[Crm-sig] Issue 326 Resolving inconsistencies between E2, E4, E52 and E92
RSanderson at getty.edu
Wed Mar 13 02:51:46 EET 2019
Let me put it a different way…
Currently the model allows us to partition physical things according to a time-span. I can easily document how to have a new identity for the Nightwatch in its 17 ft phase, and a separate identity for it in its 14.3 ft phase. I then don’t need to document how to express the width, as it’s exactly the same pattern as the “real” object. I can do the same with any descendent of physical thing, or any descendant of E4.
This seems, I have to say, like a very easy way to model Phase and State with no additional ontological features needed. I can, today, say that there is a Person which I “contain”, and has a particular temporal projection (begin of the being April 25 2016, no end date), and p2_has_type SemanticArchitect. We could provide a label mapping of P10i_contains to something like “has_phase” and it would follow the ontology and be easily usable and understandable.
I don’t regard it as a bad application, because the ontology explicitly allows it by having E18 as a subclass of E92. Compared to introducing two new classes and a bunch of new properties, instead I can simply use functionality present in CRM base today … and I can probably live with feeling dirty because of it, knowing it’s exploiting a feature that probably shouldn’t be there.
And thus, I think the feature probably shouldn’t be there :)
From: Crm-sig <crm-sig-bounces at ics.forth.gr> on behalf of Martin Doerr <martin at ics.forth.gr>
Date: Tuesday, March 12, 2019 at 4:55 PM
To: "crm-sig at ics.forth.gr" <crm-sig at ics.forth.gr>
Subject: Re: [Crm-sig] Issue 326 Resolving inconsistencies between E2, E4, E52 and E92
I agree that this is a "non-intended" model, as Guarino describes it. He also points out, that no ontology can exclude all unintended models.
However, I do not see actually why this kind of model would be disallowed with a link. If I am not mistaken, anything that can be said with the IsA can be said with the 1-1 link. Just add the links, isn't it?
Indeed, parts can have a smaller time of existence than the whole. This is intended. To declare a whole which has no portion surviving from beginning to end of the whole is also realistic. To declare a whole as E22 which has no properties poses a question about its identity.
So, I regard the example as a bad application, not a shortcoming of the schema, and a question of elaborating the identity conditions for physical objects.
On 3/12/2019 4:27 PM, Robert Sanderson wrote:
Dear Martin, all,
I agree with your assessment into the four categories, and that the first three are met, and the last is more complicated.
I also agree with the formalism for E4. It moves some of the complexity around, and doesn’t introduce inconsistency for the temporal side of things for subclasses of E2.
However, I agree with George that this does not hold true for the other sub class of E92, being E18 Physical Thing. With this subclass assertion, we can partition physical things based on time and then make assertions about those partitions using all of the sub-classes of E18. For example, to say that the Nightwatch had a width of 17 feet between its production in 1642 and 1715 when it was trimmed to fit on a wall in the Amsterdam town hall, we could have an E22 for the painting throughout time, and use P10 to reference further E22s, each of which were clarified with P160 as to their temporal projection. These projections could then have different dimensions.
<Nightwatch> a E22_Man-Made_Object ;
P10i_contains <Large_Nightwatch> , <Small_Nightwatch> .
<Large_Nightwatch> a E22_Man-Made_Object ;
a E52_Time-Span ;
P82a_end_of_the_end “1715-12-31” ] ;
P2_as_type <width-type> ;
P90_has_value 17 ;
P91_has_unit <feet-unit> ]
(and the same for Small_Nightwatch, starting 1715 with 14.3 feet as width)
This seems antithetical to the intent of the model (as I understand it) where activities (such as Modification in this case) are kept separate from the entities that they affect.
This particular pattern could be prevented by having E92 not be a sub class of E18, without affecting the P160 / P4 discussion. However, I note some issues with making only this split:
· It would still be valuable to have the STV of a physical thing, in order to calculate the intersection between the STV that a physical object projects with Periods (that are themselves STVs). So it would be valuable to introduce a relationship between E18 and E92, introducing pattern inconsistency.
· While Period and Event seem to share the identity conditions with STV, Activity and below start to seem less identical. I worry that I become the space-time volume of the sum of my activities… and then I am a STV again, even though we removed it from E18 for just this reason.
· The same issue for P160 / P4 would apply for P161 / P53 – the spatial projection of the object is its former or current location, as they have the same identity currently.
So overall, I think my position is that for consistency of the model, E92 should not be a subclass of either E4 or E18, but instead related via a property.
Hope that helps!
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,
Email: martin at ics.forth.gr<mailto:martin at ics.forth.gr>
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Crm-sig