNN基线大致相同。我们的转换大大减少了断开推理（答案F1中的19分）。它是对抗性办法的补充，共同产生进一步的削减。
PokeMQA: Programmable knowledge editing for Multi-hop Question Answering
School of Artificial Intelligence, Jilin University（吉林大学人工智能学院）；North Carolina State University（北卡罗来纳州立大学）；Texas A&M University（德克萨斯农工大学）；University of Georgia（佐治亚大学）
S$^3$HQA: A Three-Stage Approach for Multi-hop Text-Table Hybrid Question Answering
Institute of Automation, Chinese Academy of Sciences（中国科学院自动化研究所）；School of Artificial Intelligence, University of Chinese Academy of Sciences（中国科学院大学人工智能学院）
NovelQA: Benchmarking Question Answering on Documents Exceeding 200K Tokens
Westlake University（西湖大学）；University of Waterloo（多伦多大学）；Hangzhou Normal University（杭州师范大学）；Shanghai AI Lab（上海人工智能实验室）；SJTU（上海交通大学）；NYU Shanghai（纽约大学上海分校）
Consensus Attention-based Neural Networks for Chinese Reading Comprehension
iFLYTEK Research（科大讯飞研究）；Research Center for Social Computing and Information Retrieval,Harbin Institute of Technology（哈尔滨工业大学社会计算与信息检索研究中心）；Harbin Institute of Technology（哈尔滨工业大学）
