AutoLearn loader

Project Information

Project Overview

The AI4FOOD project aims to develop a comprehensive and interoperable Slovenian Food Composition Database (FCDB) enhanced with artificial intelligence methods and integrated with a knowledge base.

Food composition data is fundamental for:

  • 1. Nutrition research
  • 2. Public health
  • 3. Food safety
  • 4. Sustainable food systems
  • 5. Personalized medicine

Despite its importance, existing food composition databases face several limitations:

  • 1. Incomplete coverage of foods and nutrients.
  • 2. Inconsistent metadata.
  • 3. Limited interoperability with other datasets.
  • 4. Expensive laboratory analysis needed for full coverage.

AI4FOOD addresses these challenges by combining AI methods, semantic technologies, and domain expertise to automatically gather, integrate, and validate food composition data from multiple sources.

The project will:

  • 1. Collect missing food composition data from structured databases and scientific literature.
  • 2. Create semantic resources and ontologies for food data.
  • 3. Build a food knowledge graph linking foods, nutrients, chemicals, and diseases.
  • 4. Develop a food foundation model using large language models (LLMs).
  • 5. Validate the results through medical, industrial, and research studies.

The project aligns with major European strategies such as:

  • 1. Green Deal
  • 2. Farm to Fork Strategy
  • 3. EU Artificial Intelligence Act

and supports the digital transformation of food systems.

Objectives

The project pursues the following strategic objectives:

SO1

Develop AI methods to enhance and expand food composition data by integrating missing values from international databases and scientific literature.

SO2

Develop semantic resources and tools to enable interoperability between food composition data and knowledge resources.

SO3

Update the Slovenian Food Composition Database and create a food knowledge base and foundation model.

SO4

Validate the database and AI models through medical studies, industrial experiments, and research validation.

SO5

Ensure effective dissemination, communication, and stakeholder engagement.

Implementation Phases

The project is organized into six work packages (WPs).

WP1 — AI methods for gathering missing food composition data

Development of AI pipelines to collect and predict missing nutrient values from structured databases, scientific literature, recipes, and food product data.

WP2 — AI methods for data interoperability

Development of semantic annotation tools, knowledge graph construction pipelines, and ontology-based integration methods.

WP3 — Completing the Slovenian Food Composition Database

Integration of collected data into the national FCDB and development of a food-specific large language model (LLM) and Retrieval-Augmented Generation (RAG) system.

WP4 — Validation

Validation of the database and AI models through:

  • 1. Medical research studies
  • 2. Industrial food analysis
  • 3. AI model evaluation
WP5 — Dissemination and communication

Scientific publications, conferences, project website, and public engagement activities.

WP6 — Project management and coordination

Project coordination, reporting, risk management, and compliance with ethical and data management standards.

Project Team

The AI4FOOD project brings together researchers from Jožef Stefan Institute and Institute of Nutrition (NUTRIS).

The team combines expertise in::

  • 1. Food composition analysis
  • 2. Analytical chemistry
  • 3. Artificial intelligence
  • 4. Natural language processing
  • 3. Machine learning
  • 3. Knowledge graphs
  • 3. Nutrition science

Prof. Dr. Barbara Koroušić Seljak

Project coordinator

Computer Systems Department, Jožef Stefan Institute.

Barbara Koroušić Seljak

Prof. Dr. Nives Ogrinc

Food chemistry and isotope analysis

Department of Environmental Sciences, Jožef Stefan Institute.

Frank Hutter

Prof. Igor Pravst

Nutrition science

Institute of Nutrition.

Frank Hutter

Dr. Tome Eftimov

Artificial intelligence and machine learning

Computer Systems Department, Jožef Stefan Institute.

Frank Hutter

Dr. Eva Valenčič

Food data management

Computer Systems Department, Jožef Stefan Institute.

Frank Hutter

Dr. Ana Gjorgjevikj

AI methods for food data

Computer Systems Department, Jožef Stefan Institute.

Frank Hutter

Dr. Lidija Strojnik

Analytical chemistry

Department of Environmental Sciences, Jožef Stefan Institute.

Frank Hutter

Ana Nikolikj

PhD researcher

Computer Systems Department, Jožef Stefan Institute.

Frank Hutter

Gjorgjina Cenikj

PhD researcher

Computer Systems Department, Jožef Stefan Institute.

Frank Hutter

Jan Drole

MSc researcher

Computer Systems Department, Jožef Stefan Institute.

Frank Hutter

Matevž Ogrinc

Research assistant

Computer Systems Department, Jožef Stefan Institute.