AI tool on automatic data extraction from RASFF -SA10

11/2022-12/2024

Funding programme / funding institution: Europäische Behörde für Lebensmittelsicherheit (EFSA) - Italien

Grant number: GP/EFSA/AMU/2020/02-SA10

Project homepage: https://zenodo.org/record/4322555

Project description:

Faster trade in feed and food and their more complex and globalized supply chains pose new challenges for consumer health protection. To meet these, efficient traceability of feed and food is needed, enabled by data exchange and processing and powerful, interoperable software tools.

The software tool "Rapid Alert Supply Network Extractor (RASNEX) is used for the automated extraction of data from notifications of the European Rapid Alert System for Feed and Food (RASFF). After a feasibility study on the applicability of Artificial Intelligence was carried out within the FRAMEWORK PARTNERSHIP AGREEMENT No. GP/EFSA/AMU/2020/02 Specific Agreements No. 7, RASNEX will be transferred to the current version RASNEX 3.0 within this project. This will provide a web application with an interface that allows member states to upload RASFF notifications, automatically extract relevant feed and food traceability data, such as different operators of the supply chains and their addresses, manually adjust extracted data, and then download it in a universal traceability data exchange (UTX) format. Neuro-Linguistic Programming (NLP) and Named-Entity Recognition (NER) models are applied for this purpose.

The existing RASNEX 1.0 application is to be converted to a suitable programming language that meets the requirements of the EFSA environment rules (e.g., Python or R). The project also plans to conduct several case studies (BfR tools, EFSA cloud technical requirements, data collection tool in the R4EU environment).

The continuous exchange with stakeholders and users is intended to transfer RASNEX into the application phase and to ensure a demand-oriented further development.