The geospatial knowledge graph: from traditional UML defined datasets to Linked Data

An abundance of geospatial datasets is available. However, most are not yet accessible via the semantic web. The geospatial knowledge graph could be a linked network of the available geo-information from these datasets. In the Netherlands, many of these datasets conform to the NEN 3610 standard, containing the base model for geo-information. This standard uses UML for the visual presentation of the data model and XML, derived by an MDA driven process, for the exchange of data. In a combined effort by a number of organizations in the Netherlands, we have formulated the upper ontology for geospatial features, that will actually be part of the NEN3610 standard. As such, it should become the upper ontology for Linked Geospatial datasets in the Netherlands. As part of this effort, general transformation rules have been defined to transform a UML model to its RDF counterpart, using RDF, SKOS, OWL and SHACL, taking into account existing related efforts on this subject. With these basic building-blocks, a geospatial knowledge graph was created to show the added value of linked geodata, while simultaneously demonstrating that it is feasible to reuse the large body of geospatial datasets to create data-on-the-web.