Re: PROV-ISSUE-253: misc issues with the ontology [mapping prov-dm <-> prov-o]

Hi Paul,

Yes, possibly, but the question is as what.
I think there is some value in having the term used to replace 'relation' for instance.
This does not make them a construct/record/expression of the data model.

In fact, I would like to see some use cases, demonstrating the usefulness of such constructs, before incorporating them.

Professor Luc Moreau
Electronics and Computer Science
University of Southampton
Southampton SO17 1BJ
United Kingdom

On 24 Feb 2012, at 19:21, "Paul Groth" <p.t.groth@vu.nl<mailto:p.t.groth@vu.nl>> wrote:

If interaction/involvement is appearing in the semantics, I wonder if "back porting" it into the dm is useful.

@tim how much would this solve in terms of the owl-rl / bidirectional issues in the ontology ?

Paul

On Feb 24, 2012, at 19:45, James Cheney <jcheney@inf.ed.ac.uk<mailto:jcheney@inf.ed.ac.uk>> wrote:


On Feb 24, 2012, at 3:55 AM, Timothy Lebo wrote:


On Feb 23, 2012, at 1:24 PM, Stephan Zednik wrote:



For the domain, Association and Delegation are different from other Involvements in that they are not expected to have time information, is this because we do not view them as instantaneous events?


I'd be curious to hear an answer to this.



We don't have instantaneous event, or event, as a class in the ontology, but it is mentioned in the text of the DM on several occasions.  It seems that some involvements are viewed as events, and others are not.

An instantaneous event class would be a logical domain for occurredAt, and provide a means to separate those involvements we view as events (Generation, Usage, Start, etc)

Added Start and End to be subclasses of Timed.
http://dvcs.w3.org/hg/prov/diff/0318fa4959d1/ontology/ProvenanceOntology.owl

from those we do not (Association and Delegation).  I am not sure if such a class would be a subclass of Involvement or not.

I've advocated to keep it outside of Involvement. It aligns with your "logical domain of occurredAt" notion above and allows us to state which belong in that domain.

 I would argue that some involvements are events, and not all events are involvements, so the classes should be kept separate in the heirarchy, but Generation, Usage, etc can be a subclass of both.

Done for Generation, Usage, Start, and End.


3) I suggest discussing inclusion of a InstantaneousEvent class which will be the domain of occurredAt and a superclass to Involvements that have time information.


InstantaneousEvent is way better than "Timed". Done.
http://dvcs.w3.org/hg/prov/diff/f2e4afae422c/ontology/ProvenanceOntology.owl


Concerning the above questions, please have a look at the formal semantics draft I uploaded yesterday:

http://www.w3.org/2011/prov/wiki/FormalSemanticsWD3

Currently, what I call Interactions include Events, Associations and Derivations.  (I think these correspond to Involvements in the ontology, but am not sure).  Only Events are instantaneous.  Associations include start and end events (i.e. Association and Event can overlap).

I haven't been able to follow the ontology changes over the last week so it is possible what I did is either redundant or completely incompatible.

--James
<Edinburgh University charitable status>