4 guides
Linked data
The standards that give public data a shared, machine-readable meaning, from small vocabularies and code lists to catalogues of every dataset a country publishes.
The guides
Ontologies and vocabularies
What vocabularies, ontologies and code lists are, and how the EU Core Vocabularies give public bodies shared meaning.
02Linked open government data
Linked data principles, RDF and persistent URIs, plus the EU Open Data Directive and high-value datasets.
03DCAT-AP and data catalogues
How DCAT and DCAT-AP describe public datasets so that catalogues across Europe can be searched together.
04Semantic web services
The 2000s vision of self-assembling online services, why it faded and which of its ideas public bodies still use.
What linked data adds
Most government data sits in tables designed for one system. Linked data takes a different approach: every thing gets a stable identifier on the web, and facts are recorded as simple statements about those things. A dataset published this way can be joined with others without a bespoke integration project each time.
The technology has been around for two decades and its record is mixed. It has become the backbone of European data catalogues and shared vocabularies, while many day-to-day exchanges still run on spreadsheets and APIs. The guides here are honest about both.
Meaning first, format second
The most durable lesson is that agreeing what data means matters more than the format it travels in. A short, well-governed vocabulary that several agencies actually use does more good than an elaborate model that nobody adopts.
How these guides fit together
Begin with ontologies and vocabularies, then see how they are used to publish open data and describe it in catalogues. The last guide looks back at semantic web services, an ambitious idea from the 2000s, and at what survived. For the policy context, see interoperability.
Articles on linked data
Shorter, practical pieces that build on these guides.
Why “address” is the hardest word in government data
An address field can hold four different ideas. Why that breaks matching, and the three fixes that work.
Read the articleFive questions to ask before you build a new public register
Whether a new register becomes a trusted source or another stale copy is decided by five early questions.
Read the articleCSV, API or linked data? Choosing how to publish public data
When a CSV file is enough, when an API earns its keep, and when linked data is worth the extra effort.
Read the articleA dataset record, before and after: fixing public data metadata
One weak dataset record, improved field by field, to show what makes public data easy to find.
Read the articleCode lists: the unglamorous fix for messy public data
Why agreed lists of values beat clever matching, and how to reuse, publish and maintain them.
Read the article