Re: Source data for provenance graph in ProvenanceExampleAndConcept1

Graham and Luc -

Thank you for the responses to my questions. Much appreciated. I now fully 
understand the history - and the constraints.

At this point, my concern will now be on how to communicate this approach to 
the GIS/geospatial/sensor community represented by the OGC. This is a 
community that has a great need for standard ways of expressing and 
communicating provenance but also has a huge legacy in the use of existing 
de-facto, vendor specific, and ISO standards, such as 19115, and the use of 
XML, binary encodings, and other serializations for expressing metadata.

As part of that outreach and communication to the OGC community, any 
thoughts regarding adding a map to the use case? This would really help!

Also, another question. If some of the elements of provenance information 
have a location element (which I hope is true), any though of using 
GeoSPARQL (geo extensions to SPARQL)? We have shared the GeoSPARQL candidate 
standard with the W3C SPARQL community for comment.

Thanks and regards

Carl


----- Original Message ----- 
From: "Graham Klyne" <GK@ninebynine.org>
To: "Carl Reed" <creed@opengeospatial.org>
Cc: "Paul Groth" <pgroth@gmail.com>; "Timothy Lebo" <lebot@rpi.edu>; 
<public-prov-wg@w3.org>
Sent: Wednesday, May 18, 2011 3:18 AM
Subject: Re: Source data for provenance graph in 
ProvenanceExampleAndConcept1


> Carl Reed wrote:
>> 2. I have to ask why we are assuming that the data is published as RDF. 
>> Typically in a use case, tools or technologies are abstracted. The data 
>> could just as easily have been published as XML (which for statistics 
>> data and map data is probably the case). I think we should simply state 
>> the GovData source publishes the data using a standard encoding language.
>
> While I agree that the conceptual model for provenance should stand 
> independently of any particular representation, the use of RDF is somewhat 
> baked into the WG charter (http://www.w3.org/2011/01/prov-wg-charter):
>
> (1) Use of RDFS and OWL for describing the formal model
>
> (2) use of SPARQL for querying provenance
>
> both of which require that the provenance information can be presented 
> with respect to the RDF abstract syntax (however it may be represented 
> internally), and suggests use of RDF/XML (this currently being *the* 
> original W3C RECommendation for exchanging RDF data, and this being a W3C 
> working group). (RDFa is also formally a recommendation, but my sense is 
> that this is primarily useful for mixing RDF and human-readable text in a 
> single document, and is not necessarily ideal for exchanging raw 
> provenance data, but that's up for debate.)
>
>> 3. I am not familiar with turtle serialization so I did a bit of 
>> research. I checked Druple and Wordpress. They do not use turtle 
>> serialization. I checked Wikipedia. No entry that I could find. So, 
>> perhaps we should again not mention a specific technology - just simply 
>> state that the analyst downloads a serialization (could just as easily be 
>> RDFa).
>
> Yes, any of the common RDF serializations could be used, but in the 
> interests of interoperability I think we should be prepared to recommend 
> one as the PIL (or whatever we call it).
>
> Choices:
> - http://www.w3.org/TR/rdf-syntax-grammar/ (RDF/XMK)
> - http://www.w3.org/TeamSubmission/turtle/ (N3/Turtle)
> - http://www.w3.org/TR/rdfa-syntax/ (RDFa)
>
> Each have different advantages in different environments, but at heart 
> they all convey the same underlying abstract syntax and semantics:
> - http://www.w3.org/TR/rdf-concepts/ (abstract syntax)
> - http://www.w3.org/TR/rdf-mt/ (semantics)
>
> #g
>
>
>
>

Received on Wednesday, 18 May 2011 15:11:42 UTC