W3C home > Mailing lists > Public > public-rdf-wg@w3.org > December 2013

Re: multiple-graph example in the Primner

From: Sandro Hawke <sandro@w3.org>
Date: Sat, 07 Dec 2013 11:01:30 -0500
Message-ID: <52A3465A.7020506@w3.org>
To: Yves Raimond <yves.raimond@bbc.co.uk>
CC: Kingsley Idehen <kidehen@openlinksw.com>, "public-rdf-wg@w3.org" <public-rdf-wg@w3.org>
On 12/07/2013 10:17 AM, Yves Raimond wrote:
> On Sat, 2013-12-07 at 10:12 -0500, Sandro Hawke wrote:
>> On 12/07/2013 09:54 AM, Kingsley Idehen wrote:
>>> On 12/7/13 8:28 AM, Sandro Hawke wrote:
>>>> Kingsley Idehen <kidehen@openlinksw.com> wrote:
>>>>> On 12/7/13 7:54 AM, Sandro Hawke wrote:
>>>>>> Pat, I agree with you about the situation except I believe there's a
>>>>> way out, which is why I stopped objecting back when we were making
>>>>> these decisions.  The way out is to define some vocabulary which
>>>>> communicates from Alice to Bob what kind of dataset semantics Alice is
>>>>> using.   That vocabulary doesn't need to be defined in a W3C
>>>>> recommendation to work.  So the primer just needs to posit that such
>>>>> vocabulary might exist, and give the example as a hypothetical.
>>>>> Alternatively, we could define that vocabulary (non rec track) right
>>>>> now and use it in the primer with a caution that this is only one of
>>>>> many possible ways to use datasets.
>>>>>>         - Sandro
>>>>> Yes, a vocabulary can be used to solve the problem. Net effect, we
>>>>> still
>>>>> don't need a confusing and contradictory example in the spec :-)
>>>> I don't see any need for it to be confusing or contradictory. I'd
>>>> suggest we choose the semantics where URLs used as graph names denote
>>>> sources which yield the associated graph.   I believe that's what the
>>>> overwhelming majority of readers will expect, so when they read the
>>>> example they will feel reassured that they can do what they
>>>> expected.   The only confusing thing will be our caveats, if we're
>>>> not careful.
>>>>     - Sandro
>>> The following examples will lead to confusion:
>>> ## Relative Resource URL serving as a Named Graph IRI
>>> <>
>>> {<#s> <#p> <#o> } .
>>> ## variation of the above
>>> <.ttl>
>>> {<#s> <#p> <#o> } .
>>> ## Absolute Resource URL serving as a Named Graph IRI
>>> <http://example.org/example.ttl>
>>> {<#s> <#p> <#o> } .
>>> ## 3 Named Graphs where two are derived from Resource URLs and one is
>>> an inferred Default (or System) Named Graph which maybe be associated
>>> with
>>> ## {<#s> <#p> <#o> } .
>>> ## OR
>>> ## a Union of the Default, <.ttl>, and <> .
>>> <.ttl>
>>> {<#s> <#p> <#o> } .
>>> <>
>>> {<#s> <#p> <#o> } .
>>> All of the examples above cover implementation and usage scenarios
>>> that are best covered via related specs such as SPARQL (for querying)
>>> and concrete syntaxes (NQuad, TriG etc.).
>>> It is best for the RDF spec, primer, and related collateral to focus
>>> on RDF semantics for triples based structured data representation. All
>>> examples should be simple for the reader to understand and follow :-)
>> I don't think the example has to be that complicated.   I think if we
>> just include in the TrigG details a triple like "<> a eg:WebSource" (and
>> mention that's a suitably defined class) then we're technically correct,
>> and users aren't particularly confused. (I'm happy to have it be
>> rdf:WebSource defined in a WG Note, if we want).
>> If the Primer also included a diagram of the dataset being discussed --
>> drawn as an RDF graph of the merge of the two named graphs and the
>> default graph, color coded and segmented by their names -- then I think
>> it wouldn't be confusing at all.
> Like
> https://dvcs.w3.org/hg/rdf/raw-file/default/rdf-primer/example-multiple-graphs.jpg

Oh, that's pretty.    Yes, something like that.  More below.
> with an additional statement <http://example.org/bob> a eg:WebSource?

I'd propose it's the dataset (<>) that's the WebSource, not /bob that's 
the WebSource.   Pat can correct me, but my sense is that nothing we say 
about the thing denoted by the graph name can affect what we're saying 
about the associated graph or how they are connected.   To say something 
about how the graph name's denotation and the associated graph are 
related, we have to say something about the dataset itself.

On the image itself, I'd suggest a few tweaks:

    - move the "Graph Name" bubbles to the bottom of their sections, and 
label them with some black text, saying "Graph Name: " before the graph 
name.   Or maybe "Named Graph, with name="
    - similarly label the default graph, at the bottom, with the black 
text "Default Graph".
    - change the styling of the default graph's http://example.org/bob 
bubble to be like the other subjects, in this case using a 
document/web-page icon like 
with the subject URL being http://example.org/bob
    - Use that same icon for the <> a eg:WebSource triple

Now I'm struggling with how to connect the two uses of 
http://example.org/bob.   One option is to put it on the border like 
MonaLisa, making it both a subject in the DG and the denotation of the 
NG.   Another option is to put a dotted arc between them, with a clever 
label I can't quite think of.   Another option is to just have them be 
visually close enough of the screen that it's obvious they are the same 

Another issue with the diagram is that by using rectangles, it's a style 
that only works for trivial datasets.     Using coloring or tagging of 
the arcs, or having them in layers, allows for arbitrary datasets.   So 
I have some concern that if people learn to visualize it this way, 
they'll get stuck when they try to think of, for instance, a subject 
being shared between 3 or more different NGs.

      -- Sandro

> Yves
>>         -- Sandro
>>> Kingsley
>>>>> Kingsley
>>>>>> Pat Hayes <phayes@ihmc.us> wrote:
>>>>>>> On Dec 5, 2013, at 4:53 AM, Guus Schreiber <guus.schreiber@vu.nl>
>>>>>>> wrote:
>>>>>>>> In the telecon yesterday there were some flames about the graph
>>>>>>> metadata examples in the Primer.
>>>>>>>> My position:
>>>>>>>> - There needs to be at least one example triple in the Primer in
>>>>>>> which a graph name is being used. Dropping this completely is for
>>>>> the
>>>>>>> editors a no-go.
>>>>>>> Including such an example is a no-go for me. I will formally object
>>>>> (or
>>>>>>> protest, or register a dissent, I am not sure of the exact W3C
>>>>> process
>>>>>>> involved here) if the WG publishes any document which implies that
>>>>> such
>>>>>>> usage is in any way supported by the RDF 1.1 specifications. That is
>>>>>>> *exactly* the semantic stumbling-point at which we were unable to
>>>>>>> provide any semantics for datasets. RDF 1.1 does NOT imply in any
>>>>> way
>>>>>>> that the use of a graph-name in an RDF triple can or should be
>>>>>>> understood to refer to the graph. On the contrary, it explicitly
>>>>> denies
>>>>>>> the validity of such an assumption.
>>>>>>>> - We are happy to consider other examples. Please suggest.
>>>>>>>> - We're happy to include other/updated caveats
>>>>>>>> Current phrasing included below. Text suggestions very much
>>>>>>> appreciated!
>>>>>>>> Guus
>>>>>>>> From
>>>>> https://dvcs.w3.org/hg/rdf/raw-file/default/rdf-primer/index.html#subsection-multiple-graphs
>>>>>>> :
>>>>>>>> [[
>>>>>>>> We can write down triples that include a graph name, for example:
>>>>>>>>     <http://example.org/bob> <is published by> <http://example.org>.
>>>>>>>>     <http://example.org/bob> <has license>
>>>>>>>>         <http://creativecommons.org/licenses /by/3.0/>.
>>>>>>>> These two triples could be interpreted as license and provenance
>>>>>>> information of the graph http://example.org/bob.
>>>>>>>> NOTE
>>>>>>>> RDF does not define the way in which the graph name and the graph
>>>>> are
>>>>>>> related. It is therefore up to application developers to decide how
>>>>> to
>>>>>>> interpret such triples.
>>>>>>> That does not deal with the central difficulty. RDF is intended for
>>>>> use
>>>>>>> in publishing data on the open Web. The issue involved here is, if
>>>>>>> Alice publishes an RDF dataset, and Bob reads it, how can Bob know
>>>>>>> whether a graph name used in RDF in the datset should be interpreted
>>>>> as
>>>>>>> referring to the graph it names? And the clear, unambiguous answer
>>>>>>> given by our RDF specs is, Bob cannot know this. There is no way
>>>>>>> specified to record or transmit this information through RDF or any
>>>>>>> means usable by an RDF engine. Alice might conform to the metadata
>>>>>>> useage needed by PROV, and Bob might read this and interpret it
>>>>>>> differently, without failing in any way to conform to RDF. So as far
>>>>> as
>>>>>>> RDF is concerned, this usage is invisible. Vague references to
>>>>>>> "application developers" does not deal with this issue.
>>>>>>> Pat
>>>>>>>> ]]
>>>>>>> ------------------------------------------------------------
>>>>>>> IHMC                                     (850)434 8903 home
>>>>>>> 40 South Alcaniz St.            (850)202 4416   office
>>>>>>> Pensacola                            (850)202 4440   fax
>>>>>>> FL 32502                              (850)291 0667 mobile
>>>>>>> (preferred)
>>>>>>> phayes@ihmc.us       http://www.ihmc.us/users/phayes
> -----------------------------
> http://www.bbc.co.uk
> This e-mail (and any attachments) is confidential and
> may contain personal views which are not the views of the BBC unless specifically stated.
> If you have received it in
> error, please delete it from your system.
> Do not use, copy or disclose the
> information in any way nor act in reliance on it and notify the sender
> immediately.
> Please note that the BBC monitors e-mails
> sent or received.
> Further communication will signify your consent to
> this.
> -----------------------------
Received on Saturday, 7 December 2013 16:01:37 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:04:36 UTC