[Crm-sig] (Geographic) Place vs. SpaceTimeVolume (?) ISSUE: make methodology clear. ISSUE add "islands" to E27 Site.

Martin Doerr martin at ics.forth.gr
Sat Oct 5 20:47:07 EEST 2019


Dear Dan, All

I do not understand what you mean by making your life "easy". The 
question is, if the identity conditions of the classes you use are 
compatible with the reality you describe, and not if some properties or 
labels appear convenient.

Nobody forces you to use the CRM. It is made for reliable information 
integration. If you use it, better not abuse it;-). Franco has made good 
arguments below, that E53 Place is not what you take it for, and that 
the distinction of bona fide and fiat cannot be verified in relevant 
cases. If you replace E53 by your understanding of a "Place", basically 
you abuse the CRM. If scope notes are not well-written, please refer to 
them,  but please do not create your own;-).

Having said that, we have the following: The Space-Time Volume takes its 
identity from either coordinates or a phenomenon, including claims in 
terms of coordinates, that stay within such, fuzzy in general, 
boundaries that form "volumes".

No "named place" exists forever, hence it changes in time. If I describe 
a dinosaur bone found in Desert Gobi, there was no Desert Gobi at that 
time. "E53 Place" is not a "place".  E53 describes a geometric extent. 
Hence, it is the projection of the (maximal or current) extent of the 
named phenomenon. It is good practice to define an instance of E53 Place 
"Extent of Desert Gobi in 2019". It is wrong to regard the Desert as an E53.

It is explicit in the scope note of E4 Period:

"A geopolitical unit as a specific case of an instance of E4 Period is 
the set of activities and phenomena related to the claim of power, the 
consequences of belonging to a jurisdictional area and an administrative 
system that establishes a geopolitical unit. Examples from the modern 
period are countries or administrative areas of countries such as 
districts whose actions and structures define activities and phenomena 
in the area that they intend to govern. The borders of geopolitical 
units are often defined in contracts or treaties although they may 
deviate from the actual practice. The spatiotemporal properties of 
Geopolitical units can be modelled through the properties inherited from 
E92 Spacetime Volume."


All examples you gave of things with a political identity are instances 
of E4 Period. Period;-). All "places" defined by boundaries of 
geological features, such as islands, are Physical Features, typically 
E27 Site. Both have spatial projections.
The island of Crete was not an island 5 million years ago (Mediterranean 
dried out), and considerably larger in the last Ice Age.
They change as all physical things.

Please read the scope notes.

It is per definitionem wrong for all CRM concept to argue with the 
meaning of the label. Labels can only be wrong wrt to the scope note. 
Per definitionem they do not constitute definitions. It is wrong to 
argue that Czechoslovakia is not a period. You may argue if 
Czechoslovakia as an E4, or if "Period" is the best label for the scope 
not of E4. It is correct to regard "Extent of Czechoslovakia 2000" as an 
instance of E53.

These are foundational principles of the CRM, hence not debatable, 
because changing them would create other "ontologies".

We have discussed and published in CRM-SIG modelling principles, which 
are under review. I kindly ask all of you that help us improving the CRM 
with your vivid interest and valuable responses, to read those before 
entering deeper philosophical discussions. We have put the principles 
now on a more visible place:

http://www.cidoc-crm.org/methodology-of-ontology-development

Unfortunately, as I see now, this principle, we have presented hundreds 
of times in meetings and tutorial, has not be formulated strong enough 
neither in the above document nor the CRM text.

What comes next in the Methodolgy is section 8.3, which is not further 
elaborated.

I therefore propose to add in the CRM, in the section Terminology, 
definition of "Class", to add an adequate variant of
"It is per definitionem wrong for all CRM concept to argue with the 
meaning of the label. Labels can only be wrong wrt to the scope note. 
Per definitionem they do not constitute definitions."

So, concluding, the solution is E4 or E27 for all those guys, life is 
easier with the CRM ;-)

I hope this makes things clearer:-)

Please contradict me;-), if necessary,

Martin



On 10/5/2019 8:40 AM, Franco Niccolucci wrote:
> Dear Dan,
>
> I am a bit scared by what you propose. Let me summarize your procedure.
>
> You have a bag of things: islands, settlements (by the way, what do you mean by settlement?), territories. They all have the same nature, and have borders separating them from the rest of the universe: some are bona fide, i.e. they are borders permanently (or almost so) involving some discontinuity or heterogeneity; others don’t, and they are called fiat borders. This classification created by Smith and Varzi in a famous paper is independent from time variability.
>
> For example, an island has bona fide borders, but they may abruptly change due to natural phenomena, and Thira is a well-known example of this. A coast may be eroded by waves, sometimes very slowly and sometimes in a way perceptible by a human. A glacier is a bona fide object because its borders are defined by the intrinsic difference between the ice and the terrain, but it changes its shape in time, being larger in the winter compared to the summer.
> In the paper by Smith and Varzi introducing such concepts, the North Sea is mentioned as a fiat object although it is reasonably stable in time; actually all fiat objects tend to be variable in time due to their social/human definition.
>
> Further, time independence is not the same as time absence: Place is a concept based on time absence. To keep the integrity of your bag content, Place should be a 4D cylinder not varying along the t-axis. According to the current CRM definition, it is instead timeless. How would you manage the above mentioned case of Thira? It starts existing as a Place, but after the eruption it becomes a Space-Time Volume?
>
> Unfortunately I have no clean solution to offer. The only escape way I see tonight is to illegally associate to every Place a Space-time volume, also called Place, which has identical time sections to the Place at any time t, from the Big Bang to the end of the universe we could say; but no CRM property exists that allows associating the cross-section of a 4D Space-time volume at a given time t0 to the corresponding 3D region, a Place. In other words, Places would (always?) be projections (P161) of Space-time volumes; when the latter does not change in time, i.e. it is a 4D cylinder, it is also called a Place.
>
> This proviso makes your distinction not illogical any more, but just illegal; which is a substantial step forward.
> Then, variability in time is a matter of granularity, and may be well chosen by you according to the scope and purpose of your modeling.
>
> I am sending you separately some considerations on Space-time volumes - which are of course available to all the interested ones. A good read for the weekend.
>
> Regards
>
> Franco
>
>
> Prof. Franco Niccolucci
> Director, VAST-LAB
> PIN - U. of Florence
> Scientific Coordinator
> ARIADNEplus - PARTHENOS
>
> Editor-in-Chief
> ACM Journal of Computing and Cultural Heritage (JOCCH)
>
> Piazza Ciardi 25
> 59100 Prato, Italy
>
>
>> Il giorno 4 ott 2019, alle ore 21:46, Dan Matei <Dan at cimec.ro> ha scritto:
>>
>> Hi friends,
>>
>> In my legacy data, beside precise geographical Places (polygons, lines, points), of course I have
>> named territories.
>>
>> Well, all Places are time-dependent: they are all post Big-Bang :-) But in order to simplify my
>> life, I am tempted to model as E53_Place the bona fide spatial objects:
>>
>> •	the "history-independent" places (e.g. Island of Crete, North America)
>> •	the settlements (yes, a brutal simplification as bona fide objects)
>>
>> and as E92_SpaceTime_Volume the fiat spatial objects:
>>
>> •	the territories of (extended) administrative units (counties..., countries, empires).
>>
>> A few territories are stable in space AND time, e.g. Czechoslovakia, almost (1918-1993, with the
>> WW2 caesura), but others... Think of the Habsburg Empire.
>>
>> Of course, there are "special" cases of almost identity, as "Malta" (the island) and the territory
>> of "the Republic of Malta" (1964-), but I could live with them :-)
>>
>> What do you think ? could that be a reasonable enough decision ?
>>
>> Dan
>>
>> _____________________________________________________________
>> Dan Matei, bibliograf
>> Institutul Național al Patrimoniului, Secția Biblioteci Digitale
>> Piața Presei Libere nr. 1, 013701 București
>> tel. 0725 253 222, 021 317 90 72, fax: 021 317 90 64
>> dan.matei at patrimoniu.gov.ro; dan at cimec.ro
>>
>>
>> _______________________________________________
>> Crm-sig mailing list
>> Crm-sig at ics.forth.gr
>> http://lists.ics.forth.gr/mailman/listinfo/crm-sig
>
> _______________________________________________
> Crm-sig mailing list
> Crm-sig at ics.forth.gr
> http://lists.ics.forth.gr/mailman/listinfo/crm-sig


-- 
------------------------------------
  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: martin at ics.forth.gr
  Web-site: http://www.ics.forth.gr/isl

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ics.forth.gr/pipermail/crm-sig/attachments/20191005/20c47174/attachment-0001.html>


More information about the Crm-sig mailing list