Ontology Recommender & Linked Data Extractor (Software Fair)

Немецкий. Commodities traded in global markets are a result of a complex interaction between different industries, processes, and sectors. Ranging from the extraction and harvest of rawmaterials to sophisticated processing networks of goods. This results in various data formats and terminologies being used along the commodity chain. Thus, makes it challenging to warrant the interoperability and accessibility of commodity trade data. One solution could be the resource description framework (RDF) which translates different data sources into linked data. However, the process of generating linked data especially from unstructured data is labour intensive. Therefore, we present our Software Tools as a Proof of Concept for creating linked data from unstructured sources such as trade related documents or regulatory texts. Our services, the Ontology Recommender and the Linked Data (LD) Extractor, integrate Large Language Models (LLMS) and ontology lookup services in a Human in the Loop Process. This process involves extracting named entities, matching the entities with external ontologies and structuring them systematically as RDF. Thereby, our services scan effectively unstructured texts, organizing the information according to ontologies specified in a YAML file. This file acts as a blueprint, outlining the entities and relationships that are transformed into RDF triples. Once the files are transformed, they can be uploaded in a graph data base to retrieve relationships across the different documents and data. Our services address the challenges of data management along with data interoperability and compliance. This is essential for organizations looking to enhance operational efficiency and meet regulatory standards.