Research project Extreme Food Risk Analytics (EFRA)
The aim of this EU project is to increase food safety for citizens. Today there is a wide array of data sources holding crucial information about the food that we eat. The problem is that these sources are heterogeneous – and sometimes hidden. We explore how data can be mined, aggregated and analysed using AI.
The food supply chains are impacted by a web of drivers (economic, socio-economic, environmental and climate) that may pose direct or indirect development of food safety risks at a short- or long-term timeframe. To be able to anticipate future food safety risks a holistic or systems approach should be used that takes into account all of these factors.
Useful information for the drivers may appear in disparate, heterogeneous sources (both public and private), often unstructured and in multiple languages. For example, in world-wide public food safety authority web sites (for example recall or border rejection announcements, inspection results, lab test results), in expert content pieces (for example scientific publications, news items, opinion pieces), in consumer discussions (for example product reviews, consumer discussion boards, social media), and in economic/governmental data (for example food trade volumes, price fluctuations, country risk and corruption indicators). Finally, many useful food safety data are “hidden” in sources that are hard to discover and mine, for example in websites or databases at the municipality/local authority level.
Not taking full advantage of this wealth of public and private data comes at a great cost: Despite best efforts and modern techniques, consumers world-wide still get sick from contaminated foodstuff and food companies suffer huge economic and legal penalties from product food recalls.
EFRA’s core ambition is to overcome these boundaries by exploring novel, experimental and promising approaches in extreme data mining, aggregation, and analytics technologies. By recognizing and collecting this wealth of heterogeneous data, scattered throughout the internet, and convert it into a “universal language” of high-quality risk food data, EFRA will be able to train AI-models to proactively provide risk mitigation measures (based on predictive awareness of short- and long-term risks) with an explainable, secure, sensitive, accurate, trustworthy, fair and green manner, before occurrence of potential food safety hazard.
Project description
Through digitization and developments in sensor networks and Internet of Things (IoT) connectivity, the collection of data along the food supply chain has increased to an enormous scale. EFRA, through extreme data mining, aggregation and analytics, aspires to develop the first analytics-enabled, secure-by-design, green data space for AI-enabled food risk prevention with a mission to support EU’s global leadership in the digital-led industry transition from reaction to food risk prevention.
This will be done using three specific data utilisation Use Cases representing the very important sectoral and societal AI usage purposes in Europe. EFRA will also produce several key exploitable results (KERs) that will offer extensive positive impact. We will design, test, and deploy tools and undertake appropriate initiatives to facilitate their uptake, elicit feedback, and engage stakeholders.
The EFRA tools are:
– EFRA Data Hub, offering intelligent crawlers and data annotation and linking modules to search, mine, process, annotate, and link dispersed, multilingual, heterogeneous, and deep/hidden food safety data sources.
– EFRA Analytics Powerhouse: offering modules running over a green cloud HPC that distill useful insights and signals from the EFRA Data Hub to train privacy-preserving, explainable, green food risk prediction AI models.
– EFRA Data & Analytics Marketplace: A front-facing user-friendly web app that allows interested users to discover, purchase/use, and contribute data, AI models, and analytics modules, creating an economy where data holders and data consumers engage and trade.
The EFRA consortium consists of nine partners coming from seven European countries (Croatia, Greece, Italy, Netherlands, UK, Romania, and Sweden). The group has three research technology organizations and universities (Stockholm University is one of them), five SMEs and one certification body, capable of achieving the demanding project goals.
Together all partners form a complete group uniting the necessary expertise, skills, interdisciplinary knowledge, and resources that constitutes a representative value chain of actors.
The collaborating partners in this project are:
Wageningen University and Research (Wageningen, The Netherlands)
More information is available at the EFRA website
Read about SemEval 2025 Task 9: Advancing Food Hazard Detection
Project members
Project managers
Aron Henriksson
Associate professor
Tony Lindgren
Unit head SAS
Members
Korbinian Robert Randl
PhD student
Ioannis Pavlopoulos
Affiliated researcher
