MS^2

该数据集包含了超过47万份文献文档和2万份摘要，这些资料来源于科学研究领域，旨在助力开发能够评估和整合跨多项研究中的矛盾证据的系统。此外，该数据集还包含了被分类为9种类别的标注句子，这些类别在建模过程中被简化为背景、目标和其它三个类别。同时，数据集还提供了与临床知识相关的PICO元素的 结构化表示。其规模宏大，包含超过47万份文档和2万份摘要，所涉及的任务是多文档摘要。

This dataset comprises over 470,000 scholarly documents and 20,000 abstracts sourced from the scientific research domain, with the objective of facilitating the development of systems capable of evaluating and integrating conflicting evidence across multiple studies. Additionally, the dataset includes annotated sentences categorized into 9 original classes, which are simplified into three broad categories: Background, Objective, and Other during the modeling process. Meanwhile, structured representations of PICO elements related to clinical knowledge are also provided in this dataset. Boasting a large corpus with over 470,000 documents and 20,000 abstracts, this dataset targets the multi-document summarization task.