The Knowledge Graph Conference Icon
The Knowledge Graph Conference
  • 馃彔Home
  • 馃搮Events
  • 馃懁Members
  • 馃數Announcements
  • 馃數Ask
  • 馃數Ask The Ontologists
  • 馃數Events
  • 馃數Jobs
  • 馃數Promotions
  • 馃數Share
Powered by Tightknit
Ask
Ask

Combining JSON and JSON-LD for Knowledge Graph Integration

Avatar of Einar C.Einar C.
路May 06, 2022 11:03 AM

Hi. In the world of developers JSON is a prominent member. For Linked Data space JSON-LD is important. Has anyone in this community done work to combine the two? The idea is to add LD part into the tools used by developers. To make their JSON for APIs and messages KG aware. And developers can make the connection to KG concepts, classes, properties as design is done. Thanks.

馃檶1

22 comments

路 Sorted by Oldest
      • Avatar of Ashleigh F.
        Ashleigh F.
        路

        No but for a similar reason we did combine JASON-LD with JATS XML schema

      • Avatar of Andreas E.
        Andreas E.
        路

        Hi Einar, you might find this article interesting. It compares JSON-LD to JSON schema which is an essential part of the REST / OpenAPI world: https://dashjoin.medium.com/json-schema-schema-org-json-ld-whats-the-difference-e30d7315686a

      • Avatar of Einar C.
        Einar C.
        路

        Thanks Ashleigh F.. I see the good need in science writing also. I look into it.

      • Avatar of Einar C.
        Einar C.
        路

        Thanks Andreas E..

      • Avatar of Quentin R.
        Quentin R.
        路

        Einar C. Based on a blog on JSON-LD, we leveraged @context to refine JSON-LD such that it could be more "naturally" be processed through JSON libraries.

      • Avatar of Einar C.
        Einar C.
        路

        Thanks Quentin R.. Interesting path to help developers.

      • Avatar of Andrew P.
        Andrew P.
        路

        ive been thinking the same thought Einar C. .

        馃憤1
      • Avatar of Tomasz P.
        Tomasz P.
        路

        IMO JSON-LD is snake oil because it tricks teams into treating RDF as JSON tree. it is all to easy to put syntax over substance and that is recipe for trouble because any graph can be represented as multiple JSON documents (tree < graph)

      • Avatar of Tomasz P.
        Tomasz P.
        路

        it works for a read-only use case of turning existing JSON into RDF by augmenting it with a @context

      • Avatar of Tomasz P.
        Tomasz P.
        路

        otherwise I stand with what Katariina K. said in her talk. it is crucial to think in graph and, by extensions, also code graph.

        馃憤2
      • Avatar of Tomasz P.
        Tomasz P.
        路

        the only sane way I think is to really adopt graph as the abstract model and implement against it that way. nodes & edges

      • Avatar of Andrew P.
        Andrew P.
        路

        Tomasz P. i think its a good and familiar representation layer in the comtext for use within an application runtime.

      • Avatar of Andrew P.
        Andrew P.
        路

        I was thinking though that shacl might be a good way to define rest interface and nouts/outputs that consume and/or produce data from graph. thoughts?

      • Avatar of Tomasz P.
        Tomasz P.
        路

        馃憤

      • Avatar of Tomasz P.
        Tomasz P.
        路

        I have seen that recently and was definitely planning to implement myself. my ideas include using content negotiation by profile (W3C draft) to allow clients to choose how to represent a resource. for example

        GET /person/tomasz
        Host: example.com
        Accept-Profile: https://example.com/shape/person-minimal

        alternatively, the new QUERY method (IETF draft) could offer a GraphQL-like interface so that any arbitrary shape could be requested

        QUERY /person/tomasz
        
        [
          a sh:NodeShape ;
          ... properties to include in response
        ] .

      馃憤
      1
      馃檶1