Issue 615: scope note of E13 Attribute Assignment
Post by Wolfgang Schmidle (4 October 2022)
I am struggling with the E13 scope note: This class comprises the actions of making assertions about one property of an object or any single relation between two items or concepts. The type of the property asserted to hold between two items or concepts can be described by the property P177 assigned property [of] type: E55 Type.
It seems that the word "property" is used here in the sense of "any property one can think of", as in "The condition assessment of the endband cores of MS Sinai Greek 418 (E14) assigned property of type damage (E55)", as well as in the sense of "a CRM property", is that right?
And I found the last paragraph particularly difficult to understand: All cases of properties in this model that are also described indirectly through a subclass of E13 Attribute Assignment are characterised as "short cuts" of a path via this subclass. This redundant modelling of two alternative views is preferred because many implementations may have good reasons to model either the action of assertion or the short cut, and the relation between both alternatives can be captured by simple rules.
- What does "this model" refer to? The CRM?
- What does the "two alternative views" refer to? When I model a Property directly or via E13?
- What are the "simple rules"? Does this mean a transformation like this?
- Ex Py Ez
- Ex P140i was attributed by E13 Attribute Assignment (P177 assigned property of type "Py") P141 assigned Ez
- Should "short cut" be "shortcut" (and possibly without " ")? (Yes, there is a place for small edits, but I am really not sure about this.)
- Do the "any CRM property" paths (via E13) that can reify any CRM property have any relationship with the paths of "explicit" shortcuts (not via E13) at all? For example:
- shortcut: E72 Legal Object P105 right held by E39 Actor
- explicit path: E72 Legal Object P104 is subject to E30 Right P75i is possessed by E39 Actor
- via E13: E72 Legal Object P140i was attributed by E13 Attribute Assignment (P177 assigned property of type "P105 right held by") P141 assigned E39 Actor
- shortcut: E1 CRM Entity P1 is identified by E42 Identifier
- explicit path: E1 CRM Entity P140i was attributed by E15 Identifier Assignment P37 assigned E42 Identifier
- via E13: E1 CRM Entity P140i was attributed by E13 Attribute Assignment (P177 assigned property of type "P1 is identified by") P141 assigned E42 Identifier
6. Or is this paragraph no longer about E13 but explicitly about the subclasses of E13, as the first sentence seems to suggest? If yes, is this sentence talking about "any CRM property" paths or the long forms of explicit shortcuts for the Properties from the following list?
- E14 Condition Assessment: P44 has condition
- E15 Identifier Assignment: P48 has preferred identifier, or P1 is identified by (if the E41 Appellation is also an E42 Identifier)
- E16 Measurement: P43 has dimension (if the E70 Thing is also a E18 Physical Thing and thus can be measured)
- E17 Type Assignment: P2 has type
In other words: No "explicit" shortcuts are mentioned for E13. Do the subclasses E14, E15, E16, E17 somehow inherit the ability to create "any CRM property" paths, or are they only used in the long forms of "explicit" shortcuts of the following properties?
- P38 "deassigned" is only for Identifiers. How would one express the general situation when an attribute gets deassigned?
(Both P37 "assigned" and P38 "deassigned" are in fact subproperties of P141 "assigned". On the other hand, according to its scope note P1 is a shortcut for "P140i was attributed by E15 Identifier Assignment P37 assigned" but the counterpart "P140i was attributed by E15 Identifier Assignment P38 deassigned" is not mentioned.)
- Both E15 and E17 can be applied to any E1 CMR Entity. Why is there a subproperty P41 "classified (was classified by)" of P140 "assigned attribute to (was attributed by)" for E17 Type Assignment, but no subproperty of P140 for E15 Identifier Assignment?
Post by Martin Doerr (4 October 2022)
Nice ISSUE to review!
I think this C should be deleted. When we wrote the third paragraph, we had abandoned the policy to generally regard properties as shortcuts of assertions by third parties, and deleted many shortcut expansion statements. We introduced the concept of a maintainer of the KB as default whose opinion the KB represents.
I think the last paragraph became obsolete by that time.
At least some subclasses of E13 thought of here are cases in which the shortcut property is produced, rather than recognized, by an explicit necessary action - the measurement, but also having an identifier.
The idea is indeed that E13 may also be used for describing properties not in the CRM.
In the 55th joint meeting of the CIDOC CRM and SO/TC46/SC4/WG9; 48th FRBR/LRMoo SIG meeting, the SIG talked about the questions raised by WS on the last paragraph of E13 Attribute Assignment. Said questions will no longer be relevant if the SIG accepts the proposal by MD to delete the paragraph.
- “All cases of properties in this model that are also described indirectly through a subclass of E13 Attribute Assignment are characterised as "short cuts" of a path via this subclass. This redundant modelling of two alternative views is preferred because many implementations may have good reasons to model either the action of assertion or the short cut, and the relation between both alternatives can be captured by simple rules.”
Decision: postpone until a proposal has formally been made
A summary of the other questions posed by WS to the SIG and the main discussion points can be found in the attached document.
Belval, December 2022
In the 56th joint meeting of the CIDOC CRM and ISO/TC46/SC4/WG9 &49th FRBR/LRMoo SIG, the SIG approved the proposal by MD & WS to deprecate the last paragraph in the scope note of E13 Attribute Assignment.
Details of the decision can be found in the attached document.
Crete, May 2023
Post by Martin Doerr (24 August 2023)
I propose to close issue 615 by e-vote. The new text for E13 Attribute assignment has been approved already. Trailing an issue into another entity is not good practice, but to answer the question:
As a rule, multiple properties of a superclass should not be specialized altogether by analogy. Properties are and must be specialized if and only if they convey a more specific meaning than the superproperty of the superclass in the context of the subclass. Obviously, there is nothing you can say about an entity that enables it to have an Identifier assigned and not only an Appellation. Conversely, there should exist at least one entity that, by its nature, cannot be assigned an identifier to.
This rule, even though general KR, may be worthwhile to be documented as another ISSUE. We had more frequently cases in CRM extensions, were properties were not specialized but should have been.