Re: Today's prov-o team telecon

Luc (and any logic people),


On Apr 11, 2012, at 5:49 PM, Luc Moreau wrote:

> 
> Tim
> 
> Thanks, a question below.
> 
> On 11 Apr 2012, at 18:06, "Timothy Lebo" <lebot@rpi.edu> wrote:
> 
>> Luc,
>> 
>> 
>> On Apr 10, 2012, at 3:16 PM, Luc Moreau wrote:
>> 
>>> Hi Tim,
>>> 
>>> Can you clarify the implication of this resolution:
>>>> hadActivity for a Responsibility (Agent->Agent) https://www.w3.org/2011/prov/track/issues/226 ISSUE-226
>>>> PROPOSE: relax domain of hadActivity so that a Responsibility can use it.
>>>> Derivation -> Activity via prov:hadActivity
>>>> RESOLVED: change prov:hadActivity domain from Derivation to Involvement.
>>>> TODO: make clearer in the HTML how to use it. (in 3.3)
>>> 
>>> It now seems that we can have Attribution, Quotation, etc in the domain of hadActivity. This is not in line with DM.
>> 
>> 
>> This will be a persistent tradeoff in RL, and I don't know how to resolve this kind of interpretation. It's simply not what the axioms say. The reasoning you use is incorrect.
>> I've outlined the inferences that one can obtain with rdfs:domain before, and showed that there is not an inconsistency or an inference that does not make sense.
>> 
>> 
>> prov:hadActivity rdfs:domain prov:Involvement .
>> :s prov:hadActivity :a 
>> --------
>> :s a prov:Involvement
>> 
>> 
>> this is different than saying:
>> :q a prov:Quotation
>> -------
>> :q prov:hadActivity :activity
>> 
>> which is NOT derivable in the current ontology, but how you are interpreting it.
>> A class being in the domain of a property does NOT imply that the class uses that property.
>> 
>> The axioms that you're thinking of are minCardinality 1 / someValuesFrom, etc., which WOULD imply that the class has the property.
>> 
>> 
>> 
>> 
>>> 
>>> Also, Generation is in the domain of both activity and hadActivity. Isn't this strange?
>> 
>> resources described with prov:activity are prov:ActivityInvolvements.
>> resources described with prov:hadActivity are prov:Involvements .
>> Being a subclass of either of these classes does not imply that you are described with either of these properties.
> 
> 
> 
> 
> Isn't it the case that the following statements are consistent with the ontology. 
> 
>  :g a Generation
>    prov:activity :a1
>    prov:hadActivity :a2
> 
> :q a Quotation
>    prov:hadActivity :a2
> 


What do you mean by "consistent"?
The only two logical inconsistencies (contradictions) that one can make using PROV-O are to multi-type a resource to be both an Entity and Activity, or both an EntityInvolvement and ActivityInvolvement. This is because these two pairs of classes are disjoint.
Everything else is "fair game" - no logical inconsistencies will result with any other combination of properties and classes.

(anyone with a stronger logic background can feel free to correct me, or help me address Luc's concern)

I think that you're pointing out that we are implicitly assuming the classes to be disjoint without asserting it.
I've avoided asserting too many disjoints in favor of simplicity (and knowing that users can assert it if they need it).

So, it's also "consistent" to say:

:g a prov:Generation;
  prov:used [ a prov:Trace ];
  prov:qualifiedTrace [ a prov:Entity ];
.

Because the range of prov:used is not disjoint with prov:Trace, the range of prov:qualifiedTrace is not disjoint with prov:Entity,
the domain of prov:used is not disjoint with prov:Generation, and the domain of prov:qualifiedTrace is not disjoint with prov:Generation.

I'm sure this isn't the answer you're looking for, but I'm not sure how to resolve your concern (using RL and avoiding an explosion of over constrained constructs).



> 
> I just don't how to understand these, in terms of the DM.


I agree. Having this combination does not make sense. But I don't know how to address it within RL.
In DL, I could say a bunch. But since we're stuck in RL, we're left to only use out-of-band documentation for how the RL constructions _should_ be stitched together b/c we can't use the DL axioms that are needed to say how.

Regards,
Tim


> 
> Luc
> 
> 
> 
>> 
>> 
>> -Tim
>> 
>> 
>> 
>>> 
>>> Professor Luc Moreau
>>> Electronics and Computer Science
>>> University of Southampton 
>>> Southampton SO17 1BJ
>>> United Kingdom
>>> 
>>> On 9 Apr 2012, at 17:33, "Timothy Lebo" <lebot@rpi.edu> wrote:
>>> 
>>>> We had a quick meeting due to the UK holiday.
>>>> 
>>>> Minutes are at http://www.w3.org/2011/prov/wiki/PIL_OWL_Ontology_Meeting_2012-04-09
>>>> 
>>>> Regards,
>>>> Tim
>>>> 
>>>> On Apr 9, 2012, at 10:47 AM, Timothy Lebo wrote:
>>>> 
>>>>> Thanks, Paolo.
>>>>> 
>>>>> It's not a holiday on this side of the pond, but I took the day off myself.
>>>>> 
>>>>> Rest of prov-o, will you be joining?
>>>>> Please email me directly if you'd rather keep this off list.
>>>>> 
>>>>> Regards,
>>>>> Tim
>>>>> 
>>>>> 
>>>>> On Apr 9, 2012, at 10:38 AM, Paolo Missier wrote:
>>>>> 
>>>>>> apologies, not joining, it's a UK holiday today (and I believe this is true at least of the rest of Europe?)
>>>>>> 
>>>>>> -Paolo
>>>>>> 
>>>>>> 
>>>>>> On 4/9/12 3:30 PM, Timothy Lebo wrote:
>>>>>>> prov-o team,
>>>>>>> 
>>>>>>> The agenda for today's telecon is at:
>>>>>>> http://www.w3.org/2011/prov/wiki/PIL_OWL_Ontology_Meeting_2012-04-09#Agenda
>>>>>>> 
>>>>>>> Please be prepared to:
>>>>>>> 1) raise topics from last week's feedback that the team should discuss during meeting
>>>>>>> 2) discuss the ISSUEs listed on the agenda.
>>>>>>> 
>>>>>>> Regards,
>>>>>>> Tim
>>>>>> 
>>>>>> 
>>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>> 
>>>> 
>>