Semantic Tools for Making Food Safety Research FAIR and Future-Proof - Interoperable Terminology Services and Ontologies for Food Safety re-used in FAIRagro

The German Federal Institute for Risk Assessment (BfR) is responsible for assessing the safety of food and feed, which requires consideration of both microbial and chemical hazards. Within the Risk Assessment Knowledge Integration Platform (RAKIP), the BfR developed the Food Safety Knowledge Exchange (FSKX) format, a standardized model exchange format applicable in domains such as food safety, life sciences, One Health, and artificial intelligence [1]. To support BfR's mission, a centralized knowledge base on food and feed processing is desirable. Such a resource would integrate information on process flows, production parameters, measurement data, and consumer food handling. As a prerequisite for building this semantically integrated resource, we identified and extended domain-relevant terminology resources and developed supportive tools and services to facilitate more precise and semantically enriched descriptions of food and feed processing steps. One key outcome is a first version of a feed and food safety terminology service hosting a broad collection of ontologies relevant to food, feed, and their processing. The service is accessible via a web interface and an API, and information about contained terms can be embedded in other systems using reusable widgets. Building on these efforts, the results are being further developed and integrated into the FAIRagro terminology service. It is designed to grow by incorporating additional ontologies relevant to the FAIRagro domain. The widgets have also been further developed within the NFDI-wide basic service Terminology Services 4 NFDI, increasing their reach and reusability. This integration ensures long-term availability of the service for the federal research institutions and creates a direct link between public sector research and the NFDI consortia, strengthening cross-institutional collaboration. Another key result is the development of workflows for semantic mapping between controlled vocabularies. A notable example is the automated and manually curated mapping [2] between the FoodOn ontology [3] and EFSA's FoodEx2 classification [4], encoded using the SSSOM format [5] to ensure interoperability. This mapping comprises over 2,400 high-confidence links, offering a bridge between internationally recognized food classification systems and supporting integration into platforms such as SemLookP [6] and the EMBL-EBI OxO service [7]. To support the semantic transformation of existing vocabularies further, the Ontology CV Integrator was developed – an open-source Python-based tool chain using the ROBOT toolkit [8]. This tool has been applied to convert the BfR's FSKX metadata schema into a formal ontology. Following the creation of the FSKX ontology, real-world FSKX JSON metadata was converted into machine-readable Knowledge Graphs using a JSON-LD-based method. A total of 174 metadata fields were semantically grounded using the FSKX ontology and other sources. This transformation preserved the original JSON structure while enabling integration into Linked Data infrastructures. The resulting scripts and context files are openly available and have been successfully applied to real datasets. FAIRagro will build on these achievements by expanding terminology services, integrating further agrosystem-relevant vocabularies, and reusing tools such as the Ontology CV Integrator and mapping workflows. This supports scalable, consistent development of semantic resources and fosters the broader goals of the NFDI in terms of interoperability, reusability, and knowledge standardization in agricultural and food research.