seaforestrain/Amazon-Reviews-2023

这是一个由McAuley实验室于2023年收集的大规模亚马逊评论数据集。该数据集包含丰富的特征：1. 用户评论（如评分、文本、有用性投票等）；2. 商品元数据（如描述、价格、原始图像等）；3. 链接（用户-商品/一起购买图）。数据集的新特性包括：数据量更大（共571.54M条评论，比上一版本增加245.2%）；交互时间更新（覆盖1996年5月至2023年9月）；元数据更丰富（商品描述性特征更多）；时间戳更细粒度（精确到秒或更细级别）；处理更清洁（商品元数据比之前版本更干净）；数据分割标准化（鼓励推荐系统基准测试）。数据集涵盖33个类别，基本统计显示总用户数约5451万，商品数约4819万，评论令牌数约301.4亿，元数据令牌数约307.8亿。This is a large-scale Amazon review dataset collected by the McAuley Lab in 2023. The dataset contains rich features: 1. User reviews (including ratings, review texts, helpfulness votes, etc.); 2. Product metadata (including descriptions, prices, raw images, etc.); 3. Links (user-item and co-purchase graphs). Novel characteristics of this dataset are as follows: larger data scale with a total of 571.54 million reviews, representing a 245.2% increase compared to the previous version; updated interaction time coverage spanning from May 1996 to September 2023; richer metadata with more descriptive product features; finer-grained timestamps (accurate to the second or more granular levels); more rigorous data cleaning (cleaner product metadata than prior versions); and standardized data splits to facilitate recommendation system benchmarking. The dataset covers 33 categories. Basic statistics indicate that the total number of users is approximately 54.51 million, the number of products is approximately 48.19 million, the count of review tokens is around 30.14 billion, and the count of metadata tokens is approximately 30.78 billion.