Datasets

One packaged food product.

In your code: data/products.csv

• codetextkey

  Barcode (keep as text: many start with 0)

• product_nametext · can be empty

  Name as entered

• brandstext · can be empty

  Brands, comma-separated as entered

• quantitytext · can be empty

  Package size as free text

• search_categorytext

  Which of the six categories it was collected from

• nutriscore_gradetext · can be empty

  Nutri-Score a–e, or unknown / not-applicable

• nova_groupinteger · can be empty

  How processed it is, from 1 (unprocessed) to 4 (ultra-processed)

• ecoscore_gradetext · can be empty

  Environmental score a-plus–f, or unknown

• completenessdecimal · can be empty

  How complete the record is, 0–1

• created_attimestamp · can be empty

  When the product was added, UTC

• last_modified_attimestamp · can be empty

  Last edit, UTC

• energy_kcal_100gdecimal · can be empty

  Energy, kcal per 100 g

• fat_100gdecimal · can be empty

  Fat, g per 100 g

• saturated_fat_100gdecimal · can be empty

  Saturated fat, g per 100 g

• carbohydrates_100gdecimal · can be empty

  Carbohydrates, g per 100 g

• sugars_100gdecimal · can be empty

  Sugars, g per 100 g

• fiber_100gdecimal · can be empty

  Fibre, g per 100 g

• proteins_100gdecimal · can be empty

  Protein, g per 100 g

• salt_100gdecimal · can be empty

  Salt, g per 100 g

• sodium_100gdecimal · can be empty

  Sodium, g per 100 g

One tag on one product: categories, labels, allergens, traces, countries and ingredient analysis.

In your code: data/product_tags.csv

• codetextkey→ products

  Product barcode

• kindtextkey

  category, label, allergen, trace, country or analysis

• tagtextkey

  Tag, with language prefix (en:organic)

One packaged food product.

In your code: data/products.json

One packaged food product.

In your code: data/products.jsonl