Baidu-ULTR Subset
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/lixsh6/Tencent_wsdm_cup2023
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是百度-ULTR数据集的一个子集，包含了105万2950个搜索会话，以及专家标注的带有点击信息的查询。除了包含启发式特征、分数特征和嵌入特征外，该专家标注数据集按照20%验证集和80%测试集的比例进行划分。作为一个亿级别规模的数据集，它包含了105万2950个会话和3万4047个独特的查询，其任务是进行无偏学习排序（Ultr）。
This dataset is a subset of the Baidu-ULTR dataset, which contains 1,052,950 search sessions and expert-annotated queries with click-through information. In addition to heuristic features, score features and embedding features, this expert-annotated dataset is divided into a 20% validation set and an 80% test set. As a hundred-million-scale dataset, it includes 1,052,950 sessions and 34,047 unique queries, with the task of unbiased learning to rank (Ultr).
提供机构：
Baidu
