W3C home > Mailing lists > Public > public-prov-wg@w3.org > January 2012

RE: complementOf -> viewOf: proposed text

From: Myers, Jim <MYERSJ4@rpi.edu>
Date: Tue, 17 Jan 2012 16:53:15 +0000
To: Graham Klyne <graham.klyne@zoo.ox.ac.uk>, James Cheney <jcheney@inf.ed.ac.uk>
CC: Paolo Missier <Paolo.Missier@ncl.ac.uk>, Luc Moreau <L.Moreau@ecs.soton.ac.uk>, Paolo Missier <paolo.missier@newcastle.ac.uk>, "public-prov-wg@w3.org" <public-prov-wg@w3.org>
Message-ID: <3131E7DF4CD2D94287870F5A931EFC2302A52953@EX14MB2.win.rpi.edu>
FWIW: I definitely can't keep up any more, but I feel like trying to throw a hopefully clarifying comment in again. 

On the 3-2 level thing discussion: I think the one point that has to be kept in this debate is that we (humans) often name things that do not have unambiguous provenance in our w3c model because they have not been defined/characterized in ways that distinguish their creation/destruction from changes during their lifetimes. Thus we define entities in the model which do have unambiguous characterization of their lifetimes: each entity therefore  represents one full characterization  of some thing (and only one - they are unambiguous w.r.t. everything being asserted) . I think the ambiguity we feel comes from the 'fact' that things, while we try to pretend they are real/ uncharacterized,  are already characterizations of parts of the universe according to some theory - things come from some informal or formal ontology with partial characterizations -  they just aren't entities unless that characterization includes aspects that let you unambiguously distinguish creation/destruction from lifecycle changes. 

Specialization versus alternate: I think the above way of thinking makes the distinction between specialization and alternate clear - in specialization, the only changes in characterization between the two entities has to do with lifecycle/provenance issues -(e.g. we answer the question 'does location change matter?' to distinguish the entities Luc and  Luc-in-Boston). Whereas alternates involve changes in characterization that are sufficient that some people would say we're talking about different things as well (the statue and its metal, the play and its performance (FRBR), etc.).I think that leads to some concrete definitions and implications at the level being discussed, but I don't think it defines which one is primitive:

if alternate is the primitive, e1, e2 are alternates over some interval only of there exists some single thing they both characterize (which may be different than things they nominally characterize - see note below). Specialization is then a subtype of that relationship in which e1 is defined only over a subset of the lifetime of e2, and where e1 is a specialization of e2 over its whole lifetime.

If specialization is primitive, e1 is a specialization of e2 if there is a thing they both characterize, the relationship holds for e1's lifetime and e1's lifetime is within the bounds of e2's. If e1 can also be considered a specialization of e3, then e2 and e3 are alternates at least for the duration of e1's lifetime. 

A couple notes 

- these definitions explicitly leaves open the possibility that e1, e2, e3  are defined as characterizing different things - e1 is paulo-in-cafe characterizing paulo while e2 characterizes the customer-in-red-chair with e1 being a specialization of e2,  e2 could characterize a statue and e3 the metal with e2 and e3 alternates both characterizing the same 'thing in the museum' , etc. - but there must exist something (e.g. in both these cases a thing that is ~"the stuff inside a specified space-time boundary") that they both characterize in different ways. Entities characterizing a document and a file could likewise be alternates of some thing that is ~'serialized text').

- for specialization, I think we often want specialization to have further implications (e.g. that the properties of Luc-in-Boston and Luc that are not related to location (e.g. mass) must be the same for both, or even tha tboth things are physical objects with space-time extent), but I think that becomes domain specific and out-of-scope for the model. Domain folks should be able to come up with such inferences though based on the assertion of an alternative or specialization relationship and their domain ontology. 

I think all of this has clear implications for transitivity, etc. - alternate cannot be transitive, specialization is, and two entities that are specializations of another one are by definition at least alternates of each other over the intersection of their lifetimes.

I hope some of this is directly useful, but I also throw it out because I think it may help explain a few things: 

Specialization being between entities and between things are not really opposing views - if a thing has some characterization, clearly two entities fully characterizing the lifecycle aspects  of that thing will share the aspects of the characterization implicit in the thing and thus there's a relationship between the entities themselves as a consequence.

Which relationship is primitive is a modeling choice that may not have impact on what can be modeled - I think the definitions can be made consistent either way and I think it comes down to which is easier to explain/which do we expect to be the more common use case, etc. (Or said a different way - if we are coming up with definitions where which is primitive does matter in terms of expressivity, I'm concerned that we're constraining expressivity).

 -- Jim
From: Graham Klyne [graham.klyne@zoo.ox.ac.uk]
Sent: Monday, January 16, 2012 5:59 PM
To: James Cheney
Cc: Paolo Missier; Luc Moreau; Paolo Missier; public-prov-wg@w3.org
Subject: Re: complementOf -> viewOf: proposed text


I would say that e1 and e2 *denote* aspects of the same thing, with different
levels of specialty.

Thus, when we say specializationOf(e1,e2), I would suggest that the (full)
*description* of e1 subsumes the description of e2; that is, any descriptive
statement that is true of e2 is also true of e1, but that further things may be
true of e1 that are not true of e2.  That is, e1 denotes a more specialized, or
more constrained, aspect of the thing denoted by e2.

So e1 and e2 denote things, not descriptions, but the relationship of
specialization allows us to say things about the relative descriptions of those

To my mind, introducing the distinction between things and descriptions into the
domain of discourse (i.e. the range of denotations of by names like e1 and e2)
just complicates the expressions of our language without offering any useful
increase in expressive capability.


On 16/01/2012 16:09, James Cheney wrote:
> In that case, would you (or Luc) also agree with describing "specializationOf(e1,e2)" as "e1 and e2 describe the same thing, and e1 is more detailed/specific than e2"?
> The concern I have about specalizationOf is that it is about the descriptions, not the described things.  I can rationalize alternateOf as saying that "e1 and e2 refer to the same thing", which is almost what Luc wrote, but to rationalize specializationOf I need e1 and e2 to refer to descriptions, not things themselves.  (I think it is this distinction that is one of the root causes of confusion here.)
> --James
> On Jan 16, 2012, at 4:06 PM, Paolo Missier wrote:
>> thing (we just crossed in the mail)
>> -Paolo
>> On 1/16/12 4:03 PM, Luc Moreau wrote:
>>> Hi James,
>>> To add on to this, did we really mean
>>> e1 and e2 provide two different characterization of the same entity
>>> or did we mean
>>> e1 and e2 provide two different characterization of the same THING?
>>> Luc
Received on Tuesday, 17 January 2012 16:54:06 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:58:11 UTC