Shared publicly  - 
 
Follow-up of +Danny Ayers post
https://plus.google.com/u/0/114406186864069390644/posts/dxqRgTUUM4q, here is a proposal for a star rating for linked data vocabularies.

★ Publish your vocabulary on the Web at a stable URI
★★ Provide human-readable documentation and basic metadata such as creator, publisher, date of creation, last modification, version number
★★★ Provide labels and descriptions, if possible in several languages, to make your vocabulary usable in multiple linguistic scopes
★★★★ Make your vocabulary available via its namespace URI, both as a formal file and human-readable documentation, using content negotiation
★★★★★ Link to other vocabularies by re-using elements rather than re-inventing

More on the blog post on why your ★★★★★ data need ★★★★★ vocabularies.
3
5
Michael Hausenblas's profile photoEgon Willighagen's profile photoKingsley Idehen's profile photoDavy Van Deursen's profile photo
7 comments
 
And even when it becomes 5-star, we also need inclusion of rdfs:isDefinedBy relations as a best practice due to its follow-your-nose prowess re., serendipitous discovery of instance (ABox) linked data, courtesy of vocabulary traversal.

#LinkedData #SemanticWeb
 
+Kingsley Idehen Indeed! To be added somehow in the 4-star requirements. Or in some expanded version. We must keep those as short and simple as possible (but not simpler)
 
I think something like "x stars for vocabularies" would be a good thing and this is heading in the right direction. My comments on this draft:
- One really important thing is missing: "Publish your vocabulary using an open[1] license."
- I have a little problem with the "Link to other vocabularies by re-using elements rather than re-inventing"-point. If I re-use elements from other vocabularies I don't understand why I should list them in my vocab at all because I already link to that vocab just by using the properties URIs and all the description etc. should be done in the original vocab, shouldn't it?

As background for the second point: At lobid.org we will only add newly minted properties/classes to our vocab[2] while trying to re-use existing ones. But we don't add the re-used ones to our vocab but we have a properties.txt[3] [4] for all underlying data sets. Until now, these are not linked to the void descriptions.[5] [6] Does a property for linking a dataset to a list of used properties exist somewhere?

Cheers,
Adrian

[1] http://opendefinition.org/okd/

[2] http://lobid.org/vocab/lobid

[3] http://lobid.org/dataset/resource/properties.txt

[4] http://lobid.org/dataset/organisation/properties.txt

[5] lobid.org/dataset/resource/about.ttl

[6] lobid.org/dataset/organisation/about.ttl
 
+Adrian Pohl
- OK with your first point about licensing, to add to the one-star requirement.
- Regarding re-use there is a difference between use by a dataset and re-use in another vocabulary. Your lobid.org approach is perfectly OK to me as long as the few properties you define rely on existing classes, e.g. foaf:Person and foaf:Image for http://lobid.org/vocab/lobid#contactqr.
For specification of properties used by a dataset, you can use void http://rdfs.org/ns/void#propertyPartition and http://rdfs.org/ns/void#property to indicate the property that is the predicate of all triples in a property-based partition. I don't think you need to specify otherwise the partition datasets, just using them as an anchor for property used. But if you can attach to each of those partition dataset e.g., a http://rdfs.org/ns/void#triples property, you indirectly provide information about the usage stats of each property
Add a comment...