- From: Stephan Zednik <zednis@rpi.edu>
- Date: Mon, 12 Mar 2012 11:43:10 -0600
- To: Luc Moreau <L.Moreau@ecs.soton.ac.uk>
- Cc: Provenance Working Group WG <public-prov-wg@w3.org>
- Message-Id: <EAA2EA6B-14AC-4C41-AD53-1AAC4F8B812B@rpi.edu>
On Mar 12, 2012, at 11:38 AM, Luc Moreau wrote:
>
> Hi Stephan,
>
> On 03/12/2012 05:32 PM, Stephan Zednik wrote:
>>
>>
>> On Mar 12, 2012, at 11:12 AM, Luc Moreau wrote:
>>
>>> Hi Stephan,
>>>
>>> I repeat what I said before, it's not unreasonable to allow for the plan to be specified and not the agent
>>> (with the understanding that the agent exists, but has not be asserted).
>>
>> Perhaps we are closer than we thought.
>>
> Good!
>> I agree with this statement, and always have. I believe my original argument was precisely this. The plan is still being adopted/followed by ~some~ agent, even if we don't specify any information about that agent. This is an existential qualifier, and would be modeled using the owl:someValuesFrom value constraint in OWL, but I am not sure how to say it in PROV-N.
>>
>> Is there a way to make an existential quantification using PROV-N? I did not connect setting agent to optional with the statement you make above and which I believe we both agree with.
>>
>> Perhaps some confusion arises from my assumption that PROV-N was treating the model as if it were closed world. Why else would you put 'optional' on attributes when this is the default in OWA?
>>
>
> PROV-N is a concrete syntax for a data model. The term 'optional' applies to the syntax.
>
> For the semantics, in part II of prov-dm, we can say that:
>
> if wasAssociatedWith(a,-,pl) holds that there exists an agent ag, such that wasAssociatedWith(a,ag,pl) holds.
>
> We write the existential quantifier in the underpinning rules, not in the syntax.
>
> Does it help?
So if we set it to 'optional' in PROV-N, and use owl:someValuesFrom in PROV-O, is everything good?
If so I think I am happy with this issue being resolved.
--Stephan
>
> Cheers,
> Luc
>
>> --Stephan
>>
>>> It's a *convenience short-cut*,
>>> it does not change the intention of the ActivityAssociation.
>>>
>>> Why should we introduce an extra relation:
>>> hadPlan(id,a,p,-,attr)
>>> when
>>> wasAssociatedWith(id,a,-,p,attr)
>>> can do it?
>>>
>>> To some extent, this applies to all optional arguments of prov-dm,
>>> it was indicated today in the prov-o call, that the prov-rdf translation assumes that
>>> the corresponding line(s) have to be dropped for missing arguments.
>>> (If I understood correctly)
>>>
>>> Luc
>>>
>>> On 03/12/2012 04:30 PM, Stephan Zednik wrote:
>>>>
>>>>
>>>> On Mar 12, 2012, at 6:01 AM, Luc Moreau wrote:
>>>>
>>>>>
>>>>> Hi Stephan
>>>>>
>>>>> As far as I understand, the ontology defines the class Association and the properties
>>>>> agent: AgentInvolvement -> Agent
>>>>> hadPlan: Association -> Plan
>>>>>
>>>>> I don't think see any minimum cardinality required here.
>>>>> So it appears the ontology allows for
>>>>>
>>>>> :a1 prov:hadQualifiedAssociation [a prov:Association
>>>>> hadPlan :pl1]
>>>>>
>>>>> Why can't we allow it in prov-dm?
>>>>
>>>> PROV-O cardinality may not be completely aligned with the DM at present. I suggest we review this once we have a forward direction on this.
>>>>
>>>> In DM:
>>>>
>>>> Why make Agent optional in an Activity-Agent Relation?
>>>>
>>>> The plan is currently qualifying information about the relationship between an activity and the associated agent.
>>>>
>>>> plan: an optional identifier for the plan adopted by the agent in the context of this activity;
>>>>
>>>> By the current definition in DM it does not make sense to have a plan in the association ("... plan adopted by the agent in the context of this activity") without a corresponding agent.
>>>>
>>>> Does an Activity-Agent relation make sense with no agent?
>>>>
>>>> If we are intent to associate a plan to an activity directly, then we should use a specialization of Entity-Agent Relation in which agents are optional and are used to qualify the relationship between the activity and plan. This may be the only way forward.
>>>>
>>>> hadPlan(id,a,p,ag,attr)
>>>>
>>>> --Stephan
>>>>
>>>>>
>>>>> Luc
>>>>>
>>>>>
>>>>>
>>>>> On 03/09/2012 08:47 PM, Stephan Zednik wrote:
>>>>>>
>>>>>> Did you get my last email on this?
>>>>>>
>>>>>> The email with:
>>>>>>
>>>>>> "1) If we take an open world view, then I don't think there is an issue where we know that a specific plan was adopted by an otherwise unknown agent. We can represent the agent, we just won't have any characterizing information about the agent except that it was the agent that adopted this specific plan in this activity.
>>>>>>
>>>>>> 2) If that is not desirable, I suggest adding an Activity-Entity Relation to link Plans to Activities with optional information about which Agent(s) used the plan."
>>>>>>
>>>>>> It appears some of my emails weren't going out for a while, so you may never have gotten it. The email never showed up on the list, and I never got a reply so I am not sure you have seen it.
>>>>>>
>>>>>> I my preference is 1) and I expect your preference is 2), but I think having an Agent-Activity Association without an Agent will be confusing and it goes against the current definition of the relation. If a pure Plan-Activity relation is desired we may have to just mint a new relation.
>>>>>>
>>>>>> --Stephan
>>>>>>
>>>>>> On Mar 8, 2012, at 10:20 PM, Luc Moreau wrote:
>>>>>>
>>>>>>> Hi Tim,
>>>>>>> Yes, see my non-converging discussion with Stephan on ISSUE-203.
>>>>>>>
>>>>>>> Professor Luc Moreau
>>>>>>> Electronics and Computer Science
>>>>>>> University of Southampton
>>>>>>> Southampton SO17 1BJ
>>>>>>> United Kingdom
>>>>>>>
>>>>>>> On 9 Mar 2012, at 02:22, "Timothy Lebo" <lebot@rpi.edu> wrote:
>>>>>>>
>>>>>>>>
>>>>>>>> On Mar 8, 2012, at 6:01 AM, Stian Soiland-Reyes wrote:
>>>>>>>>
>>>>>>>>> I'm not sure - now we can't have a planned activity without an agent -
>>>>>>>>> so there will be phantom agents appearing.
>>>>>>>>
>>>>>>>> ^^ is there a separate issue for this? It seems odd that an Activity can't mention a plan without also bringing an Agent to the game.
>>>>>>>>
>>>>>>>> -Tim
>>>>>>>>
>>>>>>>>>
>>>>>>>>> There is also no way to say that the associated agent is actually
>>>>>>>>> *performing* the activity. And so we only know that an agent performed
>>>>>>>>> something with relation to the activity, and that something might or
>>>>>>>>> might not have been following the associated plan.
>>>>>>>>>
>>>>>>>>> These are DM issues, though.. so you can close this issue. I would
>>>>>>>>> have to think of a good use-case of a plan/recipe which there is no
>>>>>>>>> agent following - perhaps that's not possible?
>>>>>>>>>
>>>>>>>>> On Mon, Mar 5, 2012 at 15:41, Daniel Garijo
>>>>>>>>> <dgarijo@delicias.dia.fi.upm.es> wrote:
>>>>>>>>>> Hi Stian,
>>>>>>>>>> this issue is still raised and pending review.
>>>>>>>>>> now we have Plans to link an agent and a plan to an activity,
>>>>>>>>>> with an Association.
>>>>>>>>>>
>>>>>>>>>> I think that we have addressed this issue, and it could be closed. Thoughts?
>>>>>>>>>> Thanks,
>>>>>>>>>> Daniel
>>>>>>>>>>
>>>>>>>>>> 2011/9/28 Myers, Jim <MYERSJ4@rpi.edu>
>>>>>>>>>>
>>>>>>>>>>> I don’t know that it’s a big deal, but I think of hadRecipe as potentially
>>>>>>>>>>> very indirect rather than a subclass of used. I’d like to assert that the
>>>>>>>>>>> “software development” PE was intended to satisfy the plan as documented in
>>>>>>>>>>> “Work Breakdown Structure element 2.7” but in a use case like that, it seems
>>>>>>>>>>> a stretch to say the PE used the plan versus that I’m just asserting that
>>>>>>>>>>> the PE was intended to fulfill the plan (perhaps just the selection of this
>>>>>>>>>>> PE versus another one was affected by the plan and, after the selection of
>>>>>>>>>>> the PE, the plan was not directly used to guide it, etc.).
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> Jim
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> From: public-prov-wg-request@w3.org [mailto:public-prov-wg-request@w3.org]
>>>>>>>>>>> On Behalf Of Jim McCusker
>>>>>>>>>>> Sent: Wednesday, September 28, 2011 11:21 AM
>>>>>>>>>>> To: Stian Soiland-Reyes
>>>>>>>>>>> Cc: Paolo Missier; public-prov-wg@w3.org
>>>>>>>>>>> Subject: Re: PROV-ISSUE-102 (hadRecipe): Ontology is missing recipe link
>>>>>>>>>>> [Formal Model]
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> If we do adopt a hadPlan/hadRecipe property, it should be a subproperty of
>>>>>>>>>>> used. In which case, if the plan/recipe had a class of Recipe/Plan already
>>>>>>>>>>> (this is a role for an entity, by the way), then why do we need anything
>>>>>>>>>>> other than used?
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> Jim
>>>>>>>>>>>
>>>>>>>>>>> On Wed, Sep 28, 2011 at 8:45 AM, Stian Soiland-Reyes
>>>>>>>>>>> <soiland-reyes@cs.manchester.ac.uk> wrote:
>>>>>>>>>>>
>>>>>>>>>>> On Wed, Sep 28, 2011 at 11:11, Paolo Missier <Paolo.Missier@ncl.ac.uk>
>>>>>>>>>>> wrote:
>>>>>>>>>>>> I recall a discussion with example as part of ISSUE-95 (now part of
>>>>>>>>>>>> formal
>>>>>>>>>>>> model): http://www.w3.org/2011/prov/track/issues/95
>>>>>>>>>>>> isn't that thread relevant?
>>>>>>>>>>>
>>>>>>>>>>> It is marked as relevant, but from the discussion it seems to still
>>>>>>>>>>> rely on "hadRecipe" to say that a plan existed. Using that plan as a
>>>>>>>>>>> class as well merely adds information, such as what kind of attributes
>>>>>>>>>>> you could expect to find, or the hint that it *did* go according to
>>>>>>>>>>> the plan.
>>>>>>>>>>>
>>>>>>>>>>> I get the feeling that ISSUE-95 is slightly controversial as it relies
>>>>>>>>>>> on some OWL2 semantics, but that we are generally positive, however
>>>>>>>>>>> the formal model as it stands does have a recipe as a simple link, and
>>>>>>>>>>> I don't think this ISSUE-102 should be controversial or be much in
>>>>>>>>>>> conflict with ISSUE-95.
>>>>>>>>>>>
>>>>>>>>>>> I have therefore put prov:hadRecipe into
>>>>>>>>>>>
>>>>>>>>>>> http://dvcs.w3.org/hg/prov/raw-file/default/ontology/ProvenanceFormalModel.html#hadrecipe
>>>>>>>>>>> - we can then later fill in what that blank resource is if we go for
>>>>>>>>>>> ISSUE-102 - or remove it if 102 finds a better approach.
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> We can argue about the name in this thread - recipe/plan, etc..
>>>>>>>>>>>
>>>>>>>>>>> --
>>>>>>>>>>> Stian Soiland-Reyes, myGrid team
>>>>>>>>>>> School of Computer Science
>>>>>>>>>>> The University of Manchester
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> --
>>>>>>>>>>> Jim McCusker
>>>>>>>>>>> Programmer Analyst
>>>>>>>>>>> Krauthammer Lab, Pathology Informatics
>>>>>>>>>>> Yale School of Medicine
>>>>>>>>>>> james.mccusker@yale.edu | (203) 785-6330
>>>>>>>>>>> http://krauthammerlab.med.yale.edu
>>>>>>>>>>>
>>>>>>>>>>> PhD Student
>>>>>>>>>>> Tetherless World Constellation
>>>>>>>>>>> Rensselaer Polytechnic Institute
>>>>>>>>>>> mccusj@cs.rpi.edu
>>>>>>>>>>> http://tw.rpi.edu
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> --
>>>>>>>>> Stian Soiland-Reyes, myGrid team
>>>>>>>>> School of Computer Science
>>>>>>>>> The University of Manchester
>>>>>>>>>
>>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>
>>>>>
>>>>> --
>>>>> 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
>>>>>
>>>>
>>>
>>> --
>>> 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
>>>
>>
>
> --
> 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
Received on Monday, 12 March 2012 17:43:45 UTC