Can we create Virtual Graphs of other Virtual Graphs - potentially where referenced Virtual Graphs are remote to the core, central graph - ie a graph of graphs?
Pretty sure that the answer is "yes", for example, you can virtualise Stardog endpoints which have their own virtual graphs not explicitly known to the central graph. But it'd help if you expand on what specifically you're trying to achieve.
Example (but only for understanding purpose - may not be true definition of use case).
We have a series of businesses that all deal in "Clients". Each business wants to run and manage their own instance of stardog - for use as an internal capability and to fulfil the below - and use the virtual graph capabilities to map Client to internal data in various systems. The businesses all want to share client information, so they are wanting to create a central stardog instance that virtualises and aggregates each of the individual businesses Client into a super graph of Client (now obviously Client isnt the only class, but for example purposes it should suffice). Now, they do not want to copy the Client details centrally - each business wants to return "ownership" of its clients data.
Is the above scenario possible - and how would you go about it with Stardog if it could.
Ok. Think I answered my own question. Each of the other "remote" Stardog virtual graphs become Sparql Data Sources within the central instance - have to say that the documentation is a bit ambiguous when it comes to the data source configurations - especially for non-JDBC data sources.
Yes, you're right. You can create a VG out of a Stardog-backed SPARQL endpoint . You can have multiple VGs like that in the central instance and run aggregate queries over them.
Agreed that it's not super easy to find this part in the docs. Thanks, we'll look into that.