adisyonist/14-allergens-v2

Adisyonist Multilingual EU-14 Allergen Dataset

A multilingual dataset for predicting the 14 EU food allergens from restaurant menu item text.

The dataset contains approximately 294,500 examples consisting of menu item names and, where available, short menu descriptions.

It is designed for multi-label text classification, where a single menu item may contain zero, one, or multiple allergens.

---

Dataset Structure

Each record contains only three fields:

• id: Unique identifier for the example.
• text: Menu item name, optionally followed by a short description.
• allergens: List of EU-14 allergen labels associated with the item.

Example

{"id":"example_001","text":"Spaghetti Carbonara. Pasta with egg, parmesan cheese and pancetta.","allergens":["gluten","eggs","milk"]}

---

Allergen Labels

The dataset uses the following 14 canonical allergen labels:

gluten
crustaceans
eggs
fish
peanuts
soy
milk
tree_nuts
celery
mustard
sesame
sulphites
lupin
molluscs

An empty list indicates that none of the 14 allergen classes are assigned to the example:

{"id":"example_002","text":"Fresh Fruit Plate.","allergens":[]}

---

Intended Use

This dataset is intended for training and evaluating multilingual models that predict food allergens from restaurant menu text.

Typical applications include:

• Multi-label allergen classification
• Restaurant menu analysis
• Food information systems
• Multilingual food NLP
• Allergen detection research

---

Important Notice

Allergen labels represent dataset annotations and should not be treated as a substitute for verified ingredient, manufacturer, or restaurant allergen information.

Real recipes and preparation methods may vary.

---