Wikidata:Classification noticeboard

A noticeboard for discussion of apparent issues in the classification tree -- for example, instance of / subclass of chains that appear to lead to paradoxical results; best choices of instance of (P31) and subclass of (P279) on individual items; and other such issues.

Resources: Wikidata:Item classification, Help:Basic membership properties, Help:Classification, Help:Modeling

Watford DC line <--< Event

• Watford DC Line (Q19882) appears as an instance in the class-tree of occurrence (Q1190554). -- Jheald (talk) 10:38, 6 September 2015 (UTC)

Looking at these trees, the weakest link seems to be transport (Q7590) as a subclass of motion (Q79782), putting it into the class-tree of occurrence (Q1190554).

Other oddities would include

• how Watford DC Line (Q19882) can end up in the class-trees of abstract entity (Q7184903) and of indefinite legal concept (Q1512368);
• in the "railway line" tree railway line (Q728937) surely shouldn't be in the class-tree of physical property (Q4373292),
• plus, if the class-tree of work (Q386724) is so encompassing as to include this, does that leave much distinction between work (Q386724) and artificial object (Q16686448) ?

Do people have any comments or see any good fixes to sort any of these?

(BTW, I presume that removing the classification of commuter rail service (Q1412403) as a sub-class of transport (Q7590) would be non-controversial, as it is already implied via the classification of urban rail transit system (Q3491904) ?)

Thanks in advance, Jheald (talk) 10:38, 6 September 2015 (UTC)

A couple of points to add here:

• I think that transport being a subclass of motion is probably correct actually (or at least the tree would intersect later anyway if some steps were added in between). The act of transporting something is ultimately movement/motion.
• ⟨ commuter rail service (Q1412403)    ⟩ subclass of (P279) ⟨ transport (Q7590)    ⟩
  is the bit that's wrong in my opinion. It should instead be
  ⟨ commuter rail service (Q1412403)    ⟩ subclass of (P279) ⟨ transport infrastructure (Q376799)    ⟩
   ? As transport infrastructure (Q376799) is not a subclasss of transport (it's a subclass of infrastructure (Q121359)), this takes it down the 'physical object' route rather than the 'movement/motion' route.
• Is it correct to say that
  ⟨ social space (Q7551384)    ⟩ subclass of (P279) ⟨ geographic location (Q2221906)    ⟩
  ? I think this is wrong, because we are jumping from a construction/installation to the concept of space as a physical property (besides which, a social space (Q7551384) is not always a geographic location (Q2221906) because they can also be virtual). Perhaps it's better to cut the chain there and instead say
  ⟨ social space (Q7551384)    ⟩ subclass of (P279) ⟨ Gathering place (Q5527376)    ⟩
  ? (note that Gathering place (Q5527376) has no statements at the moment so would need fitting into the tree).

Any thoughts? NavinoEvans (talk) 19:31, 7 September 2015 (UTC)

@NavinoEvans, Yair Rand: :

• What's an instance of transport ? When I went to work by bus, I was set into motion by the bus and went from the point A to the point B thanks to the bus, so I say it's correct that transport is a specific kind of motion.
• I think this is partially true. First I think that we need to take into account the fact that there is processes going on that infrastructure : trains moving, people drinving the cars ... I think this is a complex object not made only of physical object. Is not this ... a business anyway ? With a social usefulness. author  TomT0m / talk page 09:26, 16 February 2016 (UTC)
• Physical space / Numerical space : Has a space to be physical ? I think it's worth creating a "digital (social) space" anyway, something like a superclass of "internet forum". Maybe we need a superclass for all of these, but that become phylosophical :) I think we need a notion of "virtual space" for spaces that are born from humain interpretation of datas : a fictional space (that can "live" when someone read a book in thatone imagination), a video game (that lives thanks through coumputer interpretation of code and human interactions), an internet forum, ... author  TomT0m / talk page 09:26, 16 February 2016 (UTC)

Publication

• publication (Q732577) currently appears as a subclass of media (Q340169) and work (Q386724), and an instance of information (Q11028) and process (Q10843872)

Are these relations right? And also, is the item in reality currently doing double-duty for two different concepts: both a publication, as a physical object; and the process of publication, as an act of publishing? Jheald (talk) 11:04, 6 September 2015 (UTC)

Background: This came up because a combined P31/P279 TREE search placed The Guardian (Q11148) as a leaf of occurrence (Q1190554) (as reported here at project chat).

Slightly more careful searching reveals that newspaper (Q11032) is not in the strict sub-class tree of occurrence (Q1190554):

In the strict sub-class tree, "publications" are clearly considered to be physical manifestations of artifacts.

But is it correct that "publication" is also given as an instance of information (Q11028) and process (Q10843872)? Is this appropriate?

Publication as an instance of "process in engineering"

The relationship with process (Q10843872) looks quite wrong, on all counts -- wrong subject item, wrong relationship, wrong object item. (Though whoever added this statement was trying to say something valid, so IMO it is worth thinking how perhaps it should have been coded.

To start with, the object item is wrong, as becomes clear when we look at process (Q10843872) on Reasonator, and in its classification tree:

process (Q10843872) is about engineering processes that turn a set of inputs into a set of outputs. It is not about the typical activity of an engineer. Instead, the target object of the statement should probably simply be engineering (Q11023).

Secondly, the subject object is wrong, if newspaper (Q11032) is intended to be for manifested artifacts. What seems to be intended here is "publication" in the sense of an engineer publishing papers. So it looks on the face of it as if this isn't the right item for the concept. Something like publishing (Q3972943) might be more appropriate, except that that appears to relate to the activity of publishers rather than authors. Perhaps something like scientific style (Q7433830) might more accurately capture the intention. (That item's article, in English at least, does appear to relate to the activity rather than the output).

Finally, the relationship surely ought to be part of (P361), giving as a suggested final statement:

scientific style (Q7433830) ... is part of (P361) ... engineering (Q11023)

(which incidentally, if I have got this right, does not imply that all scientific writing is part of engineering -- otherwise one would have used subclass of (P279)).

Update: Done Jheald (talk) 19:35, 7 September 2015 (UTC)

If we had an item for scientific-publication-by-an-author-as-an-activity, that could perhaps be part of (P361) scientific style (Q7433830).

Incidentally,

• (i) the various realtionships of items via subclass of (P279), instance of (P31) and part of (P361) to engineering (Q11023), as revealed by Reasonator could probably use a clean-up, as at the moment they appear to be all over the place (Update: several fields of engineering moved from "part of" to the stronger "subclass of" Jheald (talk) 19:22, 7 September 2015 (UTC)); and
• (ii) engineering (Q11023) should probably be somewhere in the sub-tree of activity (Q1914636), eg (looking at the classification below) perhaps by adding the statement science (Q336) ... is a subclass of (P279) ... activity (Q1914636) (Update: done Jheald (talk) 19:22, 7 September 2015 (UTC))

printing (Q11034) is a process (Q10843872). Publication is not.

Jheald: For me part of (P361) is not a weak version of subclass of (P279). subclass of (P279) is for classes of lots of things with some characteristic in common so it links a class to a class. part of (P361) is to link one thing to a bigger thing so it links an instance to a bigger instance. An aircraft carrier is part of a fleet which contains lots of thing that are not aircraft carriers. It is an instance of the class of aircraft carriers which only contains aircraft carriers. These are very different properties. Joe Filceolaire (talk) 21:14, 7 September 2015 (UTC)

Indeed, see Help:Classification for example. Subclass means special kind of. A wheel is not a special kind of car, while sports cars are. author  TomT0m / talk page 09:09, 8 September 2015 (UTC)

Publication as an instance of information

Finally there is the relationship newspaper (Q11032) ... is a instance of (P31) ... information (Q11028)

I am not sure how to approach this. What is the relationship that is trying to be expressed here, and is it actually a instance of (P31) ? I am not confident; but perhaps a better encapsulation would be

manifestation (Q286583) ... is facet of (P1269) .. information (Q11028)

But then currently we have manifestation (Q286583) ... has part(s) (P527) .. information (Q11028) -- which is the opposite relation. So I'm really not sure what is appropriate here.

Thoughts / suggestions on any of the above would be very welcome. Jheald (talk) 13:34, 6 September 2015 (UTC)

I think that facet of (P1269) is absolutely not a subproperty of part of. author  TomT0m / talk page 15:40, 6 September 2015 (UTC)

Sequence

The classification tree above for newspaper shows rather odd behaviour around sequence, which surely can't be right:

Newspapers are mathematical objects? That can't be right. But I'm not sure what the best right way to fix it is.

The mathematical idea empasises the idea of ordering. But that's surely only one facet of the nature of a sequence of concrete objects. Therefore, should one rename "sequence" to "mathematical sequence", and then create a new general item "sequence", such that

"mathematical sequence" ... is a facet of (P1269) ... "sequence"  ?

I'm still not quite comfortable with this. On the one hand, all sequences are sequences in a mathematical sense. But on the other, sequences of concrete objects have aspects of their nature which are not entirely encompassed merely in a mathematical sequence.

I think facet of (P1269) is the right relation for "this applies to everything in the class, but doesn't encapsulate their full nature".

But I'm not sure that I've got this right, so it would be good to know how people see this. Jheald (talk) 14:08, 6 September 2015 (UTC)

Update: The en-language description of sequence (Q133250) is "ordered list of elements in mathematics", and the sitelinks mostly end in "(Mathematik)" of similar, so it seems Q133250 already is specific to mathematics.

So (it seems to me) the issue is that instead of "series" being a subclass of Q133250, instead we should have sequence (Q133250) a subclass of (P279) series (Q20937557) -- i.e. precisely the other way round.

Some might I suppose object that the mathematical notion is the more general one, in that any series of objects embodies the mathematical notion of what is a sequence. However, in terms of our articles, Q133250 is specific to mathematics, whereas series (Q20937557) and its super-class class (Q17519152) can apply to anything -- so it should be Q133250, not Q20937557 which should be the sub-class.

Making this change resolves the issue of newspapers being mathematical objects -- or, at least, it does until we get to class (Q18844919) being a subclass of set (Q36161). That will have to wait until tomorrow :-) Jheald (talk) 23:43, 7 September 2015 (UTC)

Journals are not really mathematical objects. It occurs that we can model the sequence of, say daily edition of a news paper, with a mathematical sequence (assign a number to each edition for example). So to model this I would take an approach like making them subclasses of chronologically orderable works for example. author  TomT0m / talk page 09:13, 8 September 2015 (UTC)

@TomT0m: For newspapers, journals, periodicals etc, I think we already have the notion of chronologically orderable works in serial (Q2217301)

As regards journals not being mathematical objects, I think the solution of making sequence (Q133250) a subclass of series series (Q20937557) works as a way to fix this. This got reverted last night by User:Andreasmperu, but I hope that if/when he reads the above that he will be more convinced.

The question arises as to whether there should be some property from series (Q20937557) -> sequence (Q133250) eg perhaps "has a mathematical abstraction" or "has idealisation". Jheald (talk) 21:12, 8 September 2015 (UTC)

I don't really think there actually is something to fix, so I approve the revert. A property models linking a theory to a class of phenomena of the real works seems however a good idea, something like <trait theory> models <personality>.

However in programming languages there is constructions to link something abstract like orderable sequences of object of some kind to a sequence of objects of that kind, like C++ templates or generics in Java. Some usage of the of qualifier always always reminds me something like that, for example we can read stuff like :

⟨ Harry Potter ⟩ instance of (P31) ⟨ sequence ⟩
of (P642) ⟨ Book ⟩

in Wikidata (I'm not saying it's correct as is :) ) . So there may be a topic author  TomT0m / talk page 07:56, 9 September 2015 (UTC)

Series of creative works

Incidentally, series of creative works (Q7725310) as a subclass of (P279) of work of art (Q838948) is plainly wrong.

[ -- note: when I first wrote this, series of creative works (Q7725310) was originally labelled "series" ]

Presumably the correct relation is

artwork series (Q15709879) ... is a subclass of (P279) ... work of art (Q838948)

On the other hand, most of the existing subclasses of series of creative works (Q7725310) (Reasonator) are in fact series of creative works, of various kinds.

Do people think that all of these would fit as subclasses of a artwork series (Q15709879), with the property artwork series (Q15709879) ... is a subclass of (P279) ... work of art (Q838948) -- or should a new item be made, "series of creative works" ? Jheald (talk) 14:06, 6 September 2015 (UTC)

Update It seems that at the root of this was a "language muddle". As evidenced by its sitelinks, the item (which I have now re-labelled) series of creative works (Q7725310) was specifically intended to apply specifically to creative works, as shown eg by the sitelink to de:Mehrteiler in German, and all the other languages. What was incorrect was the description "ordered set of similar objects" which somebody had added in English / French / German (and maybe more that I can't see).

I have created a new item, series (Q20937557) for this more general notion of series, to be a subclass of (P279) class (Q17519152), with series of creative works (Q7725310) a subclass of (P279) series (Q20937557).

The original statement series of creative works (Q7725310) is a subclass of (P279) work of art (Q838948) was therefore not so far off, after all. But, as noted above, it seems to me that it is more appropriate to make artwork series (Q15709879) the subclass of (P279) work of art (Q838948), if we're going to distinguish work of art (Q838948) from the more general work (Q386724)

Unfortunately I didn't realise this straight away, and I first created series of creative works (Q20936662) as a new item, before merging it back into series of creative works (Q7725310). It seems that {{Item classification}} can cope with this; but Reasonator seems to find it harder to trace back sub-classes when there are redirects -- compare [1] and [2]. I don't know whether this will continue being a problem, or whether there are bots that will eventually references in all statements to Q20936662 into references to series of creative works (Q7725310). Jheald (talk) 23:05, 7 September 2015 (UTC)

It still remains to clean up various items that point to series of creative works (Q7725310), that should probably now point to series (Q20937557) instead. Jheald (talk) 23:12, 7 September 2015 (UTC)

@Jheald: I'm not sure what you did but right now it's problematic: video game series (Q7058673) isn't only a subclass of work (Q386724) anymore. So for present in work (P1441) we now have 493 constraints violations. Please, could you repair that? --Harmonia Amanda (talk) 15:52, 9 September 2015 (UTC)

@Harmonia Amanda: Good catch, but I don't understand why this should be happening. Here's the classification tree for video game series (Q7058673):

and it clearly includes video game series (Q7058673) < series of creative works (Q7725310) < group of works (Q17489659) < work (Q386724). So any video game series should still satisfy the requirement "the value for instance of (P31) should be an item that uses subclass of (P279) with value work (Q386724) (or a subclass thereof)."

I think it may be possible that the constraint report is out of date, and this will clear up the next time it is regenerated. There was a time that I made Q7058673 a subclass of a new item series of creative works (Q20936662). I quickly merged that back to series of creative works (Q7725310), but it was a while before KrBot made the corresponding change on Q7058673 diff, and it may be that the constraint checking (like WDQ) doesn't follow redirects. So there may have been a window of a few hours, before KrBot fixed it when there was a problem; but I think the problem should now be resolved, and the constraint warnings should go away the next time the list is regenerated. Jheald (talk) 16:13, 9 September 2015 (UTC)

Ok, we'll wait a few days then! But I fear it's because video game series (Q7058673) is not only a subclass of work (Q386724) but also of series (Q20937557) (via series of creative works (Q7725310)). I hope I'm wrong because I don't see how we could fix that. --Harmonia Amanda (talk) 21:18, 9 September 2015 (UTC)

All is well! Seems resolved. Thank you. --Harmonia Amanda (talk) 05:45, 10 September 2015 (UTC)

Employment

We currently have employment (Q656365) classified as a instance of (P31) of contract (Q93288)

But given that we have a separate class item employment contract (Q1221208), would it be more appropriate to express the relationships between these items with

employment contract (Q1221208) ... is a facet of (P1269) ... employment (Q656365)

and to classify employment (Q656365) as something more general, eg

employment (Q656365) ... instance of (P31) ... social relation (Q853725),

Does this seem right? One slight anxiety, looking at the the class-tree under relation (Q930933) and Reasonator for social relation (Q853725) is that the instances and sub-classes under social relation (Q853725) seem to be dominated by quite close personal relationships, rather than broader social relationships such as "employment" -- but perhaps the closer personal relationships belong more appropriately under interpersonal relationship (Q223642) ?

I'm still quite uncertainly feeling my way in this whole area of finding the right properties between the right items to place things, so I would welcome feedback and review. Jheald (talk) 14:18, 6 September 2015 (UTC)

Update: Done Jheald (talk) 19:31, 7 September 2015 (UTC)

Anime / animated film / animated cartoon

The term anime (Q1107) refers to Japanese animation, it is very general and may include television series, feature films, television films, web series, and direct-to-video. A few days ago I noticed that Autolist queries for films returned many anime television series. Puzzled, I looked at the classification tree for anime (Q1107) and saw this statement:

• anime (Q1107)subclass of (P279)animated film (Q202866)  Wrong

This made no sense because an anime television series is not a film, so I removed the statement. User:Harmonia Amanda helpfully added a new one:

• anime (Q1107)subclass of (P279)animated cartoon (Q20535815)  Correct

This is correct since animated cartoon (Q20535815) refers to animated works in general, not only films. However, I soon noticed this other statement:

• animated cartoon (Q20535815)subclass of (P279)animated film (Q202866)  Wrong

I tried to fix it but User:Infovarius reverted my change because in the Russian language the statement is correct (discussion here). Apparently the Russian meaning of animated film (Q202866) is broader than the English-Italian-Spanish-French one. So my new proposal is:

1. Move the Russian sitelink ru:Мультфильм from animated film (Q202866) to animated cartoon (Q20535815) (broader term)
2. Remove the statement: animated cartoon (Q20535815)subclass of (P279)animated film (Q202866)

This way anime television series will stop appearing in queries for films. What do you think? Mushroom (talk) 15:43, 6 September 2015 (UTC)

I like your new proposal but as the problem is with the Russian meaning of animated film (Q202866) we need a Russian speaker to resolve this. --Harmonia Amanda (talk) 17:03, 6 September 2015 (UTC)

That also sounds very good to me. Every time I run a query for finding films I have to also exclude television programs to get rid of all the wrong results. It would be really nice to root out all the places where the television programs tree seeps into the films tree. As Harmonia Amanda has said, we do need a Russian speaker to give it the all clear though. NavinoEvans (talk) 21:14, 7 September 2015 (UTC)

Mmm and the common root to all of them would be video ? The Film item would be restricted to cinematographic artwork ? author  TomT0m / talk page 09:06, 8 September 2015 (UTC)

Ok, I'm Russian speaker, let's derive what is going on here. For me аниме subclass of (P279) мультфильм is correct. Mentioned by Mushroom "television series, feature films, television films, web series, and direct-to-video" belong to characteristics which are independent (parallel) to "animation/real camera" dichotomy. So they all can be animated. @NavinoEvans:, why don't you consider television film (Q506240) as a film (Q11424)?

I don't feel the difference between Q202866 and Q20535815, can you provide example of "A not B" and "B not A"? For me, "ru:Мультфильм" (~en:Animation or Animation film) is a film which made of many pictures. But, for example, Avatar (Q24871) I'd consider as a usual film with special effects (Q381243). --Infovarius (talk) 10:14, 15 September 2015 (UTC)

@Infovarius: Sorry that should have been clearer. No problem with that, my issue was actually with television series (Q5398426) popping up in queries for films. NavinoEvans (talk) 16:04, 15 September 2015 (UTC)

@Infovarius: Wallace and Gromit (Q691528) is an animation movie without drawings, it's made with modeling clay (Q474273). author  TomT0m / talk page 14:24, 15 September 2015 (UTC)

short film (Q24862)

For what it's worth, there are also some oddities in a couple of subclasses of short film (Q24862) -- see tree -- namely web series (Q526877) and serial film (Q7751682), which includes things like The Perils of Pauline (Q2234803). While clearly related to the idea of "short film", these aren't precisely examples of it -- similar perhaps to your issues with film and film series above.

A series of creative works (Q7725310) arguably is a work (Q386724) in its own right, and an artwork series (Q15709879) arguably a work of art (Q838948), so in those cases examples of the former probably should be returned in a search for the latter. But a film series (Q24856) should arguably not be returned in a search for a film (Q11424). I'm not quite sure where that leaves serial film (Q7751682).

Is Flash Gordon Conquers the Universe (Q3746399) a film, or a series of twelve shorts? Probably the former, I guess. But not a short film (Q24862), I think. Jheald (talk) 00:05, 8 September 2015 (UTC)

horse breed (Q3745054)

horse breed (Q3745054) is the class of horse breeds like shire horse (Q40728). Is it a subclass of Horse or a subclass of breed? Joe Filceolaire (talk) 14:31, 7 September 2015 (UTC)

Subclass of breed, I think. shire horse (Q40728) is an instance of horse breed (Q3745054), not a subclass of it. horse breed (Q3745054) and everything else that's an instance of horse breed (Q3745054) are breeds, not individual horse instances. shire horse (Q40728) is a subclass of horse, because all instances of shire horse (Q40728) are horses, and not breeds. --Yair rand (talk) 21:03, 7 September 2015 (UTC)

I agree that

⟨ horse breed (Q3745054)    ⟩ subclass of (P279) ⟨ breed (Q38829)    ⟩

is right and

⟨ horse breed (Q3745054)    ⟩ subclass of (P279) ⟨ horse (Q726)    ⟩

leads to controversy. For example, shire horse (Q40728) is a class of horses, so it is a subclass of class "Horses". How it can be (and should it be?) modelled? --Infovarius (talk) 10:05, 10 September 2015 (UTC)

How about <shire horse (Q40728) instance of (P31) horse breed (Q3745054)> and

⟨ shire horse (Q40728)    ⟩ subclass of (P279) ⟨ horse (Q726)    ⟩

? Joe Filceolaire (talk) 13:35, 12 September 2015 (UTC)

This is the way to go, see Help:Classification for a rationale. author  TomT0m / talk page 16:19, 12 September 2015 (UTC)

Seems correct to me, but maybe somehow redundant (2 properties about class "horses"). --Infovarius (talk) 10:09, 14 September 2015 (UTC)

@Infovarius: Don't get what you mean, can you be more specific ? author  TomT0m / talk page

This is the way to go imho, per the rationale presented in Help:Classification. author  TomT0m / talk page 11:07, 14 September 2015 (UTC)

abstract entity (Q7184903)

(Class tree) Needs some attention as to what is properly a instance of (P31) as opposed to a subclass of (P279) of this. Jheald (talk) 23:28, 8 September 2015 (UTC)

occurrence (Q1190554)

(Class tree) The top of this class tree seems particularly confused -- I think it was a bit of a dumping ground in the GND classification, and that maybe has been imported here. Jheald (talk) 23:36, 8 September 2015 (UTC)

This needs some splitting up. natural phenomenon (Q1322005) and subclasses fits for some of these. (Maybe also physical phenomenon (Q1293220). The top of that tree has some ambiguities between phenomenon (Q483247) and phenomenon (Q16722960), so I don't know where the boundaries for these items are.) I don't know if there is an item for general calendar event, but there probably should be. Some things can be subclasses of action (Q4026292), intentional human activity (Q451967), or activity (Q1914636) (ambiguous boundaries again...). I don't know what the generic term for any legal occurrence is, but there probably is one that includes a lot. There is incident (Q18669875), which currently has no English description but might be usable as a class if well defined. Maybe also social event?

And then the issue of remaining subclasses that don't belong anywhere in the event tree: evil (Q15292), did not finish (Q1210380), failure (Q1121708), nudity (Q10791), and poverty (Q10294). I have no idea how to categorize any of them. --Yair rand (talk) 06:38, 27 October 2015 (UTC)

"I don't know if there is an item for general calendar event, but there probably should be." A subclass of periodic event then. holiday (Q1445650) and annual event (Q18574946) seems to fit pretty well. author  TomT0m / talk page 08:53, 27 October 2015 (UTC)

evil (Q15292)    seems to me a kind of "morality classification" of event. poverty (Q10294) is interesting, it seems more a state of someone or of a social class compared to another. Is this more like a charactérisation of "poor people" that an event per se, even if we can have a set of events or period when anyone is judged "poor" by some rules. I'd say there is a social class "poor people", and that "poverty" is the name of the property the people fulfills to be classified in this class. A metaclass like "poverty charaterisation" or "poverty definition" would be useful. But of course someone can be poor and later rich, so this is a function of time. There is also "poverty periods" then who fits more easily in the "event" class tree. author  TomT0m / talk page 08:53, 27 October 2015 (UTC)

culture (Q11042)

See  . Some cultures are listed as instance of (P31) culture (Q11042), some as subclass of (P279) culture (Q11042). Unclear which is correct. I'd lean towards instance of (P31), since that's how we'd typically say it, at least. "French culture is a culture." --Yair rand (talk) 21:12, 19 October 2015 (UTC)

Yair rand: I would say subclass of (P279) in these cases. An 'instance of (P31):culture (Q11042)' would be something a lot more specific than 'French culture'. One specific traditional dance; one specific cooking method; one specific style of traditional headdress: these could each be an instance of (P31) 'Gascon culture' (for instance) which would, in turn, be a subclass of (P279) of 'French culture' etc. Joe Filceolaire (talk) 04:10, 25 October 2015 (UTC)

I think it could be more clear if we have more specific items like "Gascon's dance", "Gascon's story", "Gascon's event", and if we decide that "Gascon's culture" should be aliased "Gascon's token/instance" everything should become clear.

However what's clear we must avoid however is having a boundary cross beetween the instance/class/metaclass structure, like for example both an instance of an irish famous instance of a feast and the itself, like "New years eve 2000" and "New years eve" both beeing instances of the cultural item. So indeed I think "culture" is by essence a metaclass : it refers not to instances like "New years eve 2000" but onto things that are transmitted, by definition of culture, but to "New year eve", the feast we repeat every year because our ancesters also did.

Another option if the token/class/metaclass principle actually does not always work (we must find other examples) would be to create a property like "associated culture" with no real constraint in case of doubt. author  TomT0m / talk page 08:10, 25 October 2015 (UTC)

• A specific traditional dance should have a specific field "culture" pointing to "French culture", we should not abuse instance of (P31) for that. A dance is an instance of Dance, not of Culture.
  • Please also note that the Getty aat:300264088 "Styles and Periods Facet" has 5.5k entries, mixing culture, ethnicity, archaeological site, style, period, movement... I think they have a good reason for mixing it that way, since the boundaries between these things are not always clear-cut --Vladimir Alexiev (talk) 15:43, 18 November 2015 (UTC)
    • @Vladimir Alexiev: This is exactly why instance of (P31) and subclass of (P279) are exactly what we need : we just classify cultural object. If what we call a culture is a class of object, everything fits pretty well. The "class/token" distinction works very well in this case as in others. Objects like "French culture" or "Scottish Culture" can then be (metaclasses)-classed as something as a "national culture" if needed. author  TomT0m / talk page 12:58, 19 November 2015 (UTC)

More thoughts about culture (Q11042)

@Yair rand, TomT0m, Vladimir Alexiev: I've thought about this for a week and come to the conclusion that instance of (P31) should never be used with culture (Q11042). Instead we should use indigenous to (P2341).

• a dance is an <instance of:dance> or maybe <instance of:'traditional dance' or 'couples dance' or 'polka'>
• it is also <indigenous to (P2341):Romani people (Q8060)' or 'Scotland (Q22)>. (P2341 can be used with locations or with ethnic groups.)

At least that is what I think. Joe Filceolaire (talk) 08:40, 25 November 2015 (UTC)

Where can we see the discussion about the property creation ? It seems I missed it. I tend to agree but I'm puzzled by the fact you only argument by "I thought a lot about it". A little bit more details about the reasons would be welcome :) author  TomT0m / talk page 09:36, 25 November 2015 (UTC)

I've read the prop discussion page and it doesn't provide a lot more justification. But the case is very clear: we need a specific property (like indigenous to (P2341)), we can't use instance of (P31) to link things to Culture. --Vladimir Alexiev (talk) 08:27, 8 December 2015 (UTC)

object (Q488383)

What exactly is an object? What's the difference between this and an entity? Can any miscellaneous thing be a subclass/instance of object? --Yair rand (talk) 15:40, 19 November 2015 (UTC)

I guess the big dichotomy at sake is "temporal" / "physical" here : an e event involves physical objects, usually. The french description says Both are entity. But the fr and en seems inconsistent, as "object" seems to be possibly abstract in the en def while in fr it is supposed to be a concrete entity ... author  TomT0m / talk page 15:58, 19 November 2015 (UTC)

We don't deal with the meaning of words here. We deal with the scope of application of wikidata items - in this case item Q488383.

To find out what an item is look at the statements it has and the statements that use this item (see 'what links here' in the side menu).

These tell us that object (Q488383) is a class of different meanings. The English Wikipedia articles gives an overview of a lot of these meanings and the widely different items that are <instance of (P31):object (Q488383)> backs up the idea that this item is a broad class. Note that having a broad, poorly defined class is useful for wikidata for classifying the narrower more specific classes for "Object".

This means that if you are looking for exact meanings then you need to create items with exact meanings which are <subclass of (P279)|object (Q488383). Many of these items will also be called "object" but will have statements making it clear what their domain of application is limited to.

One of the specific meanings does make a disctinction between 'Object' and 'Entity'. In wikidata a claim has a 'subject' (the page the claim is on) and an 'object' (the item the claim links to - if the property links to an item). Both the 'subject' and the 'object' are 'items'. Wikidata 'items' are a subclass of wikidata 'entities' (properties are also wikidata 'entities'). This meaning of 'object' is different from Q488383.

In short the exact meaning of item Q488383 is a broad class which includes many specific meanings of "Object". Hope this helps. Joe Filceolaire (talk) 09:01, 25 November 2015 (UTC)

@Filceolaire: I was not talking about the meaning of the word, I was talking of the scope of the Wikipedia article. Wikipedia articles are a reference to define an item that should be reflected on the item definition and usage, overwise this is an interwiki conflict and there might be a lot of confusion in interlanguage communication. author  TomT0m / talk page 09:40, 25 November 2015 (UTC)

case of death (Q2438548), death (Q4)

There are many items about deaths of specific individuals that have instance of (P31) one of these two items, and it's not clear what the difference is. I imagine one of them should be subclass of (P279) the other. Another issue is, what should the upper level be a subclass of? Currently they have occurrence (Q1190554) and property (Q937228), respectively. Is there any more specific item for a kind of change in status that these imply? --Yair rand (talk) 03:10, 26 November 2015 (UTC)

case of death (Q2438548) has one linked article, in German. death (Q4) has 125 linked articles, including one in German. I think you probably need to read the german articles to understand the distinction that German wikipedia is trying to make here. I suspect that most of the links to case of death (Q2438548) should link to death (Q4). Sorry this probably doesn't help. Joe Filceolaire (talk) 22:30, 26 November 2015 (UTC)

Classification of occupation

Right now an occupation is a biological process and a human behavior. Does that seem correct to you ? I don't really find something really convincing to back this up in https://en.wikipedia.org/wiki/Occupation for example. A second question is "what about the class of person who have this occupation" ? We often in everyday life say "he is a doctor" for example. Do we choose to give this up on Wikidata ? author  TomT0m / talk page 09:06, 16 February 2016 (UTC)

prose and other form of languages

What about the classification of prose (Q676) ? author  TomT0m / talk page 11:14, 7 March 2016 (UTC)

Event

occurrence (Q1190554) is currently a subclass of era (Q6428674) -> time interval (Q186081) -> interval (Q185148) -> set (Q36161) -> mathematical object (Q246672). These relations seems weird to me, saying basically that all instances of an event (e.g. war, sports competition) are a mathematical object. --Pasleim (talk) 16:40, 12 September 2016 (UTC)

The concept of an "era (Q6428674)" has a lot of associations that would make it too specific to be a superclass of occurrence (Q1190554), I think. mathematical object (Q246672) for time-space ranges in general sounds about right, though. --Yair rand (talk) 02:50, 13 September 2016 (UTC)

Sites

Do we have any equivalent of BFO's "site" class, as in spatial regions demarcated by other objects? Currently, the items hole (Q20204151), cavity (Q28367692), and others are listed as being subclasses of absence (Q19829125) and other classes, which seems incorrect. --Yair rand (talk) 16:51, 2 January 2020 (UTC)