I have been looking at ways to improve our releases, so that it is clear just what state the vocabulary was in at each point in time. There are some N-Quad files under data/releases/ and a utility in scripts/data/rdfa2nq which will emit quads given one of our RDFa files.

To validate this exploration I have imported an aggregate of the v2.1 release into a SPARQL database at Dydra.com:

http://dydra.com/danbri/schema-quads/@query#graphs-and-terms

In this experimental representation,

• triples in the core get a quad named graph URI based on a release, e.g. http://schema.org/versions/2.1/
• triples from extension files (are aggregated then) get a quad like http://auto.schema.org/#v2.1
• we continue the practice from v2.0 and v2.1 of not currently including extension triples in a the 'schema-all.nt' per-release dumps.
• we don't know yet whether hosted extensions will be on a release/update/edit cycle coupled to the core, or independent.
• whether, where or how to link these from the public site. access is via Github raw links currently.

Links:

Github view of a single release dump file:

• https://raw.githubusercontent.com/schemaorg/schemaorg/master/data/releases/2.1/all-layers.nq