QMSum
收藏资源简介:
QMSum数据集是由复旦大学等机构的研究人员创建,专注于多领域会议摘要任务。该数据集包含232个会议的1808个查询-摘要对,覆盖产品、学术和委员会会议等多个领域。数据集的创建过程涉及主题分割、查询生成和基于查询的摘要标注,旨在通过灵活的查询机制满足不同用户对会议内容的不同需求。QMSum数据集的应用领域广泛,旨在帮助用户高效获取会议关键信息,支持自动化会议摘要技术的研究与开发。
The QMSum dataset was created by researchers from Fudan University and other institutions, focusing on the multi-domain meeting summarization task. It contains 1808 query-summary pairs from 232 meetings, covering multiple domains such as product, academic, and committee meetings. The construction process of the QMSum dataset involves topic segmentation, query generation, and query-based summary annotation, aiming to meet diverse user needs for meeting content through a flexible query mechanism. The QMSum dataset has wide application scenarios, designed to help users efficiently obtain key meeting information and support the research and development of automated meeting summarization technologies.




