At the Norwegian Maritime Authority (NMA) there was a need for replacing their legacy supervision system. The purpose of the new solution is to facilitate interaction between all stakeholders that are involved in ensuring that a vessel is compliant with relevant regulations. The new implementation should rely on machine readable, linked data, reflecting current and historic regulations. Including tools for enrichment of additional sources and facilitate data maintenance for domain experts.||||This presentation will walk you through our initial challenges, and the proposal of a hypothesis of automatic knowledge graph generation from regulations. Our journey through a Proof of Concept, leading up to our pipeline running in production on Azure. Exploring the synergies of Natural Language Processing (NLP) and Semantic Technologies. Including tooling for graph enrichment with Reasonable Ontology Templates, and why we chose the Shapes Constraint Language as our data model for the domain of maritime law. The presentation will also cover tooling for data maintenance performed by domain experts at the NMA. We will discuss lessons learned, benefits, and values from our experiences at the Norwegian Maritime Authority with combining NLP and RDF.||
