Answers are grounded in this paper and cite their sources.
Turning a food description into the equivalent FoodEx2 code has so far been a job for a person working with a tool, in several manual steps. FoodEx2 Encoder, a new tool I developed, uses a large language model to remove those steps: it takes an English food description and returns a complete FoodEx2 code, with a base term, facet descriptors, and labelled free text for whatever the catalogue cannot express. The model only compares the description with catalogue terms; what the resulting code means is defined by the catalogue, not by the model. The encoder is open source (MIT) and also available as a paid hosted service that plugs into AI chat tools through MCP. This article describes why it was built, how an encoding works, how to judge a result, and what the encoder cannot do yet.
No takes yet. Share an insight, caveat, or question.
Mineur Jorn (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: