Re: PROV-ISSUE-467 (activity-start-req-trigger): Do activity start/end always require trigger? [prov-dm-constraints]

On Mon, Aug 6, 2012 at 4:53 PM, James Cheney <jcheney@inf.ed.ac.uk> wrote:
> Since you don't give an indication of what would resolve the issue, I'm not sure if this is a problem with PROV-DM, or with PROV-CONSTRAINTS?
>
> If the latter, would making the trigger parameters non-expandable (so that you can write "-" for a missing trigger) fix the problem?  If so, I propose we do so.  (This will also require being careful about these arguments later, as for derivation and association.)

I was just not sure if we had WG consent on the requirement of the
trigger as I don't remember seeing a discussion about this. I think
this is a problem with PROV-Constraints, as Prov-DM does not *require*
the trigger, it is optional there:

>  An end may refer to an entity, known as trigger ◊, that terminated the activity, or to an activity, known as ender ◊ that generated the trigger.
> trigger: an optional identifier (e) for the entity triggering the activity ending;
> ender: an optional identifier (a1) for the activity that generated the (possibly unspecified) entity (e);

which sounds sane to me.


However PROV-Constraints requires the trigger by existentialising (?)
it, which I find slightly odd, because I believe activities could
possibly be started/ended by other ways than triggers (for instance
activities representing phenonema, which just 'are'). Or do we require
this as a mirror to how entities must be generated by an activity, and
bite the bullet with phantom triggers/instructions? If we have to do
this, then I just need to be sure that this is OK even at the time
boundaries of the activity, so it can self-start and self-terminate.


If there are no protests, I would however prefer the - for missing
trigger, but I know this would cause more editorial changes, similar
to the blank "-" for the activity of wasDerivedFrom and plan of
wasAssociatedWith. However that is what I feel is most natural, that
both the trigger and its activity could be - for "unspecified" rather
than "unknown"; for both wasEndedBy and wasStartedBy.


If the inference rule is changed to imply
wasStartedBy(_id1;a,-,-,t1,[])  and similar for wasEndedBy(), then
this might be enough to stop the recursive looping as well. (Check!)



> Are you also arguing for removing the wasEndedBy part of the conclusion?  If so, any objections?

Yes, my argument is that the activity end might or might not be there.
For an activity that as far as the provenance asserter is concerned is
ongoing (which he has no way to say in PROV), this would be in the
future, or it might never happen (runs forever; for instance we don't
know if physics:expansionOfUniverse or bio:life will terminate). We
don't know, and should not then infer it. We can infer that the
activity is still alive at a later point (say because an entity was
generated by the activity), and hence know that there is no wasEndedBy
earlier than that point, but we don't know there will be an
wasEndedBy.


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

Received on Wednesday, 8 August 2012 08:27:53 UTC