Issue 675: missing part of type
Post by Martin Doerr (16 October 2023)
Dear All,
In the discussion about typed negative properties, I have the impression that a property: "misses part of type" may be utterly useful for finding archaeological object in a global search, such as the head or arms of a statue, characteristic elements of buildings etc. Admittedly, it poses the question where to stop the non-existence, and what missing parts would have a chance to be found. Would a part lost by accident be a part removal? Would that be an alternative way of documenting missing parts?
Opinions?
Best,
Martin
Post by Oyvind Eide (17 October 2023)
Dear Martin,
E80 Part Removal has E7 Activity as a superclass, thus, a part removed without
any agent being responsible is not intended it seems.
As to accidents made by agents: we would have to distinguish between
intentional act and accident in order to say accidents are not part removals.
Is that necessary?
As for parts missing: it looks useful to me (to be confirmed by those
documenting archaeological objects), but it will then establish a norm for how
a object should be or have been. This statue had a head (as documented in ...)
or this statue belong to a type of objects that has heads (as documented by ...)
All the best,
Øyvind
Post by Athina Kritsotaki (17 October 2023)
Dear Martin,
maybe I misunderstood, but how can we explicitly know thw circumstances of leading to this state, described by the property? what I mean is, that this property seems to me related to the definition of situations and to inference (how can we assert the validity of missing parts? and what about the FOL? can it support it? It seems useful but isn't it a kind of inference? just a question or maybe I am missing something
Athina
Post by Christian-Emil Ore (18 October 2023)
Dear all,
As Øyvind point out, the part removal class is a subclass of activity and thus
all instances represents "actions intentionally carried out by instances of E39
Actor ". An accident, landslide, earthquake etc is an instence of E5 Event but
not of A7 Activity.
The negative statement "missing a part (arm, title) is based on a belief that
the object in question once was complete. So this should perhaps be modeled in
CRMinf?
Best,
Christian-Emil
Post by Martin Doerr (20 October 2023)
Dear All,
Indeed, I see two characteristic cases:
A) broken surfaces:
This is characteristic for statues, which miss heads or limbs, but also for architectural elements. The Roman statues without heads have characteristic places where to place the head. There is the reasoning that people hardly produced a statue with a broken-off arm in antiquity. These parts have not been discrete before being broken of. In other cases, there may be traces of mortar or other cement to the connected component, or damaged joining features, such as corrupted screw holes etc.
B) If an object is found in a context of /use/, rather than in a /factory/, we can assume that it contained all essential components.
I agree with Oeyvind that a part removal is not adequate for a deterioration happening when some objects down etc. Therefore I raised the issue, because there is no obvious workaround in CRM currently.
The property should be used when there is enough plausibility that the object was complete. I do not assume someone went to a battle field with a chariot without wheels. Even if, the cases are so marginal they are irrelevant for the purpose of the CRM.
See also our paper, in which we analyzed a lot of situations:
DOI:10.1007/3-540-45581-7_31 <https://doi.org/10.1007/3-540-45581-7_31>
Corpus ID: 46464138
A Metamodel for Part - Whole Relationships for Reasoning on Missing Parts and Reconstruction M. Doerr <https://www.semanticscholar.org/author/M.-Doerr/38587181>, D. Plexousakis <https://www.semanticscholar.org/author/D.-Plexousakis/1705358>, C. Bekiari <https://www.semanticscholar.org/author/C.-Bekiari/2861757> Published in International Conference on Conceptual Modeling <https://link.springer.com/conference/er er> ER 2001: Conceptual Modeling — ER 2001 <https://link.springer.com/book/10.1007/3-540-45581-7> pp 412–425
Best,
Martin
Post by Thanasis Velios (20 October 2023)
Dear all, Martin,
Looking for things without certain types of features is indeed very useful. NTP46 and NTP56 are meant to do exactly that (if you have the time check https://www.semantic-web-journal.net/content/typed-properties-and-negati... and also here: https://github.com/linked-conservation-data/crmntp )
I agree with Athina, if I understand Martin's point correctly, the proposed property "misses part of type" brings the added semantics of the part being there originally on purpose. I do not think it is necessary to introduce such a property for two reasons:
1) In terms of retrieval, being able to juxtapose
E18 Physical Thing → P46 is composed of → E18 Physical Thing
E18 Physical Thing → TP46 is composed of physical thing of type → E55 Type
E18 Physical Thing → NTP46 is not composed of physical thing of type → E55 Type
is enough for typical research scenarios. Also my understanding is that from P46 or P56 we cannot deduct any intentionality during production. It is just a statement about parts.
2) One can point to technique types or design and procedures to express the original elements of the object if necessary.
The choice of the type of things that are marked as non-existing depends on the expertise of the observer, the mini closed world that they decided to set, any existing types that are known etc. It is a question of types, not of properties, right?
For damage, I often use S18 Alteration from CRMsci to express things being changed without an agent. Isn't that appropriate?
All the best,
Thanasis