approach:
- use python rdflib to perform a SPARQL query on the SKOS vocabulary and transform the relevant info to the datahub business glossary yaml format
- use datahub ingest to load business glossary yaml file
alternative approach would be to integrate rdflib python code directly with the datahub python emitter.
issues:
- source url, source_ref currently not visible in UI. Already reported. work around defining source_url as custom property
first step: see jupyter notebook
second step:
datahhub ingest -c business_glossary_to_datahub.yml
result:
Potential improvements:
- still a bit 'hacky' with hardcoded splits and mapping to a specific glossary...
- have the skos relations (to national vocabulary) in the documentation
- also import documentation from national vocabulary including images etc.
have a concept browser, included in the business glossary so you can visually browse through related concepts