Re: Actuation and Actuators in SOSA (issue-91)

Please, Kerry, you need to stop using words like “silly”. Group members are offended by this.

From: Kerry Taylor <kerry.taylor@anu.edu.au>
Date: Wednesday, 15 February 2017 at 7:48 am
To: "Simon.Cox@csiro.au" <Simon.Cox@csiro.au>, "maxime.lefrancois@emse.fr" <maxime.lefrancois@emse.fr>, "jano@geog.ucsb.edu" <jano@geog.ucsb.edu>, Armin Haller <armin.haller@anu.edu.au>, "janowicz@ucsb.edu" <janowicz@ucsb.edu>
Cc: "public-sdw-wg@w3.org" <public-sdw-wg@w3.org>
Subject: RE: Actuation and Actuators in SOSA (issue-91)

I’m sorry I have to disagree with making up a new word (and such an ugly one – although I admit that is totally a personal reaction).   There are plenty of good English words out there for what we are trying to explain here. We are doing no service to anyone to invent such a meaningless new term.  What is wrong with looking at previous work in this area? What is wrong with a nice useful word like “affects” that seems to carry the right idea (I suppose, assuming I ‘get’ the right idea).


Ø  We need similar concepts for actuation.


Not sure…. Could we please see  an explanation for this idea, and some worked examples ? Most ideally with reference to our own use cases!

Let’s not get too carried away with the idea actuation is just like observation –that might be true at a surface level but certainly is not with deeper analysis. And “pretending” they are the same by making up silly names to make them “look” the same at the surface level   does not make any sense to me.

Of course, SENSORML also has something to say in this area. Should we ignore it? And if so, why?

-Kerry

From: Simon.Cox@csiro.au [mailto:Simon.Cox@csiro.au]
Sent: Wednesday, 15 February 2017 6:55 AM
To: maxime.lefrancois@emse.fr; jano@geog.ucsb.edu; Armin Haller <armin.haller@anu.edu.au>; janowicz@ucsb.edu
Cc: Simon.Cox@csiro.au; public-sdw-wg@w3.org
Subject: RE: Actuation and Actuators in SOSA (issue-91)

> Kerry: ActuableProperty is also not English. What is meant here? Perhaps an explanation of the concept would help to choose the term. SEAS uses "Property" which suits me, but I guess we are stuck in a pattern since we have "ObservableProperty elsewhere. SAN uses ImpactedProperty which is certainly better, and that would also suggest actuatedProperty could be 'impacts'. Or, better still (becuase impacts is too forceful, in general) how about "affects" and "AffectedProperty"

In all this we need to preserve the distinction between the class name and definition, and the associated property name and definition. For observations we distinguish Observable Properties - i.e. potentially observable by sensors - from observed properties - i.e. actually observed in an observation. A set of *observable* properties might be published in a list or register, for re-use in multiple observation instances, where their role becomes *observed*.

We need similar concepts for actuation.

And yes, "actuable" is a new word, but is clearly related to existing English and new coinages for specific purposes are nothing new in technical contexts. Actionable may be an acceptable alternative, though to me it does not carry quite the same meaning.

Simon
________________________________
From: Maxime Lefrançois <maxime.lefrancois@emse.fr<mailto:maxime.lefrancois@emse.fr>>
Sent: Tuesday, 14 February 2017 7:01:44 PM
To: Krzysztof Janowicz; Armin Haller; Krzysztof Janowicz
Cc: Cox, Simon (L&W, Clayton); public-sdw-wg@w3.org<mailto:public-sdw-wg@w3.org>
Subject: Re: Actuation and Actuators in SOSA (issue-91)

Hi,

I added some answers to Kerry's questions in the wiki page https://www.w3.org/2015/spatial/wiki/Actuation



These are copied here:


Kerry: can we reconsider the names please? "actsOnProperty" (from SEAS) instead of "actuatedProperty" (does not follow active property naming convention, is not English)

- Maxime: +1 for "sosa:actsOnProperty/sosa:isActedOnBy" and "sosa:observesProperty/sosa:isObservedBy", for the sake of having consistent naming conventions.

Kerry: ActuableProperty is also not English. What is meant here? Perhaps an explanation of the concept would help to choose the term. SEAS uses "Property" which suits me, but I guess we are stuck in a pattern since we have "ObservableProperty elsewhere. SAN uses ImpactedProperty which is certainly better, and that would also suggest actuatedProperty could be 'impacts'. Or, better still (becuase impacts is too forceful, in general) how about "affects" and "AffectedProperty"

- Maxime: related emails in the list: https://lists.w3.org/Archives/Public/public-sdw-wg/2017Feb/0335.htmlhttps://lists.w3.org/Archives/Public/public-sdw-wg/2017Feb/0338.html https://lists.w3.org/Archives/Public/public-sdw-wg/2017Feb/0339.html .

- Maxime: propose: "sosa:ActionableProperty"

Kerry: What is a Phenomenontime in this context? As distinct from a ResultTime? Why do we need it?

- Maxime: AFAIK, resultTime can be later than phenomenonTime. As an example in the spec, maybe we could use the example of an astronomical telescope that outputs today some phenomenon that occurred many years ago?

Kerry: What is the impact on SSN?

- Maxime: should we duplicate any axiom that exists for Observation and adapt it for Actuation?

- Maxime: should we decide which of the MeasurementProperty can also apply to Actuators? As a first guess, I would say Accuracy, ActuationLimit, Drift, Frequency, Latency, Precision, Resolution, ResponseTime, all apply to Actuation

- Maxime: I believe all of the OperatingProperties also apply to Actuators.




Best,
Maxime

Le lun. 13 févr. 2017 à 10:55, Maxime Lefrançois <maxime.lefrancois@emse.fr<mailto:maxime.lefrancois@emse.fr>> a écrit :
Dear Simon, all,

From my side, it's 'yes' to your second question.

 - if requirement 5.27 [1]  is sufficient to motivate the addition of actuator/actuation, then requirement 5.16 may be sufficient to motivate the addition of the Samping side of the system.
 - as far as I know, not all of GoodRelations has been swallowed by schema.org<http://schema.org> anyways, and this is managed by the W3C Schema.org Community Group [2]. So it's not a 'all or nothing' matter there. If Samping is is SOSA and the schema.org<http://schema.org> community doesn't want sampling, then it won't make them reject Actuation.

+1 for Simon to create a wiki page with turtle snippets that explain your proposal, (potentially multiple options) ?

@Jano, could you also write turtle snippets for your proposed alternative in the Wiki ?

Kind regards,
Maxime


[1] - https://www.w3.org/TR/sdw-ucr/#ExSituSampling

[2] - https://github.com/schemaorg/schemaorg


Le lun. 13 févr. 2017 à 08:14, Krzysztof Janowicz <jano@geog.ucsb.edu<mailto:jano@geog.ucsb.edu>> a écrit :
Hi Simon, Armin, all,
I fully agree with keeping SOSA as minimalistic as possible. This is a key design goal. The changes I proposed are a reaction to issue-91 and other change requests and they are minimal in nature by only introducing one class and one property. They are also in line with other work on actuators. The fact, that such minimal changes were sufficient to address the outstanding issues shows that by now SOSA seems to stabilize and is well designed. One could even fix these issues by an even more minimalistic change, I will implement this tomorrow as alternative.

As far as sampling is concerned, I absolutely agree that Sample needs to be in SOSA. Whether it is of equal importance compared to observations and actuations is difficult to say. Simon, may I suggest that you create a similar example for sampling? If all we need would be just one or two more classes, then I would support to add it. Otherwise, we could leave Sample in there as stub and add more axioms to the new SSN.

More generally speaking (and leaving the sampling issue aside), my big concern is that we will start doing this for 10 more cases, thereby ruining the entire idea of a lightweight SOSA. To be very clear about this, I created this proposal because I was tasked to do so. I believe that SOSA will be fine with said changes (as they are minimal) to better support actuation but that SOSA would also remain valuable without these changes. If this opens the flood gates to tons of change requests for new classes and properties, I would strongly prefer to leave SOSA as is. SOSA was never designed to capture all use cases and all details in a balanced way as this is the task of the SSN.


Cheers,
Jano

On Sun, Feb 12, 2017 at 10:11 PM, Armin Haller <armin.haller@anu.edu.au<mailto:armin.haller@anu.edu.au>> wrote:
I will raise the question of Sampling in the core in the discussion around Actuation in our next telco.

In terms of Actuation we have several use cases that require actuation: https://www.w3.org/TR/sdw-ucr/#ModelActuation I believe we need to have a strong argument why to not include it in the core.

Personally, I think Actuation should be in SOSA as many IoT applications on the Web will include Actuation. Even many of the IoT home devices available in Apple Stores include actuation (turning light on, recording your favourite show over Siri, Cortana, Amazon Echo, changing the thermostat etc.).

On 13/2/17, 11:50 am, "Simon.Cox@csiro.au<mailto:Simon.Cox@csiro.au>" <Simon.Cox@csiro.au<mailto:Simon.Cox@csiro.au>> wrote:

    Thanks Jano.

    The proposal is exactly in line with expectations.

    However, I am concerned that we should clarify the scope (and size) of SOSA. Specifically,
    1. do the requirements for SOSA include a basic actuation model?

    If that is the case then
    2. should the Sampling side of the system also need to be fleshed out?
    I could make a proposal for this, but had been holding back because I had assumed that was probably out of scope for most SOSA users, and should rather be the subject of a vertical (richer axiomatization) + horizontal (additional scope) extension to SOSA.

    In developing SOSA until now we have generally leaned towards parsimony - lets minimise the number of concepts in SOSA to a core that might be useful to schema.org<http://schema.org> folk.

    BTW - I'm OK with the answers to these two questions being 1. Yes, and 2. No, but wanted to put the issue on the table so we are all clear about what is being ruled in, and what is out.

    And just in case there is any question, even if it is "2. No", Sample still belongs in SOSA, as it is critical for many (most?) observations.
    It would just be sampling and sample preparation that would be delegated elsewhere.

    Simon

    -----Original Message-----
    From: Krzysztof Janowicz [mailto:janowicz@ucsb.edu<mailto:janowicz@ucsb.edu>]
    Sent: Monday, 13 February, 2017 10:50
    To: Cox, Simon (L&W, Clayton) <Simon.Cox@csiro.au<mailto:Simon.Cox@csiro.au>>; armin.haller@anu.edu.au<mailto:armin.haller@anu.edu.au>; public-sdw-wg@w3.org<mailto:public-sdw-wg@w3.org>
    Subject: Actuation and Actuators in SOSA (issue-91)

    Dear all,

    I added a wiki pages that shows a concept map for the changes to be made on the Actuator and Actuation side of SOSA. The proposed changes address some shortcomings of the current model and are also in preparation for a deeper axiomatization in SSN.

    There are two major (but in no sense dramatic changes) to SOSA, namely a proposal to add the SOSA:actuatedProperty role and a class called SOSA:ActuableProperty.  These are in line with previous work and requests made on this list.

    I hope you can look at the concept map and the notes on the wiki page as I hope we can get this resolved during our next teleconference. Please keep in mind that everything that is not shown in a dashed style is already part of SOSA.

    https://www.w3.org/2015/spatial/wiki/Actuation_in_SOSA


    Best,
    Jano