Re: PROV-ISSUE-100 (Entity definition): Section 5.2.1 Entity [Conceptual Model]

Hi Luc,
I have raised points in this issue separately reflecting changes in DM
(prior to SPWD). Hence, although some of my concerns in this thread are not
completely resolved we can close this particular issue and focus on more
recently raised issues.

Thanks.

Best,
Satya

On Wed, Nov 30, 2011 at 7:41 AM, Luc Moreau <L.Moreau@ecs.soton.ac.uk>wrote:

> Hi Satya,
>
> Our message to you earlier this month is unanswered. Since then, the
> document has
> substantially evolved. We now have entity and entity records, etc.
> Furthermore, time and events are now fully discussed in section 2.
>
> Since all your concerns have been addressed, I propose to close the issue,
> pending review.
> Can you confirm it is appropriate?
>
> Regards,
> Luc
>
>
> On 11/07/2011 12:05 PM, Luc Moreau wrote:
>
>> Hi Satya,
>>
>> Paolo and I have been discussing your issue 100. (See PM/Luc comments)
>> Some of the recent accepted proposals definitely address your concerns.
>> The recent changes in prov-o also, I think, answer some of your queries.
>>
>> I am not entirely clear what is outstanding in this issue.
>> Can you clarify for us? Can we close the issue?
>>
>> Cheers,
>> Luc
>>
>>
>> > My review comments for Section 5.2.1 Entity in the current version of
>> the conceptual model document:
>> >
>> > 1. In PROV-DM, an entity expression is a representation of an
>> > identifiable characterized thing.
>> >
>> > Issue: Since the section heading is for Entity and the PROV DM
>> > component is Entity, I am confused why we are defining "Entity
>> > Expression" and not "Entity"?
>>
>> An instance of an entity expression is syntactally written as
>> 'entity', but we use the term 'entity record' (was 'entity
>> expression') to make it clear that we refer to a PROV-DM construct and
>> not a thing in the world (see intro of section 5.1)
>>
>> We also tried to make the distinction between the model and language
>> clearer.
>> So, we use 'records' to talk about prov-dm.
>>
>> In section 2.2, we now write:
>>
>>  This specification also relies on a language, PROV-ASN, the
>>  Provenance Abstract Syntax Notation, to express instances of that
>>  data model. For each construct of PROV-DM, a corresponding ASN
>>  expression is introduced, by way of a production in the ASN grammar.
>>
>>
>> >
>> > 2. An instance of an entity expression, noted entity(id, [ attr1=val1,
>> > ...]) in PROV-ASN contains an identifier id identifying a
>> > characterized thing; contains a set of attribute-value pairs [
>> > attr1=val1, ...], representing this characterized thing's situation in
>> > the world.
>> >
>> > Issue: When we refer to an entity in provenance assertions (in
>> > different applications), do we use the identifier to refer to it or
>> > both identifier + attribute-value pairs?
>>
>> Not sure I understand (in different applications).
>>
>> We refer to it with its identifier.
>>
>>
>> >
>> > 3. The assertion of an instance of an entity expression, entity(id, [
>> > attr1=val1, ...]), states, from a given asserter's viewpoint, the
>> > existence of an identifiable characterized thing, whose situation in
>> > the world is represented by the attribute-value pairs, which remain
>> > unchanged during a characterization interval, i.e. a continuous
>> > interval between two events in the world.
>> >
>> > Issue: Are the terms "characterization interval" and "continuous
>> > interval" defined by time values? What do we mean by "continuous
>> > interval" between two events?
>>
>> We assumed a partial order between events.
>> An interval between events a and b [a,b] is the set of events x  such
>> that a<= x and x<=b.
>>
>> Does it really need to be defined explicitly?
>>
>> PM there is a hidden issue here though: how do we get agreement on event
>> ordering? isn't this a way to sweep agreement on time under the rug?
>> I am not competent enough to see this through I'm afraid but I see this
>> will creep back up on us
>>
>> Luc: The reference to Lamport is crucial here.  There is ordering in
>> distributed systems because the receipt of a message always follows
>> its sending.
>>
>>  We have extended that to: the use of an entity, follows its
>>  generation.  And, the end of a PE follows its start.  All event
>>  ordering constraints build on those two.
>>
>> Luc: it would be good to identify the events that delimit a
>> characterization interval.
>>  I don't think we have a precise answer for that.
>>  Start would be a generation event.
>>  End could be the generation of a new entity  luc, age=10  terminates
>> luc, age=9
>>  End could be the destructive consumption of an entity: egg broken to
>> make a cake
>>
>> Luc: events are not observable, but time is. Not global time, but local
>> time, found
>> on local clocks, more or less synchronized.
>>
>>
>>
>> > 4. If an asserter wishes to characterize a thing with the same
>> > attribute-value pairs over several intervals, then they are required
>> > to assert multiple entity expressions, each with its own identifier
>> > (so as to allow potential dependencies between the various entity
>> > expressions to be expressed).
>> >
>> > Issue: If a thing with same attribute-value pairs exists over several
>> > time? intervals - what will be the dependencies between the various
>> > entity expressions (since entity expressions = identifier +
>> > attribute-value pairs)? If they are different versions of an entity,
>> > they will have distinguishing attributes other than the simple
>> > occurrence at different points of time. Further, we multiple entity
>> > identifiers are used to refer to the same entity, then how do we
>> > reconcile them later?
>>
>> The example of "luc in boston" in January and June has been discussed
>> extensively.
>> Theroretically, we can find distinguishing attributes, yes (luc with
>> winter clothes
>> and summer clothes). But we have no requirements that these attributes
>> are expressed.
>> So, if we have just "luc in boston" as a characterization, the constraint
>> makes sense.
>>
>> PM agreed
>>
>> >
>> > I believe this consideration is not required and adds a layer of
>> complexity.
>> >
>> > 5. A characterization interval may collapse into a single instant.
>> >
>> > Issue: Are we referring to time values. We seem to be using terms like
>> > "characterization interval", "continuous interval", "single instant"
>> > etc. as surrogates for time. I suggest that we explicitly use "time"
>> > if all these other terms are not distinguishable from time.
>>
>> Time is a can of worms, since we can have multiple clocks, not
>> necessarily synchronised.
>>
>> That's why the whole model is event based.
>>
>> PM see my earlier comment. Satya has a point when he says events are
>> "surrogates for time".
>> Have no solution, but need more discussion goinf forward. This issue will
>> continue
>>
>> Luc: I think that's the other way round. Observable time is an
>> approximation for events.
>>
>>
>> On 09/26/2011 01:58 AM, Provenance Working Group Issue Tracker wrote:
>>
>>> PROV-ISSUE-100 (Entity definition): Section 5.2.1 Entity [Conceptual
>>> Model]
>>>
>>> http://www.w3.org/2011/prov/**track/issues/100<http://www.w3.org/2011/prov/track/issues/100>
>>>
>>> Raised by: Satya Sahoo
>>> On product: Conceptual Model
>>>
>>> Hi,
>>> My review comments for Section 5.2.1 Entity in the current version of
>>> the conceptual model document:
>>>
>>> 1. In PROV-DM, an entity expression is a representation of an
>>> identifiable characterized thing.
>>>
>>> Issue: Since the section heading is for Entity and the PROV DM component
>>> is Entity, I am confused why we are defining "Entity Expression" and not
>>> "Entity"?
>>>
>>> 2. An instance of an entity expression, noted entity(id, [ attr1=val1,
>>> ...]) in PROV-ASN contains an identifier id identifying a characterized
>>> thing; contains a set of attribute-value pairs [ attr1=val1, ...],
>>> representing this characterized thing's situation in the world.
>>>
>>> Issue: When we refer to an entity in provenance assertions (in different
>>> applications), do we use the identifier to refer to it or both identifier +
>>> attribute-value pairs?
>>>
>>> 3. The assertion of an instance of an entity expression, entity(id, [
>>> attr1=val1, ...]), states, from a given asserter's viewpoint, the existence
>>> of an identifiable characterized thing, whose situation in the world is
>>> represented by the attribute-value pairs, which remain unchanged during a
>>> characterization interval, i.e. a continuous interval between two events in
>>> the world.
>>>
>>> Issue: Are the terms "characterization interval" and "continuous
>>> interval" defined by time values? What do we mean by "continuous interval"
>>> between two events?
>>>
>>> 4. If an asserter wishes to characterize a thing with the same
>>> attribute-value pairs over several intervals, then they are required to
>>> assert multiple entity expressions, each with its own identifier (so as to
>>> allow potential dependencies between the various entity expressions to be
>>> expressed).
>>>
>>> Issue: If a thing with same attribute-value pairs exists over several
>>> time? intervals - what will be the dependencies between the various entity
>>> expressions (since entity expressions = identifier + attribute-value
>>> pairs)? If they are different versions of an entity, they will have
>>> distinguishing attributes other than the simple occurrence at different
>>> points of time. Further, we multiple entity identifiers are used to refer
>>> to the same entity, then how do we reconcile them later?
>>>
>>> I believe this consideration is not required and adds a layer of
>>> complexity.
>>>
>>> 5. A characterization interval may collapse into a single instant.
>>>
>>> Issue: Are we referring to time values. We seem to be using terms like
>>> "characterization interval", "continuous interval", "single instant" etc.
>>> as surrogates for time. I suggest that we explicitly use "time" if all
>>> these other terms are not distinguishable from time.
>>>
>>>
>>>
>>>
>>
> --
> 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<http://www.ecs.soton.ac.uk/~lavm>
>
>
>