Re: [Asdf] Location information with SDF

[Ari Keränen <[e-mail]>]

Hi Christer, Hyunjeong, et al.,

First, thank you for clarifying the details of the location information proposal. I haven't had a chance to take part of the discussion much since I'm back on my leave now. But please find below some answers / comments to your questions & comments.

Also, here's attached a copy of the contents of the note at notes.ietf.org for future reference. The contents are identical to the Nov 8th revision, except I fixed one JSON error (extra brace) in the option two. 


Cheers,
Ari

> 
> From: Christer Holmberg
> Sent: Friday, November 10, 2023 2:31 PM
> To: Ari Keränen; 이현정; [e-mail]
> Subject: RE: Location information with SDF
> 
> 
> Hi,
>  
> I have some questions/comments regarding option 2.
>  
> ---
>  
> Q1:
>  
> The Properties in sdfObject "AddressInformation" provide data (location information) that applies to the "Humidimeter" sdfThing and the "Humidity" sdfObject.
>  
> I think that concept is not part of base SDF - Properties only apply to the sdfObject/sdfThing for which they have been defined.
>  
> (I know there are discussions about sdfRelations, and they are also used in Option 2. However, they are only used to map the location information to a dictionary - not to relate the information to other sdfThings/sdfObjects in the SDF model.)

Correct. Information about where the information would apply could be part of the semantics of the relation type or target of the sdfRelation but all this is currently undefined and would need to be specified if we go down the suggested path.

> ---
>  
> Q2:
>  
> Related to Q1, the location Properties in sdfObject "AddressInformation" apply to "Humidimeter" sdfThing. But, assume the following structure:
>  
>  
>                                       sdfThing "WeatherStation"
>                                          |
>                                          |
>                                       sdfThing "Humidimeter"
>                                          |
>                                          |
>                          ------------------------------------------
>                          |                                        |
>                      sdfObject "Humidity"                        sdfThing "Foo"
>                      sdfObject "AddressInformation"               |
>                                                                   |
>                                                                  sdfObject "Bar"
>  
> Does the location information also apply to "WeatherStation" sdfThing, and/or "Foo" sdfThing (and the "Bar" sdfObject)?

Same answer as to Q1: I think this would need to be specified in the relation information. Fixing this for all relations to one or the other sounds risky.

> ---
>  
> Q3:
>  
> Since (based on base SDF) the only technical difference between sdfThing and sdfObject is that sdfThing can contain other sdfThings and sdfObjects, does option 2 allow me to define sdfThing "AddressInformation"?

Yes. But you should use Thing only if you need the additional level of grouping so sdfObject seems appropriate here. 

  
> ---
>  
> Regards,
>  
> Christer
>  
>  
>  
> From: ASDF <[e-mail]> On Behalf Of Ari Keränen
> Sent: Tuesday, 7 November 2023 15.41
> To: 이현정 <[e-mail]>; [e-mail]
> Subject: [Asdf] Location information with SDF
>  
> Hi,
>  
> Regarding the discussion yesterday at the ASDF meeting about expressing location information with SDF for the Digital Twin use cases, here is a draft note of couple of different options on how the information could be expressed using sdfProperty, sdfObject, and sdfRelation constructs:
> https://notes.ietf.org/pPdkmO6nRRKEx7-IDRHhNw
>  
> The note contains three different options for structuring the postal address information shown in the example on slide 8 (https://datatracker.ietf.org/meeting/118/materials/slides-118-asdf-draft-lee-asdf-digital-twin-00). Each option has different merits, but probably the first or second one makes most sense here.
>  
> Would be good to test if this approach works for the more complex cases too and what kind of ontology terms are needed to make this work well with the DT use case.
>  
>  
> Cheers,
> Ari