[WG-InfoSharing] W3C Data Privacy Vocabulary - Consent Receipt Inputs
james.g.hazard at gmail.com
Sun Jun 23 15:16:59 UTC 2019
Thanks Harsh. Comments also inline, though no guarantee that they are not out of line.
> On Jun 23, 2019, at 1:15 AM, Harshvardhan J. Pandit <me at harshp.com> wrote:
> Hi Jason, comments inline.
> Regards, Harsh
> On 21 Jun 2019, 16:34 +0100, James Hazard <james.g.hazard at gmail.com>, wrote:
>> A few quick thoughts, in line on Andrew’s list:
>>> On Fri, Jun 21, 2019, 7:32 AM Andrew Hughes <andrewhughes3000 at gmail.com <mailto:andrewhughes3000 at gmail.com>> wrote:
>>> Some more questions, if you don't mind...
>>> A) can we treat the list of terms in the vocabulary as exactly that: a controlled word list?
>> Most of legal documentation can be done as a “controlled list” of words, phrases, document forms. The problem is who is in “control.” That is, a closed system will always be incomplete and therefore force edge cases and diversity into a standard vocabularies and conformity. It can centralize “control” of the vocabulary and hence the thoughts. Prototype inheritance enables “permissionless” variations at the edge.
> Indeed. I’ll say that Prototype inheritance as you define it is in essence similar to the way inheritance or hierarchy works in RDF as both allow specifying a common parent or top-level concept which can be specialised as required within the use-case. But there needs to be a common vocabulary for the top-level concepts for everyone to align their specialised information under it.
It’s been a long time since I looked at RDF, but what I found then (and everywhere else I looked) was that the top level bound the lower levels. In my non-IT terms, the parent set limits on the child. With prose objects (which I iterated my way into) the child can do what it wants and picks its parents, which can be multiple. So an ordinary consent would have parents that included a form of consent, each of the parties, the place the consent was done, a taxonomy of purposes, etc. Most of those could be bundled into an info page and then the consent would need only the specifics.
So, no you don’t need a common vocabulary for the top-level, unless you want to use that to impose rules within some domain.
>>> B) what is supposed to happen when a word has more than one definition? Or is the vocabulary not about definitions but rather about "list of words”?
>>> C) regarding the RDF - if one were to use, for example, JSON-LD and refer to schema.org <http://schema.org/> context and also this RDF - should it work? (Recognizing that this question is really stretching the limits of my knowledge on semantic web-ish topics - so please rephrase the question if needed)
>> I am far from an expert on this subject, but I found that RDF over-solves the problem of managing vocabularies. Schema.org <http://schema.org/> is great, but also too limited, so it either needs a way to fork and build, or one needs to start otherwise and connect to it.
>> JSON-LD seems really useful, though I’ve found that you only need a very limited set to do most of the work.
> As I pointed out to Andrew in the previous email, JSON-LD *is* a way to express RDF, so when you’re using JSON-LD - you are using RDF. I agree that RDF is clunky and too verbose for most needs - but it does provide a terse way to express graphs, and with JSON-LD, a way to share graphs in a native web format. Besides that, it is academically sound in that it represents a formal model for expressing knowledge. If it’s just a list of words - a vocabulary, RDF is definitely not the way to go. However, the moment you have any relationship to express between them - you have a graph, and RDF does help to express it.
Yes. We found that a whole system of text and data (prose and parameters) can be made with only Records which have key/values of “Data” (actually strings) and “Edges” (references to other Records). The automation can be included by referencing hashes. On Github it is done by plain text files, but can also be done in JSON, JSON-LD (using only a tiny subset of its capabilities), XML or, say, a graph database. We did a bit with IPLD (IPFS’s linked data) but it imposed a “/“ as the inheritance crossed file boundaries. (That also was years ago.)
> I would like to point out that I’m not claiming RDF is a silver-bullet, it is not. However, I’m hard pressed to identify another ’standard’ that can be expressed in web-native serialisations (JSON-LD), that can be queried (SPARQL,GRAPH-QL), and is capable of expressing reasoning and logic (OWL2). So I’ll say it depends on what the intended use is.
Agreed, and I’ve love to see it. There is no inconsistency, I think, between those and prose objects. Prose objects are, as I understand this, at a different layer of abstraction. JSON, XML, etc. are what @xmlgrrl called “punctuations.”
The important point is to avoid centralizing rules or complexity, to permit edge-based adaptation and decision-making.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the WG-InfoSharing