<p>The European Plate Observing System (EPOS, www.epos-ip.org) is a multidisciplinary pan-European research infrastructure for solid Earth science. It integrates a series of domain-specific service hubs such as the Geological Information and Modelling Technical Core Service (TCS GIM) dedicated to access data, data products and services on European boreholes, geological and geohazards maps, mineral resources as well as a catalogue of 3D models. These are hosted by European Geological Surveys and national research organisations.</p><p>Even though interoperability implementation frameworks are well described and used (ISO, OGC, IUGS/CGI, INSPIRE …), it proved to be difficult for several data providers to deploy in the first place the required OGC services supporting the full semantic definition (OGC Complex Feature) to discover and view millions of geological entities. Instead, data are collected and exposed using a simpler yet standardised description (GeoSciML Lite & EarthResourceML Lite). Subsequently, the more complex data flows are deployed with the corresponding semantics.</p><p>This approach was applied to design and implement the European Borehole Index and associated web services (View-WMS and Discovery-WFS) and extended to 3D Models. TCS GIM exposes to EPOS Central Integrated Core Services infrastructure a metadata catalogue service, a series of “index services”, a codeList registry and a Linked Data resolver. These allow EPOS end users to search and locate boreholes, geological maps and features, 3D models, etc., based on the information held by the index services.</p><p>In addition to these services, TCS GIM focussed particularly on sharing European geological data using the Linked Data approach. Each instance is associated with a URI and points to other information resources also using URIs. The Linked Data principles ensure the best semantic description (e.g. URIs to shared codeList registries entries) and also enrich an initial “information seed” (e.g. a set of Borehole entries matching a search) with more contents (e.g. URIs to more Features or a more complex description). As a result, this pattern including Simple Feature and Linked Data has a positive effect on the IT architecture: interoperable services are simpler and faster to deploy and there is no need to harvest a full OGC Complex Feature dataset. This architecture is also more scalable and sustainable.</p><p>The European Geological Services codeList registries have been enriched with new vocabularies as part of the European Geoscience Registry. In compliance with the relevant European INSPIRE rules, this registry is now part of the INPIRE Register Federation, the central access point to the repository for vocabulary and resources. European Geoscience Registry is available for reuse and extension by other geoscientific projects.</p><p>During the EPOS project, this approach has been developed and implemented for the Borehole and Model data services. TCS GIM team provided feedback on INSPIRE through the Earth Science Cluster, contributed to the creation of the OGC GeoScience Domain Working Group in 2017, the launch of the OGC Borehole Interoperability Experiment in 2018, and proposed evolutions to the OGC GeoSciML and IUGS/CGI EarthResourceML standards.</p>

The resources published on the Web of data are often described by spatial references such as coordinates. The common data linking approaches are mainly based on the hypothesis that spatially close resources are more likely to represent the same thing. However, this assumption is valid only when the spatial references that are compared have been produced with the same positional accuracy, and when they actually represent the same spatial characteristic of the resources captured in an unambiguous way. Otherwise, spatial distance-based matching algorithms may produce erroneous links. In this article, we first suggest to formalize and acquire the knowledge about the spatial references, namely their positional accuracy, their geometric modeling, their level of detail, and the vagueness of the spatial entities they represent. We then propose an interlinking approach that dynamically adapts the way spatial references are compared, based on this knowledge.

<p>In geosciences, where nomenclature naturally has grown from regional approaches with limited cross-border harmonization, descriptive texts are often used for coding data whose meanings in the international context are not conclusively clarified. This leads to difficulties when cross border datasets are compiled. On one hand, this is caused by the national-language, regional and historical descriptions in geological map legends. On the other hand, it is related to the interdisciplinary orientation of the geosciences e.g. when concepts adopted from different areas have a different meaning. A consistent use and interpretation of data to international standards creates the potential for semantic interoperability. Datasets then fit into international data infrastructures. But what if the interpretation to international standards is not possible, because there is none, or existing standards are not applicable? Then efforts can be made to create machine-readable data using knowledge representations based on Semantic Web and Linked Data principles.</p><p>With making concepts reference able via uniform identifiers (HTTP URIs) and crosslinking them to other resources published in the web, Linked Data offers the necessary context for clarification of the meaning of concepts. This modern technology and approach ideally complements the mainstream GIS (Geographic Information System) and relational database technologies in making data findable and semantic interoperable.</p><p>GeoERA project (Establishing the European Geological Surveys Research Area to deliver a Geological Service for Europe, https://geoera.eu/) therefore provides the opportunity to clarify expert knowledge and terminology in the form of project specific vocabulary concepts on a scientific level and to use them in datasets to code data. At the same time, parts of this vocabulary might be later included in international standards (e.g. INSPIRE or GeoSciML), if desired. So called “GeoERA Project Vocabularies” are open collections of knowledge that, for example, may also contain deprecated, historical or only regionally relevant terms. In an ideal overall view, the sum of all vocabularies results in a knowledge database of bibliographically referenced terms that have been developed through scientific projects. Due to the consistent application of the data standards of Semantic Web and Linked Data nothing stands in the way of further use by modern technologies such as AI.</p><p>Project Vocabularies also could build an initial part of a future EGDI (European Geological Data Infrastructure, http://www.europe-geology.eu/) knowledge graph. They are restricted to linguistic labeled concepts, described in SKOS (Simple Knowledge Organization System) plus metadata properties with focus on scientific reusability.  In order to extend this knowledge graph, additionally they also could be supplemented by RDF data files to support project related applications and functionality.</p>

Assessing the quality of the main linked data sources on the Web like DBpedia or Yago is an important research topic. The existing approaches for quality assessment mostly focus on determining whether data sources are compliant with Web of data best practices or on their completeness, semantic accuracy, consistency, relevancy or trustworthiness. In this article, we aim at assessing the accuracy of a particular type of information often associated with Web of data resources: direct spatial references. We present the approaches currently used for assessing the planimetric accuracy of geographic databases. We explain why they cannot be directly applied to the resources of the Web of data. Eventually, we propose an approach for assessing the planimetric accuracy of DBpedia resources, adapted to the open nature of this knowledge base.

The OGC Environmental Linked Feature Interoperability Experiment (ELFIE) sought to assess a suite of pre-existing OGC and W3C standards with a view to identifying best practice for exposing cross-domain links between environmental features and observations. Environmental domain models concerning landscape interactions with the hydrologic cycle served as the basis for this study, whilst offering a meaningful constraint on its scope. JSON-LD was selected for serialization; this combines the power of linked data with intuitive encoding. Vocabularies were utilized for the provision of the JSON-LD contexts; these ranged from common vocabularies such as schema.org to semantic representations of OGC/ISO observational standards to domain-specific feature models synonymous with the hydrological and geological domains. Exemplary data for the selected use cases was provided by participants and shared in static form via a GitHub repository. User applications were created to assess the validity of the proposed approach as it pertained to real-world situations. This process resulted in the identification of issues whose resolution is a prerequisite for wide-scale deployment and best practice definition. Addressing these issues will be the focus of future OGC Interoperability Experiments.