ESCI
收藏arXiv2025-09-30 收录
下载链接:
https://github.com/shuttie/esci-s/
下载链接
链接失效反馈官方服务:
资源简介:
该数据集名为ESCI,是一个专注于产品搜索用例的大型基准数据集,涵盖了英语、日语和西班牙语查询。该数据集包含了一个较小的排名数据集,拥有48,300个独特的查询和1,118,011个相关性判断。数据集还包括了额外的元数据以及产品表示的图片。其规模超过300万条项目(主要关注英语)。该数据集的任务涵盖了查询-产品排名、多类别产品分类以及产品替代品识别。
This dataset, named ESCI, is a large-scale benchmark dataset focused on product search use cases, covering queries in English, Japanese and Spanish. It includes a smaller ranking dataset with 48,300 unique queries and 1,118,011 relevance judgments, as well as additional metadata and images representing products. It encompasses over 3 million items, with a primary focus on English. The tasks supported by this dataset cover query-product ranking, multi-class product classification, and product substitute identification.



