Wikidata:Property proposal/not

not subclass of

Originally proposed at Wikidata:Property proposal/Generic

   Not done

Description	item may be confused for being a subclass of this class
Data type	Item
Domain	item
Example 1	Pepsini (Q10326375)nothawk (Q846664)criterion used (P1013)name (Q82799)
Example 2	strawberry (Q14458220)notberry (Q13184)criterion used (P1013)misnomer (Q6875856) (botanical sense)
Example 3	Buffalo wings (Q1072203)notbison meat (Q58173066) (name refers to the place, not sure what to use as criterion)
Example 4	ulnar nerve (Q254580)notbone (Q265868)criterion used (P1013)misnomer (Q6875856)applies to name of subject (P5168)funny bone
See also	different from (P1889), negates property, does not have characteristic (P6477)
Wikidata project	WikiProject Properties (Q60008075)

not instance of

Originally proposed at Wikidata:Property proposal/Generic

   Not done

Description	item may be confused for being an instance of this class
Data type	Item
Domain	item
Example 1	Goat Simulator 3 (Q112435648)notthreequel (Q115768274)
Example 2	St. Peter's Basilica (Q12512)notmother church (Q19361673)criterion used (P1013)common misconception (Q25345174) (List of common misconceptions)
Example 3	EN 301 549 V3.1.1 (Q107313717)not instance ofharmonised European standard (Q115707781)
Example 4	🏴‍☠️ (Q100587671)not instance ofemoji regional indicator sequence (Q28840786)
Example 5	Sorgenfrey line (Q117921997)not instance ofsecond-countable space (Q1363919)
See also	different from (P1889), negates property, does not have characteristic (P6477)
Wikidata project	WikiProject Properties (Q60008075)

Motivation

The different from (P1889) property implies the two items could be confused for each other, according to its property constraints, while not would only be used when one item could be confused as being a subclass or instance of the value. It can be thought of as "this item is sometimes incorrectly thought to be a type of..."

Adding this information will be useful for clearing up misconceptions when different from (P1889) isn't applicable, without adding it to the item's description. Open to suggestions on a different name. -wd-Ryan (Talk/Edits) 04:30, 10 December 2022 (UTC)

Discussion

•  Comment such statements should just have a deprecated rank with the reason for deprecated rank (P2241) qualifier, e.g. strawberry (Q14458220)subclass of (P279)berry (Q13184)reason for deprecated rank (P2241)misnomer (Q6875856). --Push-f (talk) 05:55, 10 December 2022 (UTC)

  @Push-f: That works for strawberry maybe, but the others? There must be thousands of these, do we really want things like a deprecated Pepsini (Q10326375)subclass of (P279)hawk (Q846664) everywhere? -wd-Ryan (Talk/Edits) 15:48, 10 December 2022 (UTC)

  There is absolutely no problem with having many deprecated statements spread across many items. I find deprecated statements to be vastly superior to a "not" property because of two reasons:

  1. It actually tells you which statement specifically should not be made.
  2. It is a value of the actual property, which means it will be more likely found and considered by humans and bots adding values to the respective property.

  You do realize that we have over 1,500 properties of the Item datatype? Should we have to create an "not X" property for every property "X" just to express that this statement does not apply? Surely not.

  I have not seen a single reason why we should not use the deprecated rank for what it is intended to be used.

  --Push-f (talk) 04:52, 11 December 2022 (UTC)

• I would prefer having subclass and instance not together in this property but rather have either one of them or both of them as separate properties. ChristianKl ❪✉❫ 21:00, 10 December 2022 (UTC)

  My name for it was simply "not" but it was changed. There are probably a lot less instance cases than subclasses. I'd be fine with making them separate properties if others agree. -wd-Ryan (Talk/Edits) 21:31, 10 December 2022 (UTC)

  @ChristianKl: Done! -wd-Ryan (Talk/Edits) 17:40, 15 December 2022 (UTC)

• WikiProject Properties has more than 50 participants and couldn't be pinged. Please post on the WikiProject's talk page instead. -wd-Ryan (Talk/Edits) 15:56, 15 December 2022 (UTC)
•  Support @Push-f I don't like using rank of statement because a depreciated rank does not mean that the statement is not valid. If it does, you would need to specify it with reason for deprecated rank (P2241). This is bad for data consumers because then they have to filter out statements when querying through classes that don't have a particular reason for deprecated rank (P2241) qualifier. That's really annoying to do. If a statement is present on an item, it should mean that statement is at least partially true at least some of the time. So, subclass of (P279) statements should be actual superclasses of the class. I'm for this property because it actually separates classes that are not superclasses and ones that are (ones in subclass of (P279)). Lectrician1 (talk) 16:26, 15 December 2022 (UTC)

   Conditional support Fair enough. I can get behind this proposal if it is split up into two properties "not instance of" and "not subclass of" ... this way we can actually properly link the properties together via negates property . --Push-f (talk) 16:28, 15 December 2022 (UTC)

  @Push-f I don't think that's needed. Technically, "subclass of" and "instance of" have a parent property called "is of class". This property would be the negation of that property. To handle cases like these where we don't have an actual superproperty, I've been thinking about proposing a "subproperty of item" property that would allow us to make superproperties as items without actually creating them as properties. Then we could make is not of classnegates item propertyis of class (of course, we'd need a "negates item property" as well to do this). Lectrician1 (talk) 16:50, 15 December 2022 (UTC)

  I can rename it to "is not of class" or split them, I'm fine with either. -wd-Ryan (Talk/Edits) 17:01, 15 December 2022 (UTC)

  What are you talking about? Such a "is of class" property does not exist and neither do I see any reason why it should. And no I don't think we should needlessly complicate the negation situation. --Push-f (talk) 17:06, 15 December 2022 (UTC)

  I decided to split them for negates property, but I need another example for instance of. -wd-Ryan (Talk/Edits) 17:30, 15 December 2022 (UTC)

  I added another example. Btw. I don't get the first example of "not instance of" ... I don't think that's something that should be stated on every thing that has a 3 in its title. --Push-f (talk) 17:47, 15 December 2022 (UTC)

  We may need an item for "third instance of a series", but Goat Simulator 3 is actually just the second one and there is no Goat Simulator 2. Perhaps not necessary but it is a confusing name which the property aims to clarify. -wd-Ryan (Talk/Edits) 17:50, 15 December 2022 (UTC)

  "is of class" property does not exist

  @Push-f It semantically does though. Maurice Theodore James (Q19337600) and human (Q5) are of class mammal (Q110551885), are they not? Lectrician1 (talk) 17:31, 15 December 2022 (UTC)

  I don't think so. One is an instance of the class and another one is a subclass of the class. You know what else can be "of a class"? Opposites. There are opposites of a class, should they also be subproperties of "of a class"? Surely not. It just doesn't make any sense. --Push-f (talk) 17:50, 15 December 2022 (UTC)

  I don't think so. One is an instance of the class and another one is a subclass of the class

  But they are both under the same class. The only reason they are separate properties is is so that one can indicates its entity is an instance while the other indicates that its entity is a class.

  You know what else can be "of a class"? Opposites. There are opposites of a class, should they also be subproperties of "of a class"? Surely not. It just doesn't make any sense.

  Can you elaborate and give an example? Lectrician1 (talk) 20:17, 15 December 2022 (UTC)

•  Oppose; use ranks instead —MisterSynergy (talk) 19:55, 15 December 2022 (UTC)

  @MisterSynergy did you see my comment about ranks in my support vote? Lectrician1 (talk) 20:17, 15 December 2022 (UTC)

  Yes I have. However:

  • Ranking can be considered as a visibility controller. If you query the right way, you do not get any claims with deprecated rank in the results (and otherwise you would need to filter for them anyways).
  • The (deprecated) rank itself does not have a particular semantic and it can be used for several scenarios. We use the reason for deprecated rank (P2241) qualifier to denote why a claim in question needs deprecated rank, and "incorrect value" is a perfect use case for this one.

  If we separate into "subclass of" and "not subclass of" (and similarly for P31), it would not need much time and we will see both properties with identical values on the same item page; pretty much as we see a lot of misuse of P279/P31 claims with identical values, even though they are always displayed right next to each other. The ranking system is much better suited for this problem in my opinion. —MisterSynergy (talk) 20:26, 15 December 2022 (UTC)

  I brought this up before, but does a deprecated Pepsini (Q10326375)subclass of (P279)hawk (Q846664) really seem appropriate? This property can also negate parent classes that may have been inherited in the subclass tree. -wd-Ryan (Talk/Edits) 20:45, 15 December 2022 (UTC)

  Ranking can be considered as a visibility controller. If you query the right way, you do not get any claims with deprecated rank in the results (and otherwise you would need to filter for them anyways).

  Could you make an example query you are suggesting people use that would avoid depreciated statements with whatever reason for deprecated rank (P2241) you want to use? For example, a query that gets all the direct subclasses of all music genres while avoiding whatever depreciated subclass of (P279) statement you will use? Lectrician1 (talk) 23:59, 15 December 2022 (UTC)

• The first example in "not instance of" doesn't make sense to me. Of course Goat Simulator 3 (Q112435648) is not an instance of third (Q28469713), because third (Q28469713) is not something that can have any instances of it. --99of9 (talk) 04:28, 19 December 2022 (UTC)

  Fixed with threequel (Q115768274). -wd-Ryan (Talk/Edits) 17:12, 19 December 2022 (UTC)

•  Oppose per MisterSynergy. Per Help:Ranking, "deprecated rank is used for statements that are known to include errors (i.e. data produced by flawed measurement processes, inaccurate statements) or that represent outdated knowledge (i.e. information that was never correct, but was at some point thought to be)." I think Lectrician1's understanding of rank is faulty. --Tagishsimon (talk) 20:51, 3 January 2023 (UTC)

  way too broad. everything is not an instance of almost everything else. BrokenSegue (talk) 00:39, 31 January 2023 (UTC)

  Everything doesn't have almost every quality, or have almost every part. -wd-Ryan (Talk/Edits) 01:51, 31 January 2023 (UTC)

•  Oppose Too broad, no limit on usage. If a statement does not exist, well then it should be assumed it is not true. For these reasons I'm against all "not" properties. Lectrician1 (talk) 15:41, 26 February 2023 (UTC)
•  Support. I need this very property for a different reason. Math students often want to find a counterexample to the claim of the form "All A is B". We have nilpotent group (Q1755242) which is a subclass of solvable group (Q759832) but not "solvable group that is not nilpotent". This is fine. But we're not confident when we meet statements like Sym(3)instance of (P31)solvable group (Q759832) to conclude that Sym(3) is not nilpotent – maybe the claim is just not made sharp enough. A statement like Sym(3)not instance ofnilpotent group (Q1755242) would clarify the answer. I agree in many cases it is better to use instance of (P31) statement with deprecated rank, but there is some other cases as I described, where it is a pure and notable fact that something does not belong to something else, we should have a "not instance of" statement with normal rank. 慈居 (talk) 17:44, 2 June 2023 (UTC)

  I added an example to explain why I need this property. The Sorgenfrey line (Q117921997) is by no means a second-countable space (Q1363919) and it has never been considered one. This is better than the rank solution considering how it is confusing otherwise when we add qualifiers such as

  • Sorgenfrey line (Q117921997)instance ofsecond-countable space (Q1363919)proved bysome mathematician
  • Sorgenfrey line (Q117921997)instance ofsecond-countable space (Q1363919)proved usingsome theorem or technique

  慈居 (talk) 18:54, 2 June 2023 (UTC)

• @Wd-Ryan, MisterSynergy, Tagishsimon, Lectrician1, 慈居, Push-f:  Not done, no consensus of proposed property at this time based on the above discussion. Regards, ZI Jony ^(Talk) 04:54, 25 January 2024 (UTC)