Linked Data validation and transformation

[draggett]

Linked Data can play a role as a lingua franca for data and metadata, as a means to bridge different platforms and standards. Ontology languages like OWL focus on inferencing and not on validation. There is however an increasing need to be able to validate RDF graphs against a set of rules, and a need for rules that transform RDF graphs. One use case is to validate descriptions of things in the Web of things, and to verify that a thing that claims to support a given semantic model actually satisfies the constraints defined by that model. Another is to transform a description of a class of things into a specific thing.

A further criteria is that the rule language should be easy to understand by non-specialists, and lend itself to editing tools, e.g. graphical user interfaces. One promising direction is based upon augmented transition networks (ATNs) which were developed in the late 60's and early 70's for natural language parsing and has a natural representation as diagrams.

[swickr]

What is the WoT perspective on the Shapes Constraint Language?

[pchampin]

I would say that the topic of validation is now well covered by SHACL and ShEx.

Regarding transformation, the N3 community group is active and making encouraging progress.

[pchampin]

What is the WoT perspective on the Shapes Constraint Language?

The interesting thing about the WoT WG is that they did not include SHACL shapes for their Thing Description (TD) vocabulary. Instead, they proposed an informal JSON schema, which makes sense as JSON-LD is the preferred representation of TD, and as JSON schema is quite popular in the JSON space.

But that raises the question of the relationship between JSON schema and W3C (#108) and the relationship between JSON schema and SHACL...