Wikit/PIRE
收藏Hugging Face2025-02-24 更新2025-04-12 收录
下载链接：
https://hf-mirror.com/datasets/Wikit/PIRE
下载链接
链接失效反馈官方服务：
资源简介：
PIRE数据集是一个由手动创建的查询和一个PDF文件语料库组成的数据集。对于每个查询，都有标注的相关文档、页面和文本段落。该数据集可用于评估PDF文件上的信息检索策略，并分为两个部分：chunk.single和chunk.multi，分别表示查询可以由单个段落或多个段落回答。每个查询的相关文本段落都来自一个或多个PDF文件，并且标有对应的页码。
The PIRE dataset consists of manually created queries and a corpus of PDF files. For each query, relevant documents, pages, and text passages have been annotated. This dataset can be used to evaluate information retrieval strategies on PDF files and is divided into two parts: chunk.single and chunk.multi, indicating that queries can be answered by a single passage or multiple passages, respectively. Each querys relevant text passages come from one or more PDF files and are marked with corresponding page numbers.
提供机构：
Wikit
