Re: ISSUE-111: Create properties prov:hadLocation and prov:assumedBy

On Mon, Oct 3, 2011 at 19:05, Jim McCusker <mccusj@rpi.edu> wrote:
> I'm not 100% sure that we need the prov:wasAssumedBy property. Assumption of
> a Role is only one kind of contextualization that can occur in a provenance
> document. AliceAsAuthor (with type Author) is simply a contextualization of
> Alice. The rest of these are done using complementOf. I think that we should
> explicitly say that role assignment is an intended use of complementOf, but
> having a separate property is redundant, especially if we are trying to make
> a consistent representation that requires minimal inference to understand.

I would agree if it was not that we also have the "Role" concept in
PROV-DM and other potential serialisation. If it's just snuck in as an
entity subclass :Author and only have a weaker prov:wasComplementOf
linking to the "real" entity , then we can't easily extract out again
neither the entity nor the "role" if we were to go PROV-O -> PROV-DM.

(If :Author is a subclass of prov:EntityInRole - then you could  find
all the classes of an entity which are also a subclass of
prov:EntityInRole, but if the roled entity is a complement of 5 other
entities, you can't pick one of them to put into the PROV-DM
statements)


We did in the telcon yesterday and last week agree that we'll try this
explicit approach with EntityInRole and prov:wasAssumedBy for now.

I think the jury is still out on prov:assumedRole vs. just subclassing.

-- 
Stian Soiland-Reyes, myGrid team
School of Computer Science
The University of Manchester

Received on Tuesday, 4 October 2011 09:16:50 UTC