Product data lives here.
Committed. The 100-product zip provided by the organizers, unpacked. Four categories:
1_美妆护肤/— beauty / skincare2_数码电子/— digital / electronics3_服饰运动/— clothing / sports4_食品生活/— food / lifestyle
Each category has data/ (JSON) and images/ (JPG).
Caveat: confirmed AI-generated by the recruiters. Use for smoke-test and pipeline development; not for the final demo. See ../docs/DATA.md for how to source real data.
Gitignored. Real datasets sourced from HuggingFace / Kaggle / scraped sources go here. Each subdirectory should match the seed schema.
Canonical reference: seed/1_美妆护肤/data/p_beauty_001.json. Required top-level keys:
{
"product_id": "p_<category>_<n>",
"title": "...",
"brand": "...",
"category": "...",
"sub_category": "...",
"base_price": 0.0,
"image_path": "<category>/images/<file>.jpg",
"skus": [...],
"rag_knowledge": {
"marketing_description": "...",
"official_faq": [{"question": "...", "answer": "..."}],
"user_reviews": [{"nickname": "...", "rating": 0, "content": "..."}]
}
}