sarthakrastogi/amazon-search-dataset

亚马逊搜索数据集是通过SerpApi抓取的真实亚马逊产品数据，覆盖了44个搜索查询，包含1094个（查询，产品）对。查询设计为复杂且多约束的（价格+功能+颜色排除，品牌+规格等），以测试检索系统的性能。数据集用于基准测试search-expert的混合搜索管道与亚马逊原生搜索排名的对比。数据集包含多个字段，如查询、亚马逊排名、产品标题、描述、品牌、价格、折扣、评分、评论数量等。

The Amazon Search Dataset consists of real Amazon product data scraped via SerpApi across 44 search queries, covering 1094 (query, product) pairs. Queries are deliberately complex and multi-constraint (price + feature + color exclusion, brand + spec, etc.) to stress-test retrieval systems. The dataset is used to benchmark search-experts hybrid search pipeline against Amazons native search ranking. It includes various fields such as query, Amazon rank, product title, description, brand, price, discount, rating, number of reviews, etc.