Wikidata:Property proposal/object of action
object of action edit
Originally proposed at Wikidata:Property proposal/Generic
Not done
Description | thing an action is done to |
---|---|
Data type | Item |
Domain | action (Q4026292) |
Example 1 | Date cultivation in Dar al-Manasir (Q5227606)object of actiondate (Q1652093) |
Example 2 | window cleaning (Q3124765)object of actionwindow (Q35473) |
Example 3 | German December 16 suprise attack in the Battle of the Bulge (Q116504918)object of actionAllies in the Battle of the Bulge (Q116511025) Source |
Motivation edit
This is the second property of an action data model. See more details at the other proposal: Wikidata:Property proposal/agent of action
This property should also help clear up some usages of of (P642) Lectrician1 (talk) 10:51, 31 January 2023 (UTC)
Discussion edit
- Support paper shredding (Q18387525)object of actionpaper (Q11472), without using of (P642), I could only link it with named after (P138). -wd-Ryan (Talk/Edits) 19:58, 31 January 2023 (UTC)
- Comment In case it helps with pondering this proposal, related concepts that come to mind are and ‘patient’ in the linguistic sense (patient (Q170212)). ―BlaueBlüte (talk) 03:06, 1 February 2023 (UTC)
- Weak oppose It would seem to me that this proposal is only a very small improvement over the use of of (P642) in that it still requires a lot of (typically human) common sense to interpret due to class–instance confusion.
Consider the examples given (note: I’m saying ‘suffering’ instead of ‘object of’ to avoid confusion with ‘objects’ as part of statements):
Example statement | ‘semantic type’ of ‘object of action’ (Pxxx), inferred from context in example | Are all implied instances (on the object side of the statement) suffering the action? | Implied meaning of the proposed property ‘object of action’ (Pxxx) |
---|---|---|---|
Date cultivation in Dar al-Manasir (Q5227606)object of actiondate (Q1652093) | instance × class | no (assuming there exist dates outside of Dar al-Manasir (Q176047)) | the things suffering the action given by subject-of-statement are of type (class) given by object-of-the-statement |
window cleaning (Q3124765)object of actionwindow (Q35473) | class × class | maybe (cleaning might occur on all windows simultaneously, or statement might be taken to be contingent on the occurrence of a specific window-cleaning) | there are things suffering an action of type (class) given by subject-of-statement, and those things are instances of the type (class) given by object-of-the-statement |
instance × instance | yes, necessarily (only one instance) | the thing given by object-of-statement is suffering the action given by subject-of-statement |
This is clearly an issue for inferencing, text generation, etc., but the only solutions I can think of would be ➊ to split this proposed property into three or ➋ to again rely on qualifiers. (Ideally I would like to see this discussed more broadly in a request for comment or similar …)
―BlaueBlüte (talk) 04:08, 1 February 2023 (UTC)
- Comment By the way, kudos, User:Lectrician1, for choosing your examples so well, almost as in anticipation of the above analysis ;) ―BlaueBlüte (talk) 04:15, 1 February 2023 (UTC)
- @BlaueBlüte Okay, I'm a bit confused as to why the explanations you provided above cause issues... Can you elaborate more on "inferencing, text generation, etc."? Lectrician1 (talk) 05:27, 1 February 2023 (UTC)
- @Lectrician1: Well, I guess this proposed property wouldn’t cause any more severe issues that are already usually caused by Wikidata being kinda vague about classes versus instances most of the time anyhow. But it might miss an opportunity to solve some of those.
To elaborate on the example of text generation: If one wanted to render the example statements from this property proposal in natural language—say, English—, one would need to know number and definiteness so as to pick the right forms of the lexemes associated with the objects in those statements. In examples 1 and 2, that would often be plural and indefiniteness: “[no article in English] Dates [plural] are being cultivated …”, “[no article in English] Windows [plural] are being cleaned …”. In example 3, that would often be unmodified number and definiteness: “The [definite article] Allies in the Battle of the Bulge [plural = number from label/lexeme] are being attacked …”. (Ignoring tense etc. here.) If this proposed property were to be split into one for each case, one could make rules for that based on which property is used. ―BlaueBlüte (talk) 05:52, 1 February 2023 (UTC)- @BlaueBlüte This could be solved with qualifying the statements with a "relative number of object: more than one" statement which is similar to what I'm proposing here: Wikidata:Property proposal/number of part. Lectrician1 (talk) 05:39, 20 March 2023 (UTC)
- @Lectrician1: Well, I guess this proposed property wouldn’t cause any more severe issues that are already usually caused by Wikidata being kinda vague about classes versus instances most of the time anyhow. But it might miss an opportunity to solve some of those.
- @BlaueBlüte Okay, I'm a bit confused as to why the explanations you provided above cause issues... Can you elaborate more on "inferencing, text generation, etc."? Lectrician1 (talk) 05:27, 1 February 2023 (UTC)
- @Lectrician1, BlaueBlüte, Wd-Ryan:, Not done, no consensus of proposed property at this time based on the above discussion. Regards, ZI Jony (Talk) 13:07, 24 January 2024 (UTC)