JDRec dataset

该数据集名为JDRec，是首个大规模的组合物品推荐数据集，源自京东推荐系统的决策日志。该数据集包含超过752万个样本，分为训练集和测试集。数据集中既包含候选物品也包含已选物品，同时为确保敏感信息的安全，对敏感特征进行了混洗和转换处理。该数据集的规模属于大型，拥有752万个样本，其任务是组合物品推荐。

This dataset is named JDRec, the first large-scale combined item recommendation dataset derived from the decision logs of the JD.com recommendation system. It contains over 7.52 million samples, which are divided into training and test sets. The dataset includes both candidate items and selected items. To ensure the security of sensitive information, sensitive features have been shuffled and transformed. As a large-scale dataset with 7.52 million samples, its core task is combined item recommendation.

搜集汇总

背景与挑战

背景概述

JDRec数据集是一个用于推荐系统离线实验的数据集，包含53个字段，覆盖商品、用户交互和统计特征。每个样本由44个商品项组成，其中前4个为在线重排选中的商品，数据提供CSV和JSON两种格式，适用于强化学习等推荐算法研究。