Re: agenda for tomorrow's DPVCG call

Hello.

Regarding adding the suggested classes into DPV, I would suggest the 
following:

1> Personal Data Category: TV Viewing Behaviour

- Instead of directly creating a category under Behavioral, we can have 
the concept of "Service Consumption Behavior" where information of how 
the individual consumes a service is captured, either by observation or 
as analytics. This can include watching TV, netflix, whatever.

- This is far too broad a category for my comfort. Does it also include 
information about the individual, for example, my heartbeat when 
watching TV? Or my facial recording while I watch TV? Or does it only 
include my interactions with the TV - e.g. when did I change channels, 
what was I watching, timestamps, etc? How to specify this in the definition?

2> Purpose: CreateEventRecommendations

- AFAIK, Personalisation and Generating Recommendations are two separate 
and distinct purposes. Something can be personalised without 
recommendations, and something can be recommended without 
personalisation. I suggest we create a top-level purpose for Generating 
/ Provision of Recommendations. We currently have a 
CreateProductRecommendations as subclass of ServicePersonalisation, 
which I now feel needs to be corrected based on the above argument.

- IMO, CreateEventRecommendations does not indicate whom the 
recommendations are being created for. If they are to be given to the 
user, a better wording could be "provision of recommendations". If they 
are being used for the organisation, "generating event recommendations" 
would be more technically apt.

Regards,

Harsh

On 11/11/2019 17:14, Bert Bos wrote:
> On 11/11/19 18:01, Bert Bos wrote:
>> And one more agenda item, see below:
>>
>> On 11/11/19 16:34, Axel Polleres wrote:
>>> Dear all,
>>>
>>> Apologies for dropping out short notice last time, but I am confident I
>>> can chair tomorrow, November 12
>>>
>>> Again, we'll continue our quest to close ISSUES and ACTIONS to arrive at
>>> a final CG report until the end of the year!
>>> Apart from that, *we will have a "guest talk: by Christoph Lange to
>>> indroduce the IDS  Data model [1]*
>>> *
>>> *
>>> *1 . **http://ids.semantic-interoperability.org/ *(@christoph, that's
>>> the latest version I found, please feel free to send another link if
>>> there's a newer one)
>>>
>>> ---------------
>>> Agenda & dial-in information:
>>>
>>> Dial-in info: see
>>> https://lists.w3.org/Archives/Member/internal-dpvcg/2018Aug/0000.html
>>>
>>> IRC: ircs://irc.w3.org/dpvcg (irc.w3.org <http://irc.w3.org> channel #dpvcg)
>>> or connect a Web browser to http://irc.w3.org/
>>>
>>> Agenda:
>>>
>>> 1) Roll call, select scribe, agenda
>>>
>>> 2) approve last time minutes:
>>> https://www.w3.org/2019/10/29-dpvcg-minutes
>>>
>>> 3) Liaisons: Christoph Lange will tell us about the Industrial Data
>>> Spaces project, and their Information Model, which has some
>>> overlapping/complementary requirements in terms of describing privacy
>>> aspects of IDS Data Resources and of IDS Data Apps. (there was a long
>>> standing action on this on me ;-))
>>>
>>> 4) Action items
>>> https://www.w3.org/community/dpvcg/track/actions/open
>> - How to encode TV viewing behavior with the dpv vocabulary?
>>
>> For one of the use cases in SPECIAL, the event recommendations app by
>> Proximus, some types of user behavior are collected (for the purpose of
>> deriving interests and ultimately recommend events to the user). Two
>> types of behavior are collected: web browsing (what sites the user
>> visited) and TV viewing (what programs the user watched).
>>
>> We have dpv:BrowsingBehavior[1] (a subclass of dpv:Behavioral[2]) for
>> the former, but how do you encode the latter?
>>
>> Should we add something like TvViewingBehavior? Or are there other
>> solutions?
>>
>> [1] https://www.w3.org/ns/dpv#browsing-behavior
>> [2] https://www.w3.org/ns/dpv#dpv:Behavioral
> Here is a mail from Rudy Jacob (which somehow didn't make it to the
> list) with a concrete proposal:
>
> ----------------------------------------------------------------------
> Dears Community members,
>
> Proximus is one of the three industry partners of SPECIAL and we
> contributed to the W3C working group at its birth in Vienna 2018 with
> our use case:
>
> https://www.w3.org/community/dpvcg/wiki/SPECIAL/Proximus_use_case
>
>
> SPECIAL is ending soon and we have our last deliverable by the end of
> this month. Upon using and documenting the use case we noticed that two
> key vocabulary words are missing.
>
> Our suggestion (for discussion) is listed below. I will be personally
> attending the call on Nov 12th, and we hope that these two issues can be
> resolved before the end this month so that we can properly link to them.
>
>
> Many thanks,
>
>
>
> Rudy JACOB
> Business Program Manager Innovation
> www.proximus.be
>
>
>
>
>      Add a DATA class for TVViewingBehaviour (subclass of
> PersonalDataCategory).
>
> Class:	
> 	dpv:TVViewingBehavior
> Description:
> 	Information that relates to an individual's TV viewing behavior.
> is SubClass of:
> 	dpv:Behavioral
> Class:
> 	dpv:Behavioral
> Description:
> 	Information that describes an individual’s behavior or activity,
> 	on-line or off
> is SubClass of:
> 	dpv:External
>
> Class:
> 	dpv:External
> Description:
> 	Personal Data that can be observed by another person i.e. has
> 	external characteristics that make it visible
> is SubClass of:
> 	dpv:PersonalDataCategory
>
>
>
>      Add a PURPOSE class for EventRecommendations (subclass of
> ServicePersonalization).
>
> Class:
> 	dpv:CreateEventRecommendations
> Description:
> 	To create event recommendations
> is SubClass of:
> 	dpv:ServicePersonalization
>
> Class:
> 	dpv:ServicePersonalization
> Description:
> 	Personalization of service or activity for the user
> is SubClass of:
> 	dpv:Purpose
>
> ----------------------------------------------------------------------
>
>
>>> 5) Issues
>>> https://www.w3.org/community/dpvcg/track/issues
>>>
>>> 6) AOB
>
>
> Bert

-- 
---
Harshvardhan Pandit
PhD Researcher
ADAPT Centre
Trinity College Dublin

Received on Monday, 11 November 2019 21:20:49 UTC