recsys-slates
收藏资源简介:
本数据集名为recsys-slates,由奥斯陆大学和FINN.no创建,记录了用户与在线市场之间的序列交互。数据集包含3740万条记录,涵盖了用户在30天内与市场的互动,包括推荐和搜索结果。数据集详细记录了每次交互中展示的商品列表(称为slates),用户是否点击以及点击的商品。此数据集旨在通过实际展示的商品列表,帮助构建更真实的用户偏好模型,并促进使用重要性采样等方法进行更稳健的离线评估。
This dataset, named recsys-slates, was created by the University of Oslo and FINN.no, and documents sequential user interactions with an online marketplace. It contains 37.4 million records covering user-market interactions over a 30-day period, including both recommendation and search results. The dataset meticulously records the item lists (termed slates) displayed during each interaction, as well as whether the user clicked any item and the specific clicked item. This dataset is designed to help build more realistic user preference models using the actual displayed item lists, and to facilitate more robust offline evaluation via methods such as importance sampling.

- 1FINN.no Slates Dataset: A new Sequential Dataset Logging Interactions, allViewed Items and Click Responses/No-Click for Recommender Systems Research奥斯陆大学和FINN.no · 2021年



