Qdrant/hm_ecommerce_products

该数据集是Kaggle竞赛数据集H&M个性化时尚推荐的增强版本，经过清洗并增加了预计算的嵌入和可访问的图片URL，以促进时尚推荐研究和多模态检索应用。数据集包含约106,000种时尚产品的详细元数据、约137万客户的 demographic 信息、历史交易数据、通过S3 URL访问的产品图片、预计算的密集嵌入（BGE-small-en-1.5）和稀疏嵌入（SPLADEv1）。数据集适用于时尚推荐系统、多模态检索、嵌入评估、时尚趋势分析和向量数据库应用。

This dataset is a processed and enhanced version of the H&M Personalized Fashion Recommendations Kaggle competition dataset. The original dataset has been cleaned and augmented with pre-computed embeddings and accessible image URLs to facilitate fashion recommendation research and multimodal retrieval applications. It includes ~106,000 fashion products with detailed metadata, ~1.37 million customers with demographic information, historical transaction data, product images accessible via S3 URLs, pre-computed dense embeddings (BGE-small-en-1.5), and pre-computed sparse embeddings (SPLADEv1). The dataset is ideal for fashion recommendation systems, multimodal retrieval, embedding evaluation, fashion trend analysis, and vector database applications.