Re: PROV-ISSUE-455 (Dong): Type mismatch between PROV-O and PROV-DM [Ontology]

For the record:


http://www.w3.org/2011/prov/meeting/2012-07-26#PROV__2d_O_PrimarySource

Summary: The group decided that there was a bug in the naming of prov:Source in the ontology. The group resolved to name it prov:PrimarySource, in line with prov-dm. (Likewise, for prov:qualifiedSource which becomes prov:qualifiedPrimarySource.) The change will be implemented immediately.


-Tim


On Jul 24, 2012, at 8:34 AM, Timothy Lebo wrote:

> prov-wg,
> 
> On Jul 24, 2012, at 1:55 AM, Stephan Zednik wrote:
> 
>> The DM type of a primary source derivation would equate to the type of the qualified class for hadPrimarySource, correct?  The DM type used with the Revision and Quotation relations is the same term used in the PROV-O qualified classes for the relations.
>> 
>> I think PROV-O can and should use the term prov:PrimarySource for the qualified class associated with prov:hasPrimarySource.
>> 
> 
> 
> 
> 
>> I would recommend renaming prov:Source to prov:PrimarySource and prov:qualifiedSource to prov:qualifiedPrimarySource.
>> 
> 
> This is much more consistent naming, and I like it for that.
> I'd like to get a sense of consensus on this change from the WG, can we straw poll (or, just propose/resolve) this in this week's call?
> 
> 
>> I am curious why the current prov:Source is currently not a subclass of prov:Derivation.  Did we decide to have a flat hierarchy under the Agent/Entity/ActivityInfluence classes?
> 
> We did. The only distinction maintained in the Influence hierarchy is the type of influencer. 
> Specialization in the property hierarchy is used to establish Source, Quotation, and Revision as specializations of Derivation.
> 
>> 
>> Tim's concern of possible user confusion of prov:PrimarySource as an prov:Entity is justified.  I think we may get similar confusion with prov:Revision and prov:Quotation.  We may have to rely on the ontology class and property hierarchies, ontology annotations, and WG generated documents and primer to avoid this confusion.
>> 
>> ... so I guess I am punting on that issue for now.
> 
> 
> It seems as though we should just let the possibility stand.
> 
> Regards,
> Tim
> 
> 
>> 
>> --Stephan
>> 
>> On Jul 18, 2012, at 8:46 AM, Timothy Lebo wrote:
>> 
>>> Dong,
>>> 
>>> On Jul 18, 2012, at 10:45 AM, Huynh T.D. wrote:
>>> 
>>>> Thanks for the clarification, Tim.
>>>> I don’t mind using either prov:Source or prov:PrimarySource, but not both for the same relation.
>>> 
>>> I agree.
>>> 
>>>> I guess this is a negotiation to be had between the editors of PROV-O and PROV-N to resolve the mismatch.
>>> 
>>> I agree.
>>> 
>>> Regards,
>>> Tim
>>> 
>>> 
>>>> Best wishes,
>>>>  
>>>> Dong.
>>>>  
>>>> From: Timothy Lebo [mailto:lebot@rpi.edu] 
>>>> Sent: 18 July 2012 14:03
>>>> To: Provenance Working Group
>>>> Subject: Re: PROV-ISSUE-455 (Dong): Type mismatch between PROV-O and PROV-DM [Ontology]
>>>>  
>>>> Dong,
>>>>  
>>>> PROV-O has avoided calling the class PrimarySource to avoid the very likely situation where a modeler will assume it to be a subclass of Entity (i.e., the object of the hadPrimarySource property).
>>>>  
>>>>  
>>>> On Jul 18, 2012, at 8:50 AM, Provenance Working Group Issue Tracker wrote:
>>>> 
>>>> 
>>>> PROV-ISSUE-455 (Dong): Type mismatch between PROV-O and PROV-DM [Ontology]
>>>> 
>>>> http://www.w3.org/2011/prov/track/issues/455
>>>> 
>>>> Raised by: Trung Dong Huynh
>>>> On product: Ontology
>>>> 
>>>> In PROV-DM and PROV-N, prov:PrimarySource is used as the type for the hadPrimarySource relation
>>>>  
>>>>  
>>>> DM says:
>>>>  
>>>> "A primary source ◊ for a topic"
>>>>  
>>>> The essential bit here is "for a topic", not "in all cases". This makes it a relation and not a class.
>>>>  
>>>> The DM is clear about this distinction:
>>>> "A primary source ◊ relation is a particular case of derivation of secondary materials from their primary sources."
>>>>  
>>>> So, the name of the _concept_ is just fine.
>>>>  
>>>>  
>>>> That leaves only PROV-N, which levels the playing field a bit.
>>>> From the PROV-N perspective, it is natural to "type" the derivation relation. PROV-N does not have the class hierarchy mechanism like OWL does, so it does not face the same naming challenge that I described above.
>>>>  
>>>> while PROV-O currently uses prov:Source (http://dvcs.w3.org/hg/prov/raw-file/default/model/prov-dm.html#cross-references-to-prov-o-and-prov-n).
>>>> I think there should not be such a mismatch given that both the documents share the same namespace.
>>>>  
>>>> Agreed. A mismatch is unacceptable, I think.
>>>>  
>>>> -Tim
>>> 
>> 
>