Re: PROV-ISSUE-66 (is-execution-a-bob): Why is process execution not defined as a characterised entity? [Conceptual Model]

Hi Luc, Jim,

Luc:
> We still haven't defined process ordering, and I would argue that
> ... We haven't had time to define this in spec. Sorry.

OK. I'm not certain having a signal from one execution to another is
the same as one creating another, but am happy to leave this debate if
you think it is not for this draft.

But surely there still needs to be some resolution to the issue
itself? The model implies (or is at least very easy to read as
implying) that a PE is a kind of BOB, by the fact that a PE as defined
fits the definition of a BOB. But if this is not intended, then what
should the OWL file contain to reflect the model accurately? Again, I
believe this was Satya's point, but he should probably clarify
himself.

[ To justify the importance of clarifying this in the model, I was
working with someone last week who was using OPM and had expressed
things in their application as artifacts that I would intuitively
considered processes. They saw the final report on each process as
their only knowledge of it, and those reports are static, so the
processes were modelled as artifacts but with the names and reported
characteristics of the processes. ]

Jim, if I understand your argument correctly, you are saying that:
 - we should redefine PE/BOB to ensure a PE is not a BOB
 - we should define a relation from a PE to its 'substrate(s)', which are BOBs
 - describing the provenance of these substrates (along with
everything else PIL can already say) completes the description of the
provenance of a PE

Is that correct? If so, I can buy that, but I'm not yet clear what
that would look like in the model document, or if you are saying it is
already as such in the document.

Thanks,
Simon

On 1 August 2011 16:38, Myers, Jim <MYERSJ4@rpi.edu> wrote:
> While I suspect that the potential symmetry that Simon is identifying is
> 'there', I would argue against making PE = Bob (or making them
> subclasses of a common ancestor) from two angles:
>
> The most compelling to me is just that I cannot see how we're going to
> explain the model if we argue that what people typically think of as
> objects and what they typical call processes are really not so
> distinguishable. As I'll suggest below, I think we have ways to address
> Simon's use cases if we have IVPof (and maybe participates) and I'm
> hopeful we can argue for them without resorting to endurant/perdurant
> arguments. (If we did make them the same, or subclasses of the same
> thing (and we can find the Bob and PE that correspond via some
> relation(s) in the model) , we would be adding a duplicate mechanism to
> model their provenance...).
>
>
> For those who want to do that, here's my take on the mapping: I think we
> primarily have an endurant(Bob) - perdurant (PE) model - that's the
> used/generatedby core of the model. In the spirit of "The Water Falls
> But The Waterfall Does Not Fall", by Galton and Mizoguchi, though, pil
> recognizes that things we think of as endurants can also have a
> perdurant aspect when we consider other processes. An endurant egg that
> goes into a cake baking process (endurant in the sense that at each
> instant in time, it is indeed an egg), is also participating in a
> 'warming egg' process (perdurant in the sense that at no point in time
> is the egg actually warming, participates implying that the endurant egg
> is part of the substrate for the perdurant process). To model this, we
> have IVPof which allows us to model the warming egg by talking about the
> cold egg (endurant), the heating process (perdurant), and the warm egg
> (endurant). What we don't have in the model is the 'warming egg'
> directly - no way to say energy went into a 'warming egg' without
> identifying the 'heating process' separately and saying energy and the
> 'cold egg' were inputs.
>
> Philosophy aside, I think it is true that IVPof and participates (in the
> sense used here - not sure where the group debate has settled on
> pil:participates) allows us to cover all of the use cases where you
> might want to talk about the perdurant aspects of an endurant Bob or the
> endurant aspects of a perdurant PE. A PE's provenance would be complete
> if one described the provenance of all inputs (used) and the provenance
> of all Bobs that 'participate in'/are the substrate for the PE. A Bob's
> provenance would be complete if one described the PE that generated it
> and all of the PEs it had "participated in".
>
> So - hopefully the discussion here helps. If not, my basic sense is that
>
>
>
>> -----Original Message-----
>> From: public-prov-wg-request@w3.org [mailto:public-prov-wg-
>> request@w3.org] On Behalf Of Luc Moreau
>> Sent: Monday, August 01, 2011 6:50 AM
>> To: public-prov-wg@w3.org
>> Subject: Re: PROV-ISSUE-66 (is-execution-a-bob): Why is process
> execution
>> not defined as a characterised entity? [Conceptual Model]
>>
>> Hi Simon,
>>
>> While I mentioned, endurant/perdurant, I was using them for intuition,
> and
>> didn't mean to formally introduce them in the definition.
>> You may be right, a BOB with temperature changing constantly, may not
> fit
>> the endurant/perdurant distinction. But I am not expert in this
> ontology.
>>
>> But, I don't see why this makes a case for PE=BOB. What use case do
> you try
>> to resolve by merging the two concepts?  Its major downside is the
> unknown
>> meaning of using/generating a process execution.  It seems also to mix
>> use/generation/start/end.
>>
>> I see a process execution as a set of events using/generating BOBs,
> and
>> temporally delimited by a start and end event.  This set of event is
> said  by
>> an asserter to be an identifiable activity.
>> This activity has an "effect" on the world, by all the BOBs it
> generates.
>>
>> So can you identify an example that we can't model with  PE<>BOB but
> can
>> with PE=BOB?
>>
>> Luc
>>
>> On 08/01/2011 11:09 AM, Simon Miles wrote:
>> > Hi Luc,
>> >
>> > OK, I believe I understand your intuition, but would argue:
>> >
>> > 1. A process execution currently fits the definition of a BOB, as it
>> > has assertable characteristics, is identifiable, and is an entity,
>> > i.e. bounded.
>> >
>> > 2. I was not clear we wanted to restrict BOBs to endurants. For
>> > example, at the F2F1, Tim raised the case of expressing that
> something
>> > changed temperature by 10 degrees (and why), without asserting what
>> > the start and end temperatures were. If I remember correctly, you
> and
>> > I argued that the temperature change could be seen as an (invariant)
>> > attribute of a BOB, because BOBs are not necessarily instantaneous.
>> > And if I understand endurants and perdurants correctly, I think the
>> > change is a perdurant because at any moment, the change itself is
> not
>> > apparent.
>> >
>> > Intuitively, it seems reasonable that a process execution is
> something
>> > you can ask the provenance of ("why did this execution occur as it
>> > did?"). This might or might not be fully answered by saying what the
>> > execution used, who it was controlled by, when it started, and what
>> > recipe it followed.
>> >
>> > I recognise that it makes things complicated, in that a process
>> > execution (as a BOB) could be "generated" or "used" by another
>> > execution, but perhaps this is what we mean by the "ordering of
>> > processes" concept anyway. The existing definition "Generation
>> > represents the creation of a new characterized entity by an
> activity.
>> > This characterized entity did not exist before creation." seems to
> fit
>> > the notion of one process starting another one quite nicely.
>> >
>> > To be more constructive, here's my counter-proposal.
>> >
>> > * We add "characterized entity" to the process execution definition,
>> > e.g. "A process execution represents an activity which performs a
>> > piece of work. An activity is a kind of identifiable characterized
>> > entity."
>> >
>> > No other change seems necessary, though some explanation of the
>> > consequences may be needed. We might then have also covered
> "ordering
>> > of processes", at least in a causal sense (temporal ordering is
>> > different).
>> >
>> > Thanks,
>> > Simon
>> >
>> > On 1 August 2011 09:14, Luc Moreau<L.Moreau@ecs.soton.ac.uk>  wrote:
>> >
>> >> Hi Simon,
>> >>
>> >> To me, process executions are the "verbs", whereas BOBs are the
>> >> "nouns", and therefore belong to different categories.
>> >>
>> >> Several people have also mentioned they relate to
> perdurant/endurant
>> >> in formal ontologies.
>> >>
>> >> Being identifiable is therefore not the key characteristic!
>> >>
>> >> Regards,
>> >> Luc
>> >>
>> >>
>> >> PS. In a separate thread, you mentioned that IVPof could be used
> for
>> >> process executions.
>> >>     This may make sense, but in that case we simply need to change
>> >> the signature of IVP of:
>> >>          BOB x BOB   U   PE x PE
>> >>
>> >>
>> >>
>> >> On 07/29/2011 05:22 PM, Provenance Working Group Issue Tracker
>> wrote:
>> >>
>> >>> PROV-ISSUE-66 (is-execution-a-bob): Why is process execution not
>> >>> defined as a characterised entity? [Conceptual Model]
>> >>>
>> >>> http://www.w3.org/2011/prov/track/issues/66
>> >>>
>> >>> Raised by: Simon Miles
>> >>> On product: Conceptual Model
>> >>>
>> >>> This was mentioned by Satya in the call, but I can't see it having
> been
>> raised as an issue yet.
>> >>>
>> >>> As process executions are identified and may have attributes,
> including
>> start and end time, are they kinds of characterised entities,
> similarly to
>> agent? If not, why not?
>> >>>
>> >>>
>> >>>
>> >>>
>> >>>
>> >>>
>> >> --
>> >> Professor Luc Moreau
>> >> Electronics and Computer Science   tel:   +44 23 8059 4487
>> >> University of Southampton          fax:   +44 23 8059 2865
>> >> Southampton SO17 1BJ               email: l.moreau@ecs.soton.ac.uk
>> >> United Kingdom                     http://www.ecs.soton.ac.uk/~lavm
>> >>
>> >>
>> >>
>> >>
>> ______________________________________________________________
>> _______
>> >> _ This email has been scanned by the MessageLabs Email Security
>> >> System.
>> >> For more information please visit http://www.messagelabs.com/email
>> >>
>> ______________________________________________________________
>> _______
>> >> _
>> >>
>> >>
>> >
>> >
>> >
>>
>> --
>> Professor Luc Moreau
>> Electronics and Computer Science   tel:   +44 23 8059 4487
>> University of Southampton          fax:   +44 23 8059 2865
>> Southampton SO17 1BJ               email: l.moreau@ecs.soton.ac.uk
>> United Kingdom                     http://www.ecs.soton.ac.uk/~lavm
>>
>
>
>
> ______________________________________________________________________
> This email has been scanned by the MessageLabs Email Security System.
> For more information please visit http://www.messagelabs.com/email
> ______________________________________________________________________
>



-- 
Dr Simon Miles
Lecturer, Department of Informatics
Kings College London, WC2R 2LS, UK
+44 (0)20 7848 1166

Received on Monday, 1 August 2011 16:19:26 UTC