Re: [TF-LD] From compact representations to semantic lifting

Dear Kjetil, all,

For the lifting advertisement part, that's something I have been
investigating recently, and proposed to use a 'Content-Lifting-Rule' header
whose value is a URI that identifies the lifting mechanism.

Your application can then recognize some lifting rule URIs, or dereference
the URI and apply the rule you find there using a generic lifting mechanism
(for example, if you retrieve a application/vnd.sparql-generate, you can
use our SPARQL-Generate implementation https://w3id.org/sparql-generate/ )

See paper https://2018.eswc-conferences.org/paper_159/ submitted at ESWC
this year (rejected unfortunately, but the ideas are out there)

Best,
Maxime Lefrançois

Le mar. 24 avr. 2018 à 00:31, Kjetil Kjernsmo <kjetil@kjernsmo.net> a
écrit :

> Hello again!
>
> I've been working a bit again on the energy system in my house, also as an
> exercise for IoT. I have working an Arduino Uno with some temperature
> sensors, mostly around a heat exchanger, and I'd like to monitor the
> efficiency of heat exchanger. There exists a formula for that. :-)
>
> I have a minimal HTTP server on it, and the response is simply a JSON with
> the sensor ID as a key and the temperature as the value. Compact enough, I
> think. But being a semwebber, I'd like to have applications use RDF, so at
> some point, I would like to perform some semantic lifting.
>
> Obviously, it is pointless for an RDF template to reside on the Arduino, a
> key challenge is to attach knowledge about the data to them. In some
> cases,
> like mine, the semantics shouldn't be very far from the microcontroller,
> since those who implemented that also has immediate knowledge about the
> data. So, I would like to see a mechanism where I can connect both the TD
> and the semantic lifting from the Arduino.
>
> I was thinking about this in terms of protocol bindings, primarily. For
> HTTP, a Link header seems to come naturally, whereever the knowledge sits,
> a Link header is inserted pointing to the TD and the semantic lifting, and
> so a processor could create RDF data. In my case, I would have the Link
> header in the flash of my Arduino. I'm not too familiar with other
> protocols, though I have realized there is a whole zoo out there.
>
> I'm not sure how far the LD task force has gotten in their work, but I
> figured I'd share these thoughts now, perhaps they be the beginning of a
> use
> case?
>
>
> Code is here, BTW:
> https://github.com/kjetilk/arduino_tempmonitor
>
> Cheers,
>
> Kjetil
>
>