EFRA - incidents annotator gold standard

The dataset is a subset of 50 records extracted from the Food Safety Incidents (AFSI) dataset: https://zenodo.org/records/10891602; the records have been manually annotated with respect to the EFRA conceptual model.The EFRA conceptual model integrates several public ontologies to ensure interoperability and semantic consistency across food safety data sources: EFRA Food Safety Hazard - developed in the first year of the project and formalized using a SKOS model, provides a formal categorization of hazards. AGROVOC – A multilingual thesaurus developed by the FAO, covering food, agriculture, fisheries, forestry, and related domains. It contains over 32,000 OWL-based concepts available in up to 29 languages and supports data annotation, retrieval, and integration through standard semantic web formats (RDF, XML, JSON). FoodOn – An open-source ontology describing food sources, ingredients, processing, packaging, and product classification, enabling structured representation of food items and production processes. ChEBI (Chemical Entities of Biological Interest) – An ontology for representing molecular entities, particularly small chemical compounds relevant to biological and biochemical contexts. GS1 GPC (Global Product Classification) – A standardized global product taxonomy used for trade and product identification, organized in a four-level hierarchy (Segment, Family, Class, Brick) with optional attributes for finer classification. These ontologies constitute the knowledge base used by the LLM agent for mapping the extracted information to known concepts. The manually annotated gold standard and the dump of the KB are included in the dataset.Incidents Annotator Gold Standard: annotates the report with respect to the EFRA Food Safety Hazard and GS1 GPC (Global Product Classification)Incidents Annotator - Related Concepts Gold Standard: annotates the report with respect to AGROVOC, FoodON and ChEBI