LiveQA
收藏arXiv2020-10-02 更新2024-06-21 收录
下载链接：
https://github.com/PKU-TANGENT/LiveQA
下载链接
链接失效反馈官方服务：
资源简介：
LiveQA是一个基于NBA比赛实时直播构建的新型问答数据集，由北京大学教育部计算语言学重点实验室创建。该数据集包含117,050个多选题，源自1,670场NBA比赛的中文虎扑网站直播评论。LiveQA旨在测试模型对基于时间线的直播内容的推理能力，涉及时间线理解、事件追踪和数学计算。数据集的应用领域主要集中在提升机器阅读理解模型对时间敏感信息的处理能力，解决现有数据集在时间推理方面的不足。
LiveQA is a novel question answering (QA) dataset constructed based on real-time NBA game broadcasts, created by the Key Laboratory of Computational Linguistics, Ministry of Education, Peking University. This dataset contains 117,050 multiple-choice questions sourced from live commentary on the Chinese Hupu website for 1,670 NBA games. LiveQA aims to evaluate models' reasoning capabilities for timeline-based live content, covering timeline comprehension, event tracking and mathematical calculation. Its main application scenarios focus on improving the ability of machine reading comprehension models to process time-sensitive information, and addressing the deficiencies of existing datasets in temporal reasoning.
提供机构：
教育部计算语言学重点实验室，北京大学
创建时间：
2020-10-02
