Multi-News
收藏资源简介:
Multi-News是由耶鲁大学计算机科学系创建的第一个大规模多文档新闻摘要数据集,包含56,216个文章-摘要对。该数据集从newser.com收集新闻文章和专业编辑编写的人工摘要,旨在解决多文档新闻事件的全面和非冗余摘要挑战。数据集内容丰富,涵盖超过1,500个新闻来源,适用于训练和测试多文档摘要模型,特别是在处理不同焦点和观点的新闻事件时。
Multi-News is the first large-scale multi-document news summarization dataset created by the Department of Computer Science at Yale University. It contains 56,216 article-summary pairs. This dataset collects news articles and human-written summaries produced by professional editors from newser.com, aiming to address the challenge of generating comprehensive and non-redundant summaries for multi-document news events. The dataset has rich content, covering over 1,500 news sources, and is suitable for training and testing multi-document summarization models, especially when dealing with news events with different focuses and perspectives.




