INSPIRE in RDF

As many people who work in the field will know, the 2007 INSPIRE Directive tasks European Union Member States with harmonizing their spatial and environmental data. The relevant department of the European Commission, the JRC, has lead the definition of a complex data model that is broken down into various themes. Naturally enough, the data is modeled in UML and the implementations are based largely on OGC standards that make use of XML/GML etc. However, a number of projects are experimenting with using the model in Linked Data environments. These include GeoKnow, MELODIES and SmartOpenData (SmOD) in which W3C's European host, ERCIM, is a partner. This project has been instrumental in establishing the Spatial Data on the Web Working Group that is now racing towards the first formal publication of its use cases and requirements document (like most W3C WGs, the document is being developed in full public view on Github).

Like GeoKnow, when SmOD first started to consider using INSPIRE data in RDF we felt duty bound to try and represent the whole of the detailed data model. However, as the project enters its final phase, much of this has been rejected in favor of a simpler approach that is more in line with 'Linked Data thinking' and no longer attempts to recreate the full scope of INSPIRE in RDF. There are two principal motivations for this:

1. Experience: when creating Linked Data for use in a range if pilot projects within SmOD, a slavish following of INSPIRE proved burdensome and unhelpful. The aim of taking a different approach (Linked Data) must be to gain some benefit from that approach not available from the original (XML/GML), recognizing that the original will offer features not available in the derived work.
2. The publication of the Study on RDF and PIDs for INSPIRE by Diederik Tirry and Danny Vandenbroucke under ISA Action 1.17: A Reusable INSPIRE Reference Platform (ARE3NA). This report summarized work by three experts: Clemens Portele, Linda van den Brink and Stuart Williams. The summary proved extremely useful to the current project partners. All documents from that work are available and remain marked as 'for review' although a conversation with JRC staff suggests that no further work is foreseen on these documents.

One call to action from the ARE3NA work was that the INSPIRE Registry be extended to include SKOS concept schemes in addition to the formats already offered. This has been done and allows SmOD to use the registry's persistent URIs as identifiers for many of the concepts that are important in the current work.

It is this combination of factors that is behind the final model being at once simpler and much more comprehensive than the initial one in its coverage of the INSPIRE themes. For example, the three classes originally associated specifically with representing Geographical Names have disappeared altogether to be replaced by rdfs:label!

The work is fully available via the most stable namespace available to the project, namely http://www.w3.org/2015/03/inspire/.

Only the INSPIRE themes relevant to SmOD have been modeled in RDF as part of this work and, even within those themes, only the classes and properties needed in the project have been defined. Therefore, the 'SmOD vocabularies' should be seen only as a beginning.

Can they be added to?

Certainly. Indeed, that's the express hope.

How?

Ideally, the JRC itself will publish RDF vocabularies that mirror the INSPIRE model. In that eventuality, the ones on w3.org should almost certainly be deprecated. However, until that happens, the best vehicle W3C has for gathering people together with a common interest is the Community Group system. This is open to W3C Members and non-members alike and a Community Group could act as the forum for discussion of INSPIRE in RDF with the ability to add new terms, clarify existing ones, add new translations and, if needed, deprecate old terms.

I already have some expressions of interest in this but would like to gather more before proposing the CG be formed. If this interests you, please get in touch.

Before closing, I want to thank colleagues in the SmartOpenData project, notably Tatiana Tarasova and Jindřich Mynarz, for their help, advice and expertise.