Most of us in the library and cultural heritage communities interested in metadata are well aware of Tim Berners-Lee’s five star ratings for linked open data (in fact, some of us actually have the mug).
The five star rating for LOD, intended to encourage us to follow five basic rules for linked data is useful, but, as we’ve discussed it over the years, a basic question rises up: What good is linked data without (property) vocabularies? Vocabulary manager types like me and my peeps are always thinking like this, and recently we came across solid evidence that we are not alone in the universe.
Check out: “Five Stars of Linked Data Vocabulary Use”, published last year as part of the Semantic Web Journal. The five authors posit that TBL’s five star linked data is just the precondition to what we really need: vocabularies. They point out that the original 5 star rating says nothing about vocabularies, but that Linked Data without vocabularies is not useful at all:
“Just converting a CSV file to a set of RDF triples and linking them to another set of triples does not necessarily make the data more (re)usable to humans or machines.”
Needless to say, we share this viewpoint!
I’m not going to steal their thunder and list here all five star categories–you really should read the article (it’s short), but only note that the lowest level is a zero star rating that covers LD with no vocabularies. The five star rating is reserved for vocabularies that are linked to other vocabularies, which is pretty cool, and not easy to accomplish by the original publisher as a soloist.
These five star ratings are a terrific start to good practices documentation for vocabularies used in LOD, which we’ve had in our minds for some time. Stay tuned.