Proposed - for ternary relations in reference create two relations - R-at-all-times, R-at-some-time

[GoogleCodeExporter]

In order to support a reading of temporal relations in BFO2 it is proposed that 
for each temporal relation R specified in the reference, two properties are 
instead created:

R-at-some-time
R-at-all-times

For x R y at t,

The scope over which the some and all t are selected from is all t such that 
exists_at(x,t).

This proposal does not rule out other relations. It suggests only a minimally 
adequate change to support our goal. Further work should continue.

Original issue reported on code.google.com by alanruttenberg@gmail.com on 25 May 2012 at 3:41

[GoogleCodeExporter]

At the May 25 teleconference it was decided that we will review each of the 
ternary relationships and decide on a case by case basis whether to create an 
at-all-times, at-some-times, or both relationships. Chris pointed out that it 
may be able to prove, e.g., that quality-of-at-some-time is equivalent to 
quality-of-at-all-times. Alan pointed out that choice varies between relation 
and inverse, so quality-of-at-all-times, but has-quality-at-some-time.

Mathias, Chris, Alan, Ron will review the relations and propose what the choice 
should be

Original comment by alanruttenberg@gmail.com on 25 May 2012 at 5:27

• Changed state: Accepted

[GoogleCodeExporter]

A realization: The inverse properties need to have their scope done 
differently. 

Consider a part-of-at-all-times b
-> forall(t) exists_at(a,t) -> exists_at(b,t) and part_of(a,b,t)

The inverse of this is:
-> forall(t) exists_at(a,t) -> exists_at(b,t) and has_part(b,a,t)

But this is not necessarily the same as a has-part-at-all-times b, which 
-> forall(t) exists_at(b,t) -> exists_at(a,t) and has_part(b,a,t)

It should be the case that r(a,b) <-> inv(r)(b,a)

However in cases where a and b exist over different temporal regions this will 
not hold.

I'm not sure what a desirable solution should be yet. Two obvious ones are to 
make clear the properties are not inverse of one another and define additional 
properties that are, or to use the anonymous inverse relation in OWL.

Original comment by alanruttenberg@gmail.com on 6 Jun 2012 at 3:35

[GoogleCodeExporter]

Ron Rudnicki has submitted his review (see comment 1) and offers (after a 
clarification of the FOL expansion of the relations) that both -at-some-time 
and at-all-time versions of the relations can be included.

Mathias has submitted his review. Comments of mine in line are marked as <<AR: 
>>. 

((c-part-of c-has-part) :ternary) for both: at some times & at all times

((c-ppart-of c-has-ppart) :ternary) for both: at some times

<<AR: Why not the same as c-part-of>>

((m-part-of m-has-part) :ternary) [for both: at some times [if the b in a 
member_part_of b is defined by extension (which I think is not the case for 
BFO), then, of course, for the has-member-part it is:  at all times]
<<AR: I realize I don't quite grok m-part-of and I believe the latest reference 
draft makes some changes, so I will review further>>

((located-in  at some times & at all times has-location) :ternary)
<<AR: I read this as both at-some-time and at-all-times versions should be 
created for both relations>> 

((located-at-r  at some times r-location-of) :ternary)
<<AR: I read this as located-at-r and r-location-of are both to be defined as 
at-some-time>> 

((inheres-in  at all times bearer-of at some times) :ternary)

((s-dep-on has-s-dep at all times) :ternary)
Maybe I got that wrong, but the ref states:
" Inherence is a subrelation of s-depends on which holds between a dependent 
continuant and an independent continuant. Since dependent continuants cannot 
migrate from one independent continuant bearer to another, it follows that if a 
s-depends on independent continuant b at some time, then a s-depends on b at 
all times at which a exists.  Inherence is in this sense redundantly 
time-indexed."
this seems to be wrong since some dependent continuants CAN migrate: they are 
generically dependent.

<<AR: Generic dependents are disjoint from specific dependents and there is no 
dependent continuant class (reference typo - will report), so there is no error 
as suggested. However I also question some of the logic. Inherence is a 
subrelation of s-depends and so it doesn't seem one can conclude what is 
concluded for the superproperty based on the subproperty.
I think Mathias' intention is that s-dep-on is suggested to be at-all-time and 
has-s-dep is at some times(please verify). If inherence is redundantly time 
indexed perhaps a reference issue should be added. Simply make the relation 
binary.

There may be a need for at-all-times version of has-versions of specific 
dependence and subproperties to capture cases such as mass and other physical 
properties for which there is a single instance that persist for the entire 
lifetime of the entity. See 
http://hl7-watch.blogspot.com/2008/02/weight-of-baby.html
 >>

((g-dep-on has-g-dep) :ternary) at some times
((q-of at all times  has-q at some times) :ternary)
((f-of has-f) :ternary) analogues to quality
((r-of has-r) :ternary) analogues to quality
((d-of has-d) :ternary) analogues to quality

((has-material-basis material-basis-of) :ternary) analogues to quality
<<AR: This one I'm not sure of - need to better understand the relation. 
Examples needed>>

((concretizes concretization-of) :ternary)
given the elucidation:
" a(concretizes)[Elucidation: a concretizes b at t means: a is a specifically 
dependent continuant & b is a generically dependent continuant & for some 
material entity c, a s-depends on c at t and b g-depends on c at t, and if b 
migrates from bearer c to another bearer d than a copy of a will be created in 
d. [075-001]
I assume that "a copy of a" is a' viz. another instance, right?
In that case the relation holds at all times

<<AR: By this logic it is concretization-of that should be at-all-times. I 
think if one were be chosen for concretizes it would be -at-some-times. I'm not 
sure if the language of migration is the best choice for g-dependence - will 
consider putting in reference issue>>

((st-projects-onto-s s-projection-of-st) :ternary)
((has-participant participates-in) :ternary) at some time

<<AR: I assume that at-some-time is the choice for both cases. However for the 
participant case, earlier discussion (i should hunt for reference, but I think 
it originate with David OS) suggested that continuous participation was also a 
desired relation.>>

Original comment by alanruttenberg@gmail.com on 6 Jun 2012 at 4:09