Re: Thing Description for existing data sources

> On Jun 10, 2016, at 10:34 AM, Kovatsch, Matthias <matthias.kovatsch@siemens.com> wrote:
> 
> The schema only defines the structure. Somehow we need to attach the semantic annotations directly in the schema. I think schema.org <http://schema.org/> goes in this direction, but has the other issues you identified.
> 
[ Matthias] I think we need something completely new. It would help to properly collect the requirements, identify the closest matches and elaborate from there.
------------------------

What is wrong with using web linking techniques (hyperlinks) with link relations and link attributes to carry the application semantics? Thing Description already has the ability to include more RDF property types to indicate these relations/attributes and they can be resolved to full descriptive URIs using json-ld.

Rather than try to attach application semantics to schemas, I suggest we go the other way around, and attach schemas to semantic descriptions.

I'm not sure why the payload format and application semantics must use the same description mechanisms at all. There may be good reasons to use one method (link relations and attributes) to carry application semantics and another method (templates, json-schema) for describing payload structure and data type (number, string, boolean, time, etc.)

As I recall, the issue with schema.org <http://schema.org/> was in it's lack of ability to describe the data types we think we need. There is nothing stopping schema.org <http://schema.org/> from being easily extended with property types that describe payloads. Schema.org <http://schema.org/> is only the semantic ontology; it can describe anything we want, and can be used to describe payloads by linking to json-schema instances ot other template formats.

I am working with schema.org <http://schema.org/> and some other organizations on use cases and strawman designs for bringing schema.org <http://schema.org/> to IoT/WoT including any new requirements like payload schemas and type descriptions. There will be some examples from this activity to review and evaluate soon.

I agree that we need to have the requirements and use case clearly defined before we evaluate solutions.

Best regards,

Michael