Skip to content

Latest commit

 

History

History
142 lines (92 loc) · 2.76 KB

PITCHME.md

File metadata and controls

142 lines (92 loc) · 2.76 KB
Raw

Vocab.belgif.be

Publishing re-usable thesauri as Linked Open Data

Vocab server

  • Publishing XSD schema's
  • Publishing thesauri, controlled lists

Thesauri

  • Searchable HTML view
  • SKOS in multiple download formats
    • JSON-LD, TTL...
  • Linked Data Fragments API for machines

What is SKOS ?

+++

Features

  • Terms can have labels in multiple languages
  • Preferred, alternative labels
  • Hierarchy: broader, narrower terms
  • Mapping =/= taxonomies: close, exact match

+++

Example

@prefix skos: <http://www.w3.org/2004/02/skos/core#> .

<http://vocab.belgif.be/auth/mybe-theme/FINA#id> a skos:Concept;
    skos:notation "FINA" ;
    skos:prefLabel "Financiën"@nl, "Finance"@fr ;
    skos:broadMatch <http://publications.europa.eu/resource/authority/data-theme/ECON> .

+++

Benefits

  • Simple to create and use
  • Used by EU Publication Office, UN FAO, ...
  • Possible to link to other thesauri

Already published

  • NACEbel economic activities
  • Territories recognized by Belgium
    • linked to GeoNames, EU country codes
  • NIS regions and municipalities
    • linked to GeoNames
  • ...

Adding a new thesauri

  • Create a new SKOS file
    • dcterms:title, dcterms:description (NL-FR-DE-EN)
  • Add it to the github repository
  • Build and deploy new instance
    • Files will be indexed and converted on start-up

Creating SKOS files

+++

Why conversion

  • Taxonomies often maintained in other systems
  • No need for yet another input tool

Technology

  • DropWizard server
  • Embedded RDF4J triple store / Lucene
  • Deployed as 1 docker container
    • Read-only, scalable

+++

What is DropWizard ?

  • Open source Java REST framework by Yammer
    • Jersey annotations, Freemarker templates
  • Stand-alone micro-services
    • Similar to Spring Boot etc

+++

What is RDF4J ?

Linked Data Fragments API

  • Triple Pattern Fragments "web-scale" queries
  • Client-side SPARQL queries
  • Paginated results
  • Very basic queries on the server
    • Less server load, more network traffic

Thank you

Questions ?

@fa[twitter] @BartHanssens

@fa[envelope] opendata@belgium.be