The Common Core repo is a collaborative effort to describe and catalogue elements of a linked data approach that facilitate FAIR data principles. At present this is an effort between Ocean InfoHub, POLDER and the NSF DeCODER projects.
The first product being explored is the idea of a catologue of patterns supporting interoperability. This has the name, Pattern Asset Catalogue (PAC).
A simple JSON-LD (ie RDF) profile to describe a data graph that:
- Provides metadata about a SHACL shape (itself an RDF graph)
- related queries (SPARQL)
- Also related to the SPARQL endpoints as services that can do query or update, etc
- related services associated with the above
- related JSON-LD frames
- related release Graphs (link to example for OIH and others)
- potential JSON schema files
- example data graphs
- connects the above to resolvable and established references of associated goals and patterns
The reason?
SHACL, frames and other elements express the structure of a graph. Additionaly they can connect these assets to aspects of FAIR principles or other related principes. This can aid in the discovery and description of the intended goals of these assets.
This is NOT semantic integration. It is not an ontology and does not provide any of the deep semantic interoperability benefits that ontology and semantic integration based on ontologies can provide.
PAC is also NOT a new concept. It is a influenced by. and attempts to borrow (steal) from, things such as OGC STAC but also could be compared to things such as SQL global as view / local as view concepts.
Additionally, there are other projects such as WSMO (https://www.w3.org/Submission/WSMO-primer/) that have the potential to address these goals in a more enterprise approach.
While SHACL AF does allow some use of SPARQL including SELECT and ASK, this is a not more complex as the non-RDF SPARQL is mapped into the SHACL. Also, complex SPARQL seems like it would be at least annoying to do in this approach. For now the model of connecting the SPARQL as an external creative work or other type is proposed.
DCAT can describe catalogs and does contain a Data Service node. However, they focused more on the use of services to access data. So connections to validation or framing or other apps might become a bit forced. It would be worth exploring though.
Given that both schema.org and SHACL build on the common RDF data model it would be just as easy (and one can) integrate all this information into one file. Why have two files holding the information? Part of this is just the simplicity of the abstraction as this concept is judged for value. Blending the two files together in some ways burdens the proven SHACL work with the ShaC hypothesis.
However, while it is true this approach should allow SPARQL queries to be formed that can be federated via SERVICE, see https://www.w3.org/TR/sparql11-federated-query/ for details, this is not the primary goal.
Additional approaches would include being able to form SPARQL CONSTRUCT calls to form new triples or leveraging approach to subset or frame out data from a parent graph into a useful subset of triples for specific communities or use cases.
Rather, the goal here is not to promote any given approach to use the graphs that validate to this common core shape, but simply ensure that whatever approach is developed works against the shared graph shapes.
What are the elements of FAIR we can encode into SHACL? PIDs? Vocabulary? If not SHACL, VOID or some other way to describe? schema.org?
Extend type Org to address FAIR? Build off the RDA working group?
Or SDGs or essential variables