Great analysis!
I'll continue to elaborate this further.
Wish you all a good start in 2022!
Martin
On 1/7/2022 12:08 PM, George Bruseker wrote:
Hi Rob / Francesco / Martin,
These are all nice examples that maybe we could dig into
further, maybe they display the 'senses of outcome' problem Martin is
pointing to?
An ontological problem that seems to come up in my mind as I try to
conceptualize this is do we mean
1) outcome of type in the sense of a shortcut for a real particular
event of a type (the particular event we do not know much about expect
that it was caused by the first event and has some type)
2) outcome of a type in the sense of a shortcut for a real particular
event that had particular properties (the particular event we do not
know much about expect that it produced something, showed something,
modified something and was caused by the first event)
3) outcome as an evaluation of achievement of an event (succeeds,
fails) - we only talk about one event and evaluate whether it achieves
its goal
These can all cause trouble.
So for example the JFK Assassination:
(E7) Shooting at JFK, (E69) JFK dies
So if we choose to model these as two separate events (legitimate),
then Shooting of JFK had general purpose 'death' and we know in fact
that the shooting triggers the death of JFK (no bullets in JFK, no
dead JFK that day, the shooting caused the death).
So the shortcut 'had outcome of type' could be 'death' just in case we
didn't know anything about the particular death event of JFK and
didn't want to instantiate it as a node.
Shooting of JFK (E7) triggers Death of JFK (E69) has type "Death" (E55)
So here it is that there is an event of type X that is shortcut.
That would be sense 1.
Sense 2 would be something like
Shooting at JFK (E7) triggers Death of JFK (E69) kills JFK (E21)
So here it would be the particular property of E69 to 'kill' an E21
that would be shortcuted
We could also have sense 3, 'had outcome of type' 'success'. As in,
the assassin had general purpose 'death' and the outcome was 'success'.
How would this work in the other examples:
An archeological expedition -- resulted in outcome of type "came
home empty handed" / "found something"
So we have an initial event
Archeological Expedition (E7) has general purpose "Find Something" (E55)
Archeological Expedition (E7) had outcome of type "Found Something" (E55)
And then would the shortcut mean:
a) Archeological Expedition (E7) triggered Dig Activity (A1) has type
Found Something (E55)
or
b) Archeological Expedition (E7) triggered Dig Activity (A1)
encountered Object (E22)
(so here because E22 is 'something', the shortcut is true... that
would seem more like a rule than a property)
or
c) Archeological Expedition (E7) had purpose Find Something (E55)
Archeological Expedition (E7) had outcome of type Found Something (E55)
So here it wouldn't imply a pass through to another event but would
evaluate this event in itself.
Commission of an artwork -- resulted in outcome of type "artist
ran off with the money" / "artist produced something else" /
"artist produced what was wanted" / ...
Commission of Artwork (E7) had general purpose 'production of artwork'
Commission of Artwork (E7) had outcome of type "artist ran off with
the money" / "artist produced something else" / "artist produced what
was wanted"
And then would these shortcuts mean:
a) Commission of Artwork (E7) triggered Production (E12) has
type "artist produced something else" / "artist produced what was
wanted" (E55)
or
Commission of Artwork (E7) triggered Activity (E7) has type "artist
ran off with the money" (E55)
So in the above cases it either shortcuts an E12 or an E7 which we
don't have any details about but for which we would have
classificatory terms like 'desired production', 'undesired production'
OR 'theft/loss' or something like this. As per Martin's mail on types
it falls to the vocabulary to tell us which CRM event type is implied...
or
b) Commission of Artwork (E7) triggered Production (E12) produced Some
Object (E22)
(so here because E22 is 'something', the shortcut is true... that
would seem more like a rule than a property)
But if we do this then we would have to put the 'desired production'
or 'undesired production' categories on the E22 and the non production
/ non created thing would not be expressible.
or
c) Commission of Artwork (E7) had purpose "Build Something" (E55)
Archaological Expedition (E7) had outcome of type "Built that
Something" (E55)
This above case however seems like it would be better covered by the
Plans modelling since what makes something meet or not meet a
criterion is complicated...?
Exhibition planning -- resulted in outcome of type "exhibition" /
"no exhibition" / "revised exhibition" / ...
Exhibition Planning (E7) has general purpose "Run Exhibition" (E55)
Exhibition Planning (E7) had outcome of type "exhibition" / "no
exhibition" / "revised exhibition" (E55)
And then would the shortcut mean:
a) Exhibition Planning (E7) triggered Exhibition (E7) has type
"Exhibition" / "Revised Exhibition" (E55)
it seems here we have a problem with 'no exhibition' because we refer
to a non existent
We cannot say
Exhibition Planning (E7) triggered Exhibition (E7) has type "No
Exhibition" (E55)
b) Exhibition Planning (E7) triggered Exhibition (E7) exhibited Object
(E22)
(so here because E22 is 'something', the shortcut for the positive
exhibiting is true... that would seem more like a rule than a property)
or
c) Exhibition Planning (E7) had purpose "Exhibition" (E55)
Exhibition Planning (E7) had outcome of type "Exhibition" (E55)
If here we relate the outcome back to the domain activity, but we in
reality separate the exhibition planning from the exhibition the
statement is non sensical because exhibition planning is not the
exhibition.
Conservation of object -- resulted in outcome of type "destroyed
object by mistake" / "no change" / "repaired damage" / ...
I won't tackle this one because I'm probably getting repetitive and I
think the activity planning modelling is likely a more robust solution
for this.
So I agree that there are multiple senses that we would have to
navigate. To my original thinking in putting this forward for
discussion, the most sensible interpretation, if this is a good
property at all, would be something like sense 1 where we meant that
the shortcut shortcuts an event which we don't know much about except
for its type and that it is caused by the first event.
This would leave us with at least the problem of events that don't
occur. Like 'no sale'. I think, however, maybe the example of the
commissioning gives an idea of a way out of this. If the original
intention of the commission is to trigger an E12 that is satisfactory,
if the thing doesn't get made, but we classify the outcome as
'theft/artist ran away', it is not that the commission did not result
in any other event, it just didn't result in an E12 of any sort. It
resulted in an E7 of type theft. In the 'no sale', although we may not
be privy to it, there may have been some furtive activities (E7) that
tried to hawk the item. This anonymous E7 is a real event (attempting
to hawk the item) and is legitimately classifiable as a 'no sale'.
But maybe there are good arguments for sense 2 or 3 or yet another
solution I haven't drawn out.
Best,
George
--
------------------------------------
Dr. Martin Doerr
Honorary Head of the
Center for Cultural Informatics
Information Systems Laboratory
Institute of Computer Science
Foundation for Research and Technology - Hellas (FORTH)
N.Plastira 100, Vassilika Vouton,
GR70013 Heraklion,Crete,Greece
Vox:+30(2810)391625
Email:mar...@ics.forth.gr
Web-site:http://www.ics.forth.gr/isl
_______________________________________________
Crm-sig mailing list
Crm-sig@ics.forth.gr
http://lists.ics.forth.gr/mailman/listinfo/crm-sig
