mteb/JaCWIRRerankingLite

JaCWIR（Japanese Casual Web IR）是一个由来自Hatena Bookmark的问题和网页元描述组成的数据集。这是一个轻量级的重新排序版本，包含一个减少的语料库（188,033个文档），使用来自5个高性能模型的硬负样本构建。数据集主要用于文本排序和文档检索任务，适用于日语信息检索评估。JaCWIR (Japanese Casual Web IR) is a dataset consisting of questions and webpage meta descriptions collected from Hatena Bookmark. This is the lightweight reranking version with a reduced corpus (188,033 documents) constructed using hard negatives from 5 high-performance models. The dataset is primarily used for text-ranking and document-retrieval tasks, suitable for Japanese information retrieval evaluation.