Re: PROV-ISSUE-647 (TomDN): Make prov:pariValue a subproperty of prov:entity? [PROV-DICTIONARY]

Tom,

Apologies for rocking the boat with my off-list comment.

On Mar 22, 2013, at 8:49 AM, Tom De Nies <tom.denies@ugent.be> wrote:

> Luc raised some concerns about making prov:pairValue a sub-property of prov:entity in yesterday's telecon.
> If we decide to make prov:pairValue a sub-property of prov:entity, that would imply that prov:pairValue now has the domain prov:EntityInfluence. 

This makes sense from the qualification perspective, since the KeyValuePair is adding the detail of some "key" for some existing prov:hadMember Entity "value".

The Entity :bar existed just fine on its own, then when some Dictionary decided to come along and shove it into some "key bin" called "foo", the KeyValuePair is the (membership) qualification for how the Entity :bar influenced the Dictionary (and also includes the key used: "foo").


> Would this mean that we have to make prov:KeyValuePair a subclass of prov:EntityInfluence as well?

Yup.

> This seems weird and counter-intuitive to me.

When starting with Entity :bar and wanting to move to a KeyValuePair to place it into "key bin foo", then yes, it can seem *un*intuitive (though, not sure about *counter* intuitive).
But, when you make a KeyValuePair, you're implying some Dictionary -- and you've influenced that Dictionary by placing a new Entity into it.
The Entity influenced the Dictionary by becoming its member, with the additional detail of the key.

^^ EntityInfluence, Dictionary, prov:hadMember, KeyValuePair

> It would imply that a dictionary would have some influence on all its members.
> 

Other way around. The Entities placed into the Dictionary influenced the Dictionary.


-Tim

> Tim, could you share your views on this?
> 
> Regards,
> Tom
> 
> 2013/3/7 Provenance Working Group Issue Tracker <sysbot+tracker@w3.org>
> PROV-ISSUE-647 (TomDN): Make prov:pariValue a subproperty of prov:entity? [PROV-DICTIONARY]
> 
> http://www.w3.org/2011/prov/track/issues/647
> 
> Raised by: Tom De Nies
> On product: PROV-DICTIONARY
> 
> Came up in an off-list conversation with Tim about the PROV-O of dictionaries. It appears to be useful to make prov:pairValue a subproperty of prov:entity. This way applications could use spec-level constructs to "accidentally" "understand" part of the "brand new construct".
> 
> Nice phrasing of the rationale by Tim:
> "Having prov:pairValue is a very nice subproperty for these uninterested in the alignment with qualifications, but still provides those that do care about qualifications a treat."
> 
> I see no real problems with adding this for the next release. Is this acceptable to the group or did we miss some consequences?
> 
> 
> 
> 
>