- From: Timothy Lebo <lebot@rpi.edu>
- Date: Thu, 23 Feb 2012 08:52:40 -0500
- To: Paul Groth <p.t.groth@vu.nl>
- Cc: "public-prov-wg@w3.org" <public-prov-wg@w3.org>
Thanks, Paul.
We should also note that we are inverting one of the properties used to reference the components of the triple that we are reif.... er... qualifying.
I'm rather disappointed that they are encouraging the use of blank nodes. All they had to do was remove an underscore!
I also like that they named the referencing predicate according to the role.
In their example:
eg:bob a foaf:Person.
eg:mary a foaf:Person.
_:bobMaryMarriage a ex:Marriage;
ex:partner eg:bob;
ex:partner eg:mary;
ex:date "2009-04-01"^^xsd:date.
the reference to bob can be "inverted":
eg:bob a foaf:Person;
ex:married eg:mary;
prov:qualified :bobMaryMarriage;
.
:bobMaryMarriage # proper URI instead of bnode!
a ex:Marriage;
ex:partner eg:mary; # Reference to :bob is removed, because :bob is referring to the Marriage.
ex:date "2009-04-01"^^xsd:date
.
eg:mary a foaf:Person.
Although this looks goofy for marriage (which treats partners equally), PROV has a bias towards "newer" things as being the subjects of the involvements.
The Marriage they model makes sense for symmetric relations. PROV involvements are directional.
-Tim
On Feb 23, 2012, at 7:14 AM, Paul Groth wrote:
> Hi All,
>
> I think it would be good to link to the linked data patterns book[1] in the documents around the idea of using the qualified pattern.
>
> cheers,
> Paul
>
> [1] http://patterns.dataincubator.org/book/qualified-relation.html
>
>
Received on Thursday, 23 February 2012 13:53:14 UTC