W3C home > Mailing lists > Public > public-prov-wg@w3.org > July 2011

Re: PROV-ISSUE-30 (name-for-bob): What name do we use for the BOB construct? [Conceptual Model]

From: Ryan Golden <ryan.golden@oracle.com>
Date: Mon, 25 Jul 2011 12:57:14 -0500
Message-ID: <4E2DAE7A.3060401@oracle.com>
To: public-prov-wg@w3.org
I don't have a huge problem with "pil:snapshot," "pil:perspective," or 
"pil:view" as opposed to "pil:entity."  To me, most of the proposed 
conceptualizations on the BOB topic at this point are isomorphic (except 
for Jim's)--we're just debating on terminology.

One thing I would point out, though.  In a set of provenance assertions, 
the terms "snapshot," "perspective," and "view" may subtly imply an "IVP 
of" style relationship, which may not necessarily be the case. If I 
assert "View B is derived from View A," the natural language term "view" 
makes a subtle implication (to me at least) that View B and View A are 
different views /of the same thing/.  This is not necessarily the case.  
For example, "view" would look a bit funny in this assertion, whereas 
entity would not: "Oxygen is derived from Air (by the process of 
distillation)."

Nevertheless, I could be happy with any of these terms for now.

--Ryan

On 7/25/2011 11:39 AM, Paolo Missier wrote:
> Khalid just managed to put "snapshot", "perspective" and "view" in the 
> same sentence. I see this as a good sign :-)
>
> So, having put in a good word earlier for a variation of "snapshot", I 
> would like to go back for a moment to "view", as intended in the 
> database sense.
>
> As we all know a view is indeed a perspective on the underlying 
> database, and it has two key characteristics:
> - it projects out irrelevant attributes (out of a potentially 
> unbounded number of them, as in our case)
> - if you materialize a view, this materialization remains valid only 
> as long as none of the attributes that it contains (those that matter 
> to the view) change their value in the underlying DB. When that 
> happens, you need to recompute the view (= create a new materialized 
> view, at least conceptually).
>
> Isn't that what BOBs do?
>
> In PIL the database itself is made of "things" (in Luc's latest 
> terminology as of today) that we can only talk about and partially 
> reveal by creating (mutiple) views, each valid within some temporal 
> window, which is defined by value change events in its attributes.
>
> this also avoid the "frozen in time" connotation that snapshots have 
> in several contexts.
>
> -Paolo
>
>  On 7/25/11 11:07 AM, Khalid Belhajjame wrote:
>> On 25/07/2011 10:24, Luc Moreau wrote:
>>> The problem with Snapshot (like state, etc), is that it is the
>>> snapshot of an entity.
>>> We just don't want to distinguish an entity from its state, or an
>>> entity from its snapshot.
>> On the other hand, Snapshot has the advantage of conveying the fact that
>> it is a description from a certain perspective (view).
>>
>> khalid
>>
>>> Hence, using Entity avoids this problem.
>>>
>>> Luc
>>>
>>> On 07/25/2011 10:19 AM, Paul Groth wrote:
>>>> I thought we were getting somewhere with snapshot.....
>>>>
>>>> I don't think ENTITY really captures the intuition behind a BOB. It's
>>>> too general.
>>>>
>>>> thanks,
>>>> Paul
>>>>
>>>> Luc Moreau wrote:
>>>>> The word 'Entity' should also be considered for the construct BOB.
>>>>>
>>>>> If we do so, the text 'characterized entity' should be replaced by
>>>>> something else in the draft specification.
>>>>> Why not 'thing'?
>>>>>
>>>>>
>>>>> So, the text could become:
>>>>>
>>>>>     Section 4.
>>>>>     In the world (whether real or not), there are things, which 
>>>>> can be
>>>>> physical, digital, conceptual, or otherwise, and activities involving
>>>>> things.
>>>>>     Words such as thing or activity should be understood with their
>>>>> informal meaning.
>>>>>     This specification is concerned with characterized things, 
>>>>> that is,
>>>>> things and their situation in the world, as perceived by the 
>>>>> asserter.
>>>>>
>>>>>     Section 5.1
>>>>>     An ENTITY represents an identifiable characterized thing.
>>>>>
>>>>>
>>>>> Luc
>>>>>
>>>>> On 07/24/2011 11:43 PM, Reza B'Far wrote:
>>>>>> First, for the record Khalid was the person suggesting Snapshot :)
>>>>>>
>>>>>> The way I've seen snapshot used commercially, it's fairly consistent
>>>>>> with the current definition of BOB.  There is some murkiness on both
>>>>>> sides (how "snapshot" is used commercially and I think we're still
>>>>>> iterating here on the definition of BOB, but may be that's close 
>>>>>> to be
>>>>>> finalized).  However, I think they are close enough.  What I liked
>>>>>> about "Snapshot" is that its intuitive and is used in several 
>>>>>> domains
>>>>>> that I know of (content management, legal, configuration systems, 
>>>>>> and
>>>>>> I've also seen use-cases in microfilm production by old-school
>>>>>> librarians).  Also, I think "Snapshot" offers a huge advantage that
>>>>>> it's neither explicitly linked to the entity nor its state.  And I
>>>>>> know the distinction between entity vs. entity's state and how 
>>>>>> that's
>>>>>> articulated has been in a lot of the discussions.  Using "Snapshot"
>>>>>> sort of obsoletes that discussion.
>>>>>>
>>>>>> On 7/24/11 12:57 PM, Stephan Zednik wrote:
>>>>>>> I am not partial to snapshot, partially because of the extensive
>>>>>>> functional usage of the term.  I have always associated a snapshot
>>>>>>> with a point in time, not a duration - but this may be an incorrect
>>>>>>> association.
>>>>>>>
>>>>>>> I am open to discussing it, but my initial inclination was negative
>>>>>>> towards it.
>>>>>>>
>>>>>>> Will we use the same definition as we have been using for BOB?
>>>>>>>
>>>>>>> --Stephan
>>>>>>>
>>>>>>> On Jul 24, 2011, at 9:52 AM, "Reza B'Far"<reza.bfar@oracle.com>
>>>>>>> wrote:
>>>>>>>
>>>>>>>> I second the term "Snapshot".  This term also has functional usage
>>>>>>>> in several commercial application categories used within 
>>>>>>>> roughly the
>>>>>>>> same meaning.
>>>>>>>>
>>>>>>>> On 7/24/11 3:45 AM, Khalid Belhajjame wrote:
>>>>>>>>> Hi Stephan,
>>>>>>>>>
>>>>>>>>> Given the example you gave in your previous email, I think that
>>>>>>>>> "EntitySpanshot" or "Snapshot" should be fine, given that it
>>>>>>>>> reflect the fact that it is a description of an entity that holds
>>>>>>>>> for some period of time.
>>>>>>>>>
>>>>>>>>> Do you agree?
>>>>>>>>>
>>>>>>>>> khalid
>>>>>>>>>
>>>>>>>>> On 23/07/2011 20:24, Stephan Zednik wrote:
>>>>>>>>>> I do not feel that EntityInstance, EntityInstantiation, or
>>>>>>>>>> InstantiatedEntity make sense for the book ownership 
>>>>>>>>>> scenario, or
>>>>>>>>>> any scenario modeling the provenance of changes in 
>>>>>>>>>> characteristics
>>>>>>>>>> of a physical object.
>>>>>>>>>>
>>>>>>>>>> To reiterate the example since I haven't committed it to a wiki
>>>>>>>>>> page yet.  Book X is an entity that represents a real world
>>>>>>>>>> object.  It can be put on a shelf, loaned to friends, damaged,
>>>>>>>>>> and/or destroyed.  It has important characteristics (condition,
>>>>>>>>>> ownership, location, etc) that may change over the life of the
>>>>>>>>>> book.  We may want to represent the provenance of the book as a
>>>>>>>>>> chain of ownership.
>>>>>>>>>>
>>>>>>>>>> |<----------------------------------------------------- Book X
>>>>>>>>>> ----------------------------------------------------------------->| 
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> |<!------ Book X with owner A ---->|<----Book X with owner B
>>>>>>>>>> ---->|<---- Book X with owner A --------->|
>>>>>>>>>>
>>>>>>>>>> If a book changes ownership, is the "book with changed 
>>>>>>>>>> ownership"
>>>>>>>>>> a different EntityInstance?  A different InstantiatedEntity?  I
>>>>>>>>>> don't think what we current call a BOB is an 'instance of'
>>>>>>>>>> anything.  I think of it as a description of an entity that 
>>>>>>>>>> holds
>>>>>>>>>> for some time period (not necessarily given) for which
>>>>>>>>>> contextually important mutable characteristics of the the entity
>>>>>>>>>> are held to be known.
>>>>>>>>>>
>>>>>>>>>> --Stephan
>>>>>>>>>>
>>>>>>>>>> On 7/22/2011 5:29 AM, Curt Tilmes wrote:
>>>>>>>>>>> On 07/22/2011 03:43 AM, Khalid Belhajjame wrote:
>>>>>>>>>>>> The term "Snapshot" was suggested some time ago, and it seems
>>>>>>>>>>>> that
>>>>>>>>>>>> several people did like it.
>>>>>>>>>>>> We can also use the term "EntitySnapshot".
>>>>>>>>>>> Following from snapshot:
>>>>>>>>>>>
>>>>>>>>>>> EntityInstance
>>>>>>>>>>> EntityInstantiation
>>>>>>>>>>> InstantiatedEntity
>>>>>>>>>>>
>>>>>>>>>>> Curt
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>
>>
>
>
Received on Monday, 25 July 2011 17:57:51 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 26 April 2012 13:06:37 GMT