On April 22, 2026, Tome Eftimov presented recent research on Natural Language Processing (NLP) for food and nutrition at the EuroFIR Food Forum 2026, a key event bringing together experts working on food composition data, interoperability, and innovation in the food domain.

The presentation, titled “Food Data Management and Natural Language Processing”, provided an overview of more than 12 years of research focused on transforming unstructured food-related data (e.g., scientific texts, recipes, and dietary guidelines) into structured, interoperable knowledge. The work spans the development of information extraction pipelines (NER, NEL, relation extraction), semantic resources, and knowledge graphs, enabling advanced applications such as decision support and question answering.

The talk highlighted key milestones, including:

-- The evolution of food NLP from limited entity recognition systems to fully integrated pipelines for knowledge base construction -- The CAFETERIA project, which introduced the first systematically annotated food corpora and enabled relation extraction in the domain -- Recent advances in domain-specific AI, such as FoodyLLM, a specialized large language model for food and nutrition tasks, and FoodBench-QA, the first benchmark for grounded food and nutrition question answering -- Emerging ontology-aware and training-free approaches (e.g., RAG-based pipelines) for scalable and interoperable knowledge extraction across food and biomedical domains

These contributions address critical challenges in food data interoperability and support the development of AI-driven solutions for nutrition, health, and policy-making.

The EuroFIR Food Forum 2026 featured dedicated sessions on food data innovation and highlighted the importance of collaboration in advancing food composition data infrastructures and applications