clinia/CUREv1

CUREv1数据集由Clinia的医疗团队策划，用于评估检索器在由医疗专业人员策划的查询-段落对上的性能，涵盖10个医学学科和3种跨语言设置（英语到英语、法语到英语、西班牙语到英语）。数据集包含多个配置文件，每个配置文件对应不同的数据文件和语言设置。数据集的创建旨在解决健康信息检索领域缺乏涵盖广泛医学学科和跨语言能力的数据集的问题。数据集的结构包括多个文件夹，每个文件夹代表一个医学学科，包含查询、语料库和相关判断文件。数据集的创建过程包括数据收集、处理、注释和过滤，以确保数据的高质量。

The CUREv1 dataset, curated by Clinia’s Medical Team, is designed to evaluate the performance of retrievers on query-passage pairs curated by medical professionals, across 10 medical disciplines and 3 cross-lingual settings (English-to-English, French-to-English, Spanish-to-English). The dataset includes multiple configurations, each corresponding to different data files and language settings. The creation of the dataset addresses the lack of datasets in health information retrieval that cover a broad array of medical disciplines and cross-lingual capabilities. The dataset is organized into multiple folders, each representing a medical discipline, and contains query, corpus, and relevance judgment files. The dataset creation process involves data collection, processing, annotation, and filtering to ensure high-quality data.