PeerSum

PeerSum是一个用于抽象式多文档摘要的新数据集，使用科学出版物的同行评审。该数据集的特点是其摘要（即元评审）高度抽象，是源文档（即评审）的真实摘要，并包含了源文档之间的分歧。PeerSum is a novel dataset for abstractive multi-document summarization, constructed from peer reviews of scientific publications. A key feature of this dataset is that its summaries (i.e., meta-reviews) are highly abstract, serve as authentic summaries of the source documents (i.e., individual peer reviews), and incorporate disagreements among these source documents.