LongSumm
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/guyfe/LongSumm
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含了1705个抽取式摘要和531个抽象式摘要，这些摘要均来自于自然语言处理和机器学习领域的科学论文。黄金摘要（即标准摘要）的平均长度为767个词元。该数据集基于自然语言处理和机器学习研究者的视频演讲和博客文章编制而成。其规模包括1705个抽取式摘要、531个抽象式摘要以及22篇盲测论文，任务是对科学论文进行摘要总结。
This dataset contains 1705 extractive summaries and 531 abstractive summaries, all sourced from scientific papers in the fields of natural language processing and machine learning. The average length of the gold standard summaries, i.e., the reference summaries, is 767 tokens. This dataset is compiled from video presentations and blog posts authored by researchers in natural language processing and machine learning. It includes 1705 extractive summaries, 531 abstractive summaries, and 22 blind-test papers, with the core task being scientific paper summarization.
提供机构：
Workshop on Scholarly Document Processing @EMNLP 2020
