W3C home > Mailing lists > Public > public-dpvcg@w3.org > September 2020

Re: DPV semantics: how to specify values?

From: Harshvardhan J. Pandit <me@harshp.com>
Date: Thu, 3 Sep 2020 08:52:00 +0100
To: Víctor Rodríguez Doncel <vrodriguez@fi.upm.es>, public-dpvcg@w3.org
Message-ID: <0ab59af6-5695-6d2a-497b-75f0afa27a90@harshp.com>
Hi Victor.

On 03/09/2020 07:46, Víctor Rodríguez Doncel wrote:
> 1) I believe making a taxonomy is an interesting enterprise /per se/, 
> but it entails some additional work and it is a very important change 
> over the previous work.

Agreed.
However, we codified the DPV with the intention of providing a taxonomy 
(hence the hierarchy of concepts!) so there isn't change in the concepts 
or work per-se. Or am I missing your point here?

> If you want to better define the ontology as a data model... why dont 
> you define some RDF Shapes instead? This would be far more practical. 
> Also, we may want to specify a cookbook with the most common patterns of 
> use.

Okay, I used the wrong choice of words.
Instead of data model, I should have said ontological modelling of data 
i.e. ontology. I was trying to refer to what we currently have as 
personal data handling.
(SHACL/ShEx) Shapes (AFAIK) are good to express constraints or even 
queries (if we extend their gambit) - but they still need an ontological 
model consisting of classes and properties.

My point was that the taxonomy is separate from this ontological model 
i.e. PersonalDataHandling doesn't really care much for the taxonomy 
except for existence of top-level concepts (E.g. Personal Data, 
Purpose). So we can have the taxonomy (in SKOS) + ontology (in OWL-DL).

I agree we should be specifying some list of patterns showing use. This 
is exactly why I brought this issue in connection with specifying 
examples of the DPV.

> 
> 2) Not important and I am totally out of context but in the example 
> below... doesn't make sense simply declaring this?
> :x a :Email .

This is fine if we wanted to refer to an instance of Email. But the 
other use-case is problematic: referring to Email itself i.e. saying "we 
collect (all/any) email" and not just a specific email (which is what 
the triple specifies).

-- 
---
Harshvardhan Pandit, Ph.D
Researcher at ADAPT Centre, Trinity College Dublin
https://harshp.com/research/
Received on Thursday, 3 September 2020 07:52:15 UTC

This archive was generated by hypermail 2.4.0 : Thursday, 24 March 2022 20:27:59 UTC