Re: [core] SenML and link-format in RDF (was: Re: Designs to resolve streaming issues in SenML)

[Michael Koster <[e-mail]>]

> On Jan 26, 2016, at 11:40 PM, Christian Amsüss <[e-mail]> wrote:
> 
> Hi Michael,
> 
> On Tue, Jan 26, 2016 at 08:28:31PM +0700, Michael Koster wrote:
>> {
>>    "@context" : "http://thingschema.org",
>>    "bn":"/3303/1/",
>>    "i":[
>>    {"n":"5700","v":"31.3","u":"Cel"}, ...
>>    ],
>>    "l":[
>>    {"href":"","rel":"self","rt":"temperature","u":"Cel"},
>>    ...
> 
> thanks for the example, that's a good starting point to work with.
> 
>> And here are semantic triples generated from the SenML, having used
>> http://json-ld.org/playground/ <http://json-ld.org/playground/> to
>> process the elements into RDF subjects, properties, and values and
>> serialize into N-quads.
>> 
>> _:b0 <http://thingschema.org/schema#hasItem> _:b1 .
>> _:b0 <http://thingschema.org/schema#baseName> "/3303/1/" .
>> _:b1 <http://thingschema.org/schema#resourceName> "5700" .
>> _:b1 <http://thingschema.org/schema#unitsOfMeasure> "Cel" .
>> _:b1 <http://thingschema.org/schema#value> "31.3" .
>> _:b10 <http://thingschema.org/schema#href> "5700" .
>> _:b10 <http://thingschema.org/schema#resourceType> "currentValue" .
>> _:b10 <http://thingschema.org/schema#unitsOfMeasure> "Cel" .
> 
> These are quads that contain the information of the SenML, but not
> really in an RDFish way.
> 
> RDF is all about using URIs to describe things. What I'd expect of such
> a conversion when done right would be N-triples like
> 
> _:b1 <http://thingschema.org/schema#about> </3303/1/5700> .
> _:b1 <http://thingschema.org/schema#value> "31.3"^^xsd:float .
> _:b1 <http://thingschema.org/schema#unitsOfMeasure> "Cel" .
> </3303/1/5700> <http://thingschema.org/schema#resourceType> "currentValue" .
> </3303/1/5700> <http://thingschema.org/schema#unitsOfMeasure> "Cel" .
> 
Hmm so I can't know from the schema that baseName and href are concatenated to form /3303/1/5700?
That seems to be all that's missing.

> where statements are made about actual resources. _b:1 would still be a
> blank node, because it reflects a single description of the resource's
> state (were it a time series, the necessity for that would be more
> obvious). A converter might want to even emit a statement like
> 
> _:b1 <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://thingschema.org/schema#Item> .
> 
> to make that clear.
> 
> The output produced by the thingschema.org still reads a little like a
> combination of text fragments; to compare it with relational databases,
> it looks to me as if there were no foreign keys in use yet, and
> everything were stored in text fields. The output is RDF, but it can't
> effectively be used as RDF yet.
> 
>> I think this result is quite useful to be able to use the information
>> in the context of fully qualified references and nodes of triples that
>> describe the original SenML.
> 
> I do not understand what you mean by this, could you elaborate? AFAICT,
> were I fed the N-quad output from above, I couldn't even establish any
> relationship between the "/3303/1/" URI fragment and the "5700", let
> alone use that in a query about </3303/1/5700>.
> 
> 
> There are things brewing at W3C's Web of Things initiative[1], there has
> for some time been work on semantically connected IoT devices[2], and I
> am confident that RDF metadata are the way to go.
> 
> But both with a SenML that looks anything like the drafts so far and
> with draft-ietf-core-links-json, I think that JSON-LD does not have what
> it takes to produce the meaning of the respective JSON representations
> for the semantic web.
> 
> I've long ago asked on this list about a canonical RDF representation of
> a link-format resource. Back then, there was nothing, and I had only a
> rough idea of what I'd need as well. I'd be all in for going ahead here
> and sketching up what a generic link-format document would look like in
> an RDF representation, and possibly SenML as well; that should be a
> separate discussion, though, and not deeply interfere with the SenML
> process itself. I could sketch up some parts that've been brewing with
> me and are partially implemented as well -- would that be suitable for
> this list?
> 
> 
> Best regards
> Christian
> 
> [1] http://www.w3.org/WoT/
> [2] https://iotdb.org/
> 
> -- 
> Christian Amsüss                      | Energy Harvesting Solutions GmbH
> founder, system architect             | headquarter:
> mailto:[e-mail]  | Arbeitergasse 15, A-4400 Steyr
> tel:[phone]                | http://www.energyharvesting.at/
>                                      | ATU68476614