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

Re: RDF-ISSUE-121 (merge-is-union): Should the merge of two RDF graphs be defined as their set union? [RDF Concepts]

From: Pat Hayes <phayes@ihmc.us>
Date: Thu, 14 Mar 2013 11:39:50 -0500
Message-Id: <D1AB8180-6E15-48F9-9790-61E3862D156F@ihmc.us>
To: RDF Working Group <public-rdf-wg@w3.org>

On Mar 14, 2013, at 4:46 AM, RDF Working Group Issue Tracker wrote:

> RDF-ISSUE-121 (merge-is-union): Should the merge of two RDF graphs be defined as their set union? [RDF Concepts]
> 
> http://www.w3.org/2011/rdf-wg/track/issues/121
> 
> Raised by: Antoine Zimmermann
> On product: RDF Concepts
> 
> (Note: I'm not sure whether it belong to RDF Semantics or RDF Concepts but probably both)
> 
> RDF Semantics is currently proposing to define merge as the set union of RDF graphs. Note that if not all sets of triples are RDF graphs (see ISSUE-120) then union of RDF graphs may not be an RDF graph.
> 
> Defining the merge as the union also means that a set of RDF graphs is not equivalent to its merge.

Which is correct. Consider the following example. I will use { } to indicate graph boundaries. In this example we are operating under the assumption that any set of triples counts as a graph. So there are three graphs here, two of them being subgraphs of the larger one:

{ { :a :p _:x . }
  { :b :p _:x . } }

If we merge the two smaller graphs, what do we get? With the 2004 definition of merge, we don't get the containing graph back: what we get is

{ :a :p _:x .
  :b :p _:x1 . }

So the graph is changed to a non-equivalent graph simply by merging its subgraphs, using the 2004 definition of merge. 

The new definition of merge = union means that merging subgraphs re-creates the original graph, and it coincides with the 2004 definition for graphs that do not share bnodes (which is almost all cases in practice.) It differs only in the case where two graphs (a) share a blank node but also (b) are not considered to be part of any larger or containing graph or structure. But this idea - of blank nodes accidentally being re-used in an unrelated graph - is incoherent and cannot arise in practice. 

Pat



> 
> 
> 
> 

------------------------------------------------------------
IHMC                                     (850)434 8903 or (650)494 3973   
40 South Alcaniz St.           (850)202 4416   office
Pensacola                            (850)202 4440   fax
FL 32502                              (850)291 0667   mobile
phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes
Received on Thursday, 14 March 2013 16:40:28 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 22:02:11 UTC