Re: PROV-ISSUE-429 (hasAnchor-contextualization): how does hasAnchor relate to contextualization [Accessing and Querying Provenance]

As I've already indicated, at this stage, I strongly oppose the inclusion of 
contextualization in the DM/Ontology specs.  BUT...

To the extent that contextualization is currently indistinguishable from 
specialization, the question might be: "Is prov:hasAnchor equal to or a 
subproperty of specialization?"

I think this is a good question.  prov:hasAnchor [2] was introduced to mirror 
the capability of the anchor parameter in an HTTP Link: element with 
rel=provenance [1].  In RDF usage, the hasAnchor could probably be replaced by 
some prov:specializationOf assertions between the subject of hasProvenance and 
the object of hasAnchor, but I'm not certain that *all* such uses could be 
replaced.  And even in scenarios where the replacement is possible, it's not 
obvious to me whether hasAnchor should be super- or sub- property.

Consider a "common" use-case:

C: GET http://example.com/weatherreport/today/

S: @prefix prov: <http://www.w3.org/ns/prov#>
S:
S: <> dcterms:title      "Welcome to example.com" ;
S:    prov:hasAnchor     <http://example.com/weather/20120621/> ;
S:    prov:hasProvenance <http://example.com/weather/provenance/20120621/> ;
S:     :
S:    (RDF data)
S:     :

So in this case, we would have:

<http://example.com/weather/20120621/>
   :specializationOf <http://example.com/weatherreport/today/>

But equally, we might see

C: GET http://example.com/weatherreport/20120621/

S: @prefix prov: <http://www.w3.org/ns/prov#>
S:
S: <> dcterms:title      "Welcome to example.com" ;
S:    prov:hasAnchor     <http://example.com/weather/today/> ;
S:    prov:hasProvenance <http://example.com/weather/provenance/all/> ;
S:     :
S:    (RDF data)
S:     :

where <http://example.com/weather/provenance/all/> might be providing 
information that *all* daily weather reports at http://example.com/ are 
published by (say) the owner of example.com.  The same specialization relation 
as above would still hold, but the prov:hasAnchor use is the other way round.

Thus, I conclude that prov:hasAnchor does not provide *any* provenance-related 
semantics.  It is available for use by applications as a hint for locating 
appropriate information, but is not necessary if the relationship between terms 
used in the provenance resource and the requested resource are already understood.

Now that the spec contains specializationOf properties, which in RDF can be used 
to convey information for which hasAnchor might previously have been used, I 
would be more inclined to drop the hasAnchor property altogether.

#g
--

[1] http://www.w3.org/TR/prov-aq/#resource-accessed-by-http
[2] http://www.w3.org/TR/prov-aq/#resource-represented-as-rdf



On 21/06/2012 00:10, Provenance Working Group Issue Tracker wrote:
> PROV-ISSUE-429 (hasAnchor-contextualization): how does hasAnchor relate to contextualization [Accessing and Querying Provenance]
>
> http://www.w3.org/2011/prov/track/issues/429
>
> Raised by: Paul Groth
> On product: Accessing and Querying Provenance
>
> Is hasAnchor equal to or a subproperty of contextualization?
>
>
>
>

Received on Thursday, 21 June 2012 11:26:05 UTC