全球COVID-19信息聚合系统
计算与语言
2020-10-13 v2
摘要
COVID-19全球大流行使公众密切关注相关新闻，涵盖卫生、治疗和教育影响等多个领域。同时，各国（如政策和疫情发展）的COVID-19状况差异很大，因此民众会对国外新闻感兴趣。我们构建了一个全球COVID-19信息聚合系统，包含来自7种语言10个地区的按主题分类的可靠文章。我们通过众包收集的可信COVID-19相关网站数据集保证了文章质量。神经机器翻译模块将其他语言文章译为日语和英语。基于BERT的话题分类器在我们的文章-话题对数据集上训练，通过将文章归入不同类别，帮助用户高效找到感兴趣的信息。
引用
@article{arxiv.2008.01523,
title = {A System for Worldwide COVID-19 Information Aggregation},
author = {Akiko Aizawa and Frederic Bergeron and Junjie Chen and Fei Cheng and Katsuhiko Hayashi and Kentaro Inui and Hiroyoshi Ito and Daisuke Kawahara and Masaru Kitsuregawa and Hirokazu Kiyomaru and Masaki Kobayashi and Takashi Kodama and Sadao Kurohashi and Qianying Liu and Masaki Matsubara and Yusuke Miyao and Atsuyuki Morishima and Yugo Murawaki and Kazumasa Omura and Haiyue Song and Eiichiro Sumita and Shinji Suzuki and Ribeka Tanaka and Yu Tanaka and Masashi Toyoda and Nobuhiro Ueda and Honai Ueoka and Masao Utiyama and Ying Zhong},
journal= {arXiv preprint arXiv:2008.01523},
year = {2020}
}
备注
Accepted to EMNLP 2020 Workshop NLP-COVID