Re: PROV-ISSUE-29 (mutual-iVP-of): can two bobs be mutually "IVP of" each other [Conceptual Model]

On Thu, Sep 29, 2011 at 16:02, Khalid Belhajjame
<Khalid.Belhajjame@cs.man.ac.uk> wrote:

>> Or can it be on entities? We could offer the option to express the
>> characterization interval in the entity expression.
>
> Yes, I would go for this option, we can call it validity interval.

Although I would appreciate such a notion on entities, since the
conceptual model can only deal with "regular" xsd:dateTime times, it
is a much weaker notion of overlap. You don't know that the time
measured for both entities are done in the same way, same accuracy,
etc. You are also limited to precission. If you say a starts at
14:04:00 and b at 14:04:00, and b wasComplementOf B, then you still
don't know if both A and B existed at 14:04:00.002.

In some cases the asserter will  know for a fact that a
wasFullyContainedWithin b (because B is a Luc-in-his-lifetime-like
entity) - but don't know the start- and/or stop-time, he is not able
to state this fact.

You could use the start/stop times if you made them abstract objects
which you could then relate to each-other. You can then say that (in
OWL/Turtle):

:a a prov:Entity .
:b a prov:Entity .

:a prov:wasComplementOf :b .

:a prov2:startedAt :t0 .
:b prov2:startedAt :t0 .

:a prov2:endedAt :t1 .
:b prov2:endedAt :t2 .

:t0 a prov:Time ;
  prov:time "2011-02-23T23:23:12Z"^^xsd:dateTime ;
  application:precision "P3S"^^xsd:duration ;
  prov:followedBy :t1 .

:t1 a prov:Time ;
  prov:followedBy :t2 .

:t2 a prov:Time .


Here we don't know the end-times t1 and t2 - but they have been
asserted with temporal relationships so that we can infer that :b
existed for the whole lifetime of :a - because they have the same
start time :t0 (by owl:equals - we might need a weaker property
prov:sameTimeAs - I suspect existing time/event ontologies have got
this sorted already)  and  a's end-time :t2 is before b's end-time
:t2.

You would have to be quite hardcore to work out that over multiple
times in your SPARQL though (although you can get some help from
transitivity), so it does not remove the question of if a property of
containment would be useful.


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