Issue 625: O13 "triggers" scope note
During the 55th CIDOC CRM-48th FRBR/LRMoo SIG meeting Wolfgang Schmidle proposed starting a new issue, where to discuss the scope note of O13 triggers (in particular the last clause which places a requirement for the triggering event to have been concluded before the start of the triggered event).
The examples and the scope note were considered to be conflicting one another.
Anais Guilleme proposed that if we want to use scientific examples, we can use alterations examples in conservation science or mechanical effort test on materials
December 2022
Post by Wolfgang Schmidle (8 DEcember 2022)
Dear All,
O13 "triggers" in CRMsci requires the triggering event to be finished by the time the triggered event starts. In the example of rainfall causing a landslide in the scope note, it would mean that continuing rainfall has to be split up into the part before the landslide and after the landslide. This is not obvious and needs to be reflected in the scope note.
One example for O13 is the 1966 flood in Florence triggering mould growth on books stored in flooded library rooms. I read this to claim that the mould growth started only after the flood had receded completely, which I find a strong claim. However, Steve argued that the cited source does not contain information about the mould starting to grow while the flood was still ongoing.
Other examples we discussed were:
* an asthmatic in a room full of cats, triggering an asthma attack while still being in the room
* an earthquake triggering the destruction of houses: It is not realistic to split up the earthquake event into parts for each destroyed house.
Best,
Wolfgang
Post by Martin Doerr (20 December 2022)
Dear All,
I think we need to reconsider this. The examples somehow make clear that the triggering event to be finished by the time the triggered event starts is incidental and not essential the the intended meaning. I now support an interpretation that the triggering event is essentially of different nature from the triggered one. I think we need a spatiotemporal overlap of the triggering event with the beginning of the triggered event. Without diving into STVs,
it appears to me that P176 starts before the start of and P173i ends after or with the start of is the best definition.
In STV thinking, the triggered event should also not contain areas that occur locally before all neighbouring areas and do not overlap with the triggering event. I.e., and event could start at two different places and then merge into one, but only one starting area would be triggered. That does not make sense.
best,
Martin
Post by Thanasis Velios (20 December 2022)
Indeed and the original questions raised were about the phrase in the
introduction "The association of the two events is based on their
temporal proximity, i.e. the triggering event ends when the triggered
event starts." I think the examples mentioned in the scope note indicate
a temporal relationship of:
P176 starts before the start of
than the agreed:
P182 ends before or with the start of
Even in the case of, say, a spring-loaded mousetrap, it would be
difficult to tell that the event of stepping on the trip (of the trap)
is completed before the spring is released.
So we either explain the event splitting approach or opt for the safer
option of P176.
All the best,
Thanasis
P.S. I think any change would need to go to 2.1.
Post by Martin Doerr (20 April 2023)
Dear All,
Here my first go:
OLD
O13 triggers (is triggered by)
Domain:
E5 Event
Range:
E5 Event
Quantification:
many to many (0,n:0,n)
Scope note:
This property associates an instance of E5 Event that triggers another instance of E5 Event with the latter. It identifies the interaction between events: an event can activate (trigger) other events in a target system that is in a situation of sustained tension, such as a trap or an unstable mountain slope giving way to a land slide after a rain or earthquake. In that sense the triggering event is interpreted as a cause. However, the association of the two events is based on their temporal proximity, with the triggering event ending when the triggered event starts.
Examples:
The earthquake of Parnitha in 1999 triggered the rotational landslide that was observed along the road on the same day. (fictitious)
The explosion at the Montserrat massif in 2007 (near Barcelona, Spain) triggered the rock fall event happened on 14 February 2007 (Vilajosana et al., 2008).
The 1966 flood in Florence triggered mould growth on books stored in flooded library rooms (Rubinstein, N., 1966)
In First Order Logic:
O13(x,y) ⇒ E5(x)
O13(x,y) ⇒ E5(y)
O13(x,y) ⇒ P182(x,y)
NEW
O13 triggered (was triggered by)
Domain:
E5 Event
Range:
E5 Event
Quantification:
many to many (0,n:0,n)
Scope note:
This property associates an instance of E5 Event that triggers another instance of E5 Event with the latter. It identifies the interaction between events: an event can activate (trigger) other events in a target system that is in a situation of sustained tension, such as a trap or an unstable mountain slope giving way to a land slide after a rain or earthquake.
The distinction of the triggering event from the triggered one lies in their difference of nature: The starting of the triggered event is the result of an interaction of constituents with the triggering one, but not a continuation of the kinds of processes of the latter. Therefore the triggering event must spatiotemporally overlap with the initial time and area of the triggered event, and the spreading out of the subsequent phenomena must initiate from this area and time and not from multiple independent areas.
Examples:
The earthquake of Parnitha in 1999 triggered the rotational landslide that was observed along the road on the same day. (fictitious)
The explosion at the Montserrat massif in 2007 (near Barcelona, Spain) triggered the rock fall event happened on 14 February 2007 (Vilajosana et al., 2008).
The 1966 flood in Florence triggered mould growth on books stored in flooded library rooms (Rubinstein, N., 1966)
In First Order Logic:
O13(x,y) ⇒ E5(x)
O13(x,y) ⇒ E5(y)
Best,
Martin
Post by Wolfgang Schmidle (21 April 2023)
Here's a diff:
* label:
OLD O13 triggers (is triggered by)
NEW O13 triggered (was triggered by)
(in the examples it was already called "triggered" rather than "triggers")
* scope note:
Part 1 is unchanged:
This property associates an instance of E5 Event that triggers another instance of E5 Event with the latter. It identifies the interaction between events: an event can activate (trigger) other events in a target system that is in a situation of sustained tension, such as a trap or an unstable mountain slope giving way to a land slide after a rain or earthquake.
Part 2:
OLD In that sense the triggering event is interpreted as a cause. However, the association of the two events is based on their temporal proximity, with the triggering event ending when the triggered event starts.
NEW The distinction of the triggering event from the triggered one lies in their difference of nature: The starting of the triggered event is the result of an interaction of constituents with the triggering one, but not a continuation of the kinds of processes of the latter. Therefore the triggering event must spatiotemporally overlap with the initial time and area of the triggered event, and the spreading out of the subsequent phenomena must initiate from this area and time and not from multiple independent areas.
* FOL:
O13(x,y) ⇒ P182(x,y) removed
(Domain, range, quantification, examples are unchanged)
About the changes:
Scope note part 2: If there needs to be an interaction of constituents and thus a spatiotemporal overlap, then I am not sure I understand the 1966 flood example. There is an overlap between the flood and a book getting wet and an overlap between a book being wet as a result and the growing of the mould, but is there an obvious interaction between the flood and the mould beginning to grow on a book? I am assuming O13 is not meant to be transitive?
What is the initial time and area of "mould growth on books stored in flooded library rooms"? Is it obvious that this area is connected and not multiple independent areas?
FOL / superproperties: The new scope note suggests P132 "spatiotemporally overlaps with", as well as P176 "starts before the start of" (also suggested by Thanasis) and P173i "ends after or with the start of"?
Additional questions:
Scope note part 1: What is the sustained tension in the target system (books stored in library rooms) in the 1966 flood example? Or in a house that is destroyed by an earthquake or a wildfire?
Examples: Since we want to get rid of fictitious examples, would it make sense to replace the earthquake/landslide example? Non-fictitious examples would be https://en.wikipedia.org/wiki/
Best,
Wolfgang
Post by Wolfgang Schmidle (23 April 2023)
One more question: Can a wildfire trigger the destruction of a house, or does the burning down of the house form a part of the wildfire event? In other words, is there a difference of nature or is the destruction a continuation of the kinds of processes of the wildfire?
Post by Thanasis Velios (23 April 2023)
I think the meaning of a library flooding is that the books get wet. We
can model to a microscopic scale if the situation requires, but this is
not necessary for this example I think. In the case of the wildfire, it
could be both, i.e. both triggered and also part of it, but we cannot
assume that the triggering event always completely contains the
triggered event.
I would also propose a minor rewriting of Martin's paragraph:
"The distinction of a triggering event A from the triggered event B lies
in their difference of nature. The starting of B is the result of an
interaction of material constituents of A with material constituents of
B. However, B does not necessarily continue the kinds of processes of A.
Therefore the triggering event A must spatiotemporally overlap with the
initial time and area of the triggered event B. Any subsequent phenomena
must initiate from this area and time."
Is that an improvement at all? Or am I confusing things?
All the best,
Thanasis
Post by Martin Doerr (30 April 2023)
Dear Wolfgang,
Your questions well-taken, but please do not seek a logical surrogate of reality. It does not exist. The logic can be not more than an overlay, approximating and simplifying reality, in more detail:
On 4/21/2023 1:59 PM, Wolfgang Schmidle via Crm-sig wrote:
Here's a diff:
* label:
OLD O13 triggers (is triggered by)
NEW O13 triggered (was triggered by)
(in the examples it was already called "triggered" rather than "triggers")
* scope note:
Part 1 is unchanged:
This property associates an instance of E5 Event that triggers another instance of E5 Event with the latter. It identifies the interaction between events: an event can activate (trigger) other events in a target system that is in a situation of sustained tension, such as a trap or an unstable mountain slope giving way to a land slide after a rain or earthquake.
Part 2:
OLD In that sense the triggering event is interpreted as a cause. However, the association of the two events is based on their temporal proximity, with the triggering event ending when the triggered event starts.
NEW The distinction of the triggering event from the triggered one lies in their difference of nature: The starting of the triggered event is the result of an interaction of constituents with the triggering one, but not a continuation of the kinds of processes of the latter. Therefore the triggering event must spatiotemporally overlap with the initial time and area of the triggered event, and the spreading out of the subsequent phenomena must initiate from this area and time and not from multiple independent areas.
* FOL:
O13(x,y) ⇒ P182(x,y) removed
(Domain, range, quantification, examples are unchanged)
About the changes:
Scope note part 2: If there needs to be an interaction of constituents and thus a spatiotemporal overlap, then I am not sure I understand the 1966 flood example. There is an overlap between the flood and a book getting wet and an overlap between a book being wet as a result and the growing of the mould, but is there an obvious interaction between the flood and the mould beginning to grow on a book? I am assuming O13 is not meant to be transitive?
What is the initial time and area of "mould growth on books stored in flooded library rooms"? Is it obvious that this area is connected and not multiple independent areas?
Well, it is obvious to any expert. The silent assumption of such a case of "causality" is that the interaction would not have happened under "normal" circumstances. The books obviously became wet by the flood. No normal library would make the books wet otherwise. The statement that the flood "triggered" actually approximates and simplifies the statement that the books became wet by the flood in a way that cold not be remedied readily by the library. In general, is not possible to break down such processes into discrete atomic logical steps.
There is a considerable logical-philosophical complexity to any concept of causality. Therefore we have refused so far to introduce such a concept into CRMbase. To my understanding, the reasoning is about defaults of the environment, blaming the more exceptional to be the "cause", whereas others could equally blame the lack of foresight and protective measures, or any other random factor, just as someone getting in the path of a bullet by walking.....
Would that explanation satisfy your question?
FOL / superproperties: The new scope note suggests P132 "spatiotemporally overlaps with", as well as P176 "starts before the start of" (also suggested by Thanasis) and P173i "ends after or with the start of"?
Additional questions:
Scope note part 1: What is the sustained tension in the target system (books stored in library rooms) in the 1966 flood example? Or in a house that is destroyed by an earthquake or a wildfire?
The sustained tension in this case is the sensitivity of the material to humidity. Whatever would raise humidity sufficiently would "trigger" such a process.
Examples: Since we want to get rid of fictitious examples, would it make sense to replace the earthquake/landslide example? Non-fictitious examples would be https://en.wikipedia.org/wiki/Paradise,_California#2018_fire or
https://en.wikipedia.org/wiki/The_Way_Things_Go (an artistic cascade of triggering events)
Sure, I wonder if colleagues from FORTH could recover landslide examples from the European InGeoClouds project.
Good examples could also be some houses falling down at the seaside around Santa Barbara coast in California, because of landing erosion approaching them.
By the way, I think I just made a statement about principles.
Would you regard this as noteworthy as principles?
All the best,
Martin
In the 56th joint meeting of the CIDOC CRM and ISO/TC46/SC4/WG9 &49th FRBR/LRMoo SIG, the SIG reviewed the proposal by MD to redraft the scope note of O13 triggered (was triggered by) and agreed to it. The new scope note will appear in CRMsci v2.1
For the details the scope note update, see attached
Issue closed
Crete, May 2023.