Epicure: Navigating the Emergent Geometry of Food Ingredient Embeddings

TL;DR

Epicure introduces multilingual ingredient embeddings trained on a large recipe corpus, revealing emergent geometric structures that reflect culinary and chemical relationships.

Contribution

The paper presents a novel multilingual ingredient embedding approach using a large recipe dataset and explores their emergent geometry through different random walk schemas.

Findings

Three distinct embedding models capture different aspects of ingredient relationships.

A large, normalized multilingual recipe corpus was created and used for training.

The models reveal meaningful culinary and chemical structures in ingredient space.

Abstract

We present Epicure, a family of three sibling skip-gram ingredient embeddings retrained from scratch on a multilingual recipe corpus. We aggregate 4.14M recipes from 11 sources spanning seven languages, English, Chinese, Russian, Vietnamese, Spanish, Turkish, Indonesian, German, and Indian-English, and normalise the raw ingredient strings to 1,790 canonical entries via an LLM-augmented pipeline. A 203,508-edge ingredient-ingredient NPMI graph and an 80,019-edge typed FlavorDB ingredient-compound graph, 2,247 typed compound nodes across 15 categories, seed three Metapath2Vec variants that share architecture and hyperparameters and differ only in the random-walk schema: Cooc walks the co-occurrence graph only, Chem walks the typed compound metapaths only, and Core blends both via injected ingredient-ingredient walks at controlled mixing, placing each model at a distinct point on the chemistry-vs-recipe-context spectrum.