Re: PROV-ISSUE-126: Section 5.3.3.2 "Process Execution Independent Derivation Expression." [Data Model]

On Nov 8, 2011, at 10:56 AM, Luc Moreau wrote:

> Hi Satya,
> 
> Responses interleaved.  I propose to close the issue, let me know if it shouldn't be the case.
> The recent proposal that was circulated will not use the heavyweight terminology pe-linked/pe-independent.

It was difficult for me to grasp the "pe-linked" naming scheme in the DM, so I'm glad that it is being replaced.

What is the new terminology? "Activity Linked" ? I think this is more natural and like the change.

The anchors still reflect the old terminology.
e.g.
http://dvcs.w3.org/hg/prov/raw-file/default/model/ProvenanceModel.html#pe-linked-derivationRecord

so does the ASN:

pe-linked-derivationRecord:= wasDerivedFrom ( identifier , identifier [, identifier , generationAttributeValues , useAttributesValues] )

Thanks,
Tim




> 
> 
> On 16/10/2011 01:04, Provenance Working Group Issue Tracker wrote:
>> PROV-ISSUE-126: Section 5.3.3.2 "Process Execution Independent Derivation Expression." [Data Model]
>> 
>> http://www.w3.org/2011/prov/track/issues/126
>> 
>> Raised by: Satya Sahoo
>> On product: Data Model
>> 
>> Hi,
>> This is a review comment for Section 5.3.3.2 "Process Execution Independent Derivation Expression." in the PROV-DM document (in mercurial fpwd head on Oct 15, 2011).
>> 
>> Issue: The current definition for "Process Execution Independent Derivation Expression." Section 5.3.3.2 states that:
>> "A process execution independent derivation expression is a representation of a derivation, which occurred by any means whether direct or not, and regardless of any activity in the world."
>> 
>> a) Does the above definition mean that an Entity instance e1 can be derived from another Entity instance e2 without the existence of "transformed from, created from, or affected by" activities?
>>   
> Ativities may or they may not exist.  We don't say anything about them, and we are not trying to link the derivation with any activity/activities.
>> b)  If the above definition just means that there exists some PE linked to the derivation of e2 from e1, but a provenance application may not be aware of it, then how does it relate to the constraint "derivation-process-execution" defined for "Process Execution Linked Derivation Assertion" in Section 5.3.3.1?
>> 
>>   
> 
> No, that's not the intent. If you know there is one PE, and you don't know about it, wasDerivedFrom(e2,e1) is exactly capturing this notion.
>> The current definition of "wasDerivedFrom" states that there was an activity of "transformed from, created from, or affected by" that links the two Entity instances, which is *summarized* by the wasDerivedFrom property. Hence, "Process Execution Independent Derivation Expression" is not consistent with current definition of derivation.
>> 
>> 
>> 
>>   
> wasDerivedFrom is pe-linked, and PE independent derivation are not PE-linked. Idont' think there is any inconsistency here.
> 
> Luc
> 
> 
> 

Received on Wednesday, 16 November 2011 17:24:44 UTC