disi-unibo-nlp/feast-foodex2-terms

FoodEx2 Terms — Taxonomy Reference

This dataset contains the complete FoodEx2 taxonomy developed by the European Food Safety Authority (EFSA) — a hierarchical classification system for describing and coding food items consumed in Europe.

Dataset Structure

Single train split with one row per FoodEx2 term (~29,900 rows).

Schema

Column	Type	Description
termCode	string	Unique FoodEx2 term identifier (e.g. "A03LG")
termExtendedName	string	Full human-readable term name (e.g. "Herbal and other non-tea infusions")
termScopeNote	string	Optional free-text description/definition of the term
masterParentCode	string	Code of the direct parent term in the hierarchy
implicitFacets	string	Facet descriptors implicitly associated with this term

Taxonomy Structure

FoodEx2 organises food terms in a tree hierarchy where each term has exactly one parent (masterParentCode). The tree spans multiple levels, from broad food groups down to specific food items.

Example lineage:

Root
 └── Beverages and water-based beverages
      └── Infusions, decoctions and other water-based drinks
           └── Tea and tea-based beverages
                └── Herbal and other non-tea infusions  (termCode: A03LG)

Beyond baseterms, the taxonomy includes facet descriptors — terms that annotate properties of a food such as physical state, fat content, or production method. These are organised into facet families (e.g. F03 = Physical state, F04 = Fat/oil coverage, F10 = Fat content).

Companion Dataset

See disi-unibo-nlp/feast-foodex2 for the annotated food description dataset built on top of this taxonomy.

Source

The FoodEx2 taxonomy is developed and maintained by the European Food Safety Authority (EFSA):

• FoodEx2 system description
• EFSA food classification portal

This companion taxonomy dataset is used together with the annotated FoodEx2 case-study dataset made available in the European Commission Funding & Tenders Portal documentation:

• TED notice: 150198-2025
• European Commission Funding & Tenders Portal tender details: 745b51da-4cfd-4ce9-86f6-ae0d6738f67a-CN
• Original dataset file referenced by the tender documents: FoodEx2-CaseStudy2-Dataset_V1.xlsx

This Hugging Face repository is an unofficial research redistribution/processing of public FoodEx2 source material for reproducibility of the FEAST experiments. The authoritative sources remain the EFSA resources and European Commission portal linked above. No endorsement by, or affiliation with, the European Commission, HaDEA, or EFSA is implied. Users should consult the original sources and applicable terms before reusing the data.

Citation

If you use this dataset, please cite the paper that introduced it:

@inproceedings{molfetta2025feast,
  author    = {Lorenzo Molfetta and
               Alessio Cocchieri and
               Stefano Fantazzini and
               Giacomo Frisoni and
               Luca Ragazzi and
               Gianluca Moro},
  title     = {{FEAST:} Retrieval-Augmented Multi-Hierarchical Food Classification
               for the {FoodEx2} System},
  booktitle = {Proceedings of the 28th European Conference on Artificial Intelligence
               ({ECAI} 2025)},
  series    = {Frontiers in Artificial Intelligence and Applications},
  pages     = {4169--4176},
  publisher = {{IOS} Press},
  year      = {2025},
  doi       = {10.3233/FAIA251309}
}