基于深度交互融合网络的多跳机器阅读理解  

Deep Interactive Fusion Network for Multi-hop Reading Comprehension

作　　者：朱斯琪 过弋[1,2,3] 王业相 ZHU Siqi;GUO Yi;WANG Yexiang(Department of Computer Science and Engineering,East China University of Science and Technology,Shanghai 200237,China;Business Intelligence and Visualization Research Center,National Engineering Laboratory for Big Data Distribution and Exchange Technologies,Shanghai 200237,China;Shanghai Engineering Research Center of Big Data&Internet Audience,Shanghai 200072,China)

机构地区：[1]华东理工大学信息科学与工程学院,上海200237 [2]大数据流通与交易技术国家工程实验室-商业智能与可视化研究中心,上海200237 [3]上海大数据与互联网受众工程技术研究中心,上海200072

出　　处：《中文信息学报》2022年第5期67-75,共9页Journal of Chinese Information Processing

基　　金：国家重点研发计划(2018YFC0807105);国家自然科学基金(61462073);上海市科学技术委员会科研计划项目(17DZ1101003,18511106602,18DZ2252300)。

摘　　要：近年来,多跳机器阅读理解已经吸引了众多学者的关注,其要从多个文档中提取与问题相关的线索并回答问题。但很少有工作注重在段落选择时和回答问题时的多个段落之间的交互与融合,然而这对于多跳推理任务来说是至关重要的。因此,该文提出了一种针对多跳推理机器阅读理解的多段落深度交互融合的方法,首先从多个段落中筛选出与问题相关的段落,然后将得到的“黄金段落”输入到一个深度交互融合的网络中以聚集不同段落之间的信息,最终得到问题的答案。该文实验基于HotpotQA数据集,所提方法与基准模型相比,精确匹配(EM)提升18.5%,F_(1)值提升18.47%。Multi-hop reading comprehension requiring information from multiple documents has attracted much attention. However, the interaction between paragraphs is less addressed, no matter in the gold paragraph selection or in question answering. In this paper, we propose a multi-paragraph deep interactive fusion network for multi-hop reading comprehension. First, we filter out paragraphs irrelevant to the query to reduce the impact of distractors on model performance. Then, the selected documents are further input to a deep interactive fusion network to aggregate information from different paragraphs for the final answer. Experiment on HotpotQA dataset demonstrates that our model achieves the improvements of 18.5% according to EM and 18.47% according to F_(1)-score compared with the baseline.

关 键 词：多跳推理 机器阅读理解 多段落融合 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]