W3C home > Mailing lists > Public > public-linked-json@w3.org > February 2019

Re: Slides for Berlin Data Workshop

From: Antoine Zimmermann <antoine.zimmermann@emse.fr>
Date: Wed, 27 Feb 2019 10:19:34 +0100
To: sandro@hawke.org
Cc: public-linked-json@w3.org
Message-ID: <2ef24afe-8367-c1ec-5c64-ce8036922234@emse.fr>

In my opinion, this topic should not be debated in the JSON-LD group. It 
is of course appropriate for the N3 group though.

Since you ask about graph metadata, I would never use RDF to describe 
RDF graphs. I would describe the files, or the information resources 
that encode RDF graphs. These things have a creator, a creation date, 
access rights, etc. RDF graphs (i.e. sets of RDF triples) are not 
created. They all are existing in the set of all RDF graphs at the same 

In this regard, the interpretation of Carroll et al. is superior: by 
interpreting the graph name as the named graph pair, you can distinguish 
the creation dates of two graph-encodings:

<#mygrah> dc:created "2019-02-26T16:53:42+01:00"^^xsd:dateTime .
<#yourgrah> dc:created "1999-01-01T00:00:00Z"^^xsd:dateTime .
<#mygraph> { <timbl> a <Person> }
<#yourgraph> { <timbl> a <Person> }

If <#mygraph> and <#yourgraph> are interpreted as the graph "{<timbl> a 
<Person>}", then this graph has 2 creation dates, which is probably not 
what you want with your metadata.


Le 26/02/2019 à 14:39, Sandro Hawke a écrit :
> On 2/26/19 4:49 AM, Antoine Zimmermann wrote:
>> Hello,
>> This is my first email to the JSON-LD CG mailing list, so let me 
>> introduce myself: I am Antoine, working on semantic web technologies 
>> since 2004 as a researcher. I participated in the RDF 1.1 working 
>> group, quite actively, where my primary interest was RDF semantics.
>> Concerning your slides, Gregg, on slide 2, you say "The only 
>> reasonable interpretation of graphs named via blank node (...) is that 
>> the blank node denotes the graph it names".
>> Graph names can be interpreted in many ways, lots of which are 
>> considered reasonable by those who advocate them.
> This is a potentially large topic.  Perhaps we can take it to github?  I 
> wrote up this exact issue (in the context of the N3 CG), at 
> https://github.com/w3c/N3/issues/1, so that's probably a good place to 
> do it.
> Gregg is advocating for Option 2, which I think has a lot of merit, 
> although I agree it's not "the only reasonable interpretation".  I'd 
> love to hear how you propose to, for example, publish graph metadata, 
> Antoine.
> I separated the logic question out to 
> https://github.com/w3c/N3/issues/2, since I think it's distinguishable. 
> My current applications (credibility/disinformation) are about 
> provenance not about representing rules, so issue 1 is much more 
> important to me right now.
>    -- Sandro
>> In fact, the very people who introduced the concept of named graphs 
>> (Carroll et al. in 2004) defined a formal semantics according to which 
>> an RDF interpretation I satisfies ("conforms to", in their words) a 
>> named graph (n,g) iff I(n) = (n,g), that is, the name is interpreted 
>> as the named graph *pair*, not the graph.
>> Many people have used the idea of quads (that can be seen as a 
>> syntactic variation of the concept of named graphs) in very different 
>> ways, some of which are implemented in triple stores (e.g., 
>> spatio-temporal triple store Strabon).
>> In any case, defining the meaning of a JSON-LD document is not part of 
>> the JSON-LD group's mission. JSON-LD defines how to map a JSON-based 
>> format to the abstract RDF structures, then people interpret it as 
>> they want, possibly following other specs like RDF Semantics, OWL, 
>> SWRL, or N3logic.
>> Similarly, slide 5 is not about "Reasoning in JSON-LD": it is 
>> explaining how to map N3 formulas to JSON-LD. Then people can decide 
>> to interpret JSON-LD documents as N3, following slide 5 
>> representation, and do *N3 reasoning*, not "JSON-LD reasoning". They 
>> could also just map this representation to a normal RDF dataset and 
>> apply other kinds of reasoning.
>> Best,
>> --AZ
>> Le 23/02/2019 à 23:50, Gregg Kellogg a écrit :
>>> The format for the Berlin Data Workshop [1] remains unclear, but I’ve 
>>> prepared just a couple of slides to describe one way in which 
>>> Anonymous Named Graphs in JSON-LD could support the property graph 
>>> use case.
>>>> https://json-ld.org/presentations/JSON-LD-Support-for-Property-Graphs/ 
>>>> <https://json-ld.org/presentations/JSON-LD-Support-for-Property-Graphs/> 
>>> There’s a short overview of new things in JSON-LD 1.1, and as a 
>>> bonus, a sketch of how Notation3 reasoning might look in JSON-LD. 
>>> (Hint, we really only need to invent a way to describe universal 
>>> variables at the syntax level; reasoning should be universal based on 
>>> obvious projections from Notation 3. The required extensions to RDF 
>>> Datasets and better description of reasoning semantics are work to be 
>>> done elsewhere).
>>> Gregg Kellogg
>>> gregg@greggkellogg.net
>>> [1] https://www.w3.org/Data/events/data-ws-2019/schedule.html

Antoine Zimmermann
Institut Henri Fayol
École des Mines de Saint-Étienne
158 cours Fauriel
CS 62362
42023 Saint-Étienne Cedex 2
Tél:+33(0)4 77 42 66 03
Fax:+33(0)4 77 42 66 66
Member of team Connected Intelligence, Laboratoire Hubert Curien
Received on Wednesday, 27 February 2019 09:19:59 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 16:18:52 UTC