Talk:Q35120

Highest-level item

When classifying things, it's helpful to have a root node, a highest-level thing that all other things are said to derive from, i.e. be more specific classes or instances of. This item -- entity (description: thing) -- seems like the best candidate for that root. Given that, I don't think it makes sense to classify this item as a type of another item. Emw (talk) 14:29, 17 March 2013 (UTC)

What item was Entity classified as a type of? I don't see such a statement/link. 129.42.208.182 17:40, 6 March 2014 (UTC)

This item is not a type of anything, and so should not have any instance of (P31) nor subclass of (P279) statements. It is the most general concept, often referred to as the "top concept" -- $\top$ -- in literature on knowledge representation and description logic. It has the semantics of owl:Thing. Emw (talk) 02:07, 7 March 2014 (UTC)

It's controversial, see Help:Classification. My opinion is that this item can totally be

⟨ entity ⟩ instance of (P31) ⟨ class ⟩

. I class is well defined. TomT0m ^(talk) 12:02, 9 July 2014 (UTC)

Regarding the "controversial" nature of setting 'entity' (aka 'thing', 'owl:Thing') as the top concept in an ontology, please cite any ontology, anywhere, that does differently. The OWL specification describes owl:Thing as follows: "Two OWL class identifiers are predefined, namely the classes owl:Thing and owl:Nothing. The class extension of owl:Thing is the set of all individuals. The class extension of owl:Nothing is the empty set. Consequently, every OWL class is a subclass of owl:Thing and owl:Nothing is a subclass of every class (for the meaning of the subclass relation, see the section on rdfs:subClassOf)." Help:Classification currently mentions nothing about entity (permalink).

The rest of your post is too vague and filled with grammatical errors to understand. Emw (talk) 12:48, 9 July 2014 (UTC)

@Emw: I put words on Wikidata_talk:Statistics:Subclasses_and_instances. TomT0m ^(talk) 13:32, 9 July 2014 (UTC)

@Emw: @TomT0m: I believe Emw is right with this one in terms of this being currently used as a root class and therefore should not be a type of anything. However, I think "Entity" is a weird root. I think concept (Q151885) is a truly fundamental... well concept. Both philosophically and logically. There is no objective reality that we can perceive, there is only our mind's perception. A fundamental unit of that perception is a concept which is a product of the process of conception. This entire website is a database of collective human conception, not a collection is objective reality. Science and all it's products are just agreed upon and stabilized concepts that are meant to correspond with the collective and synchronized perception of objective reality.

Basically, Entity and Object seem like good roots, the problem is that they are both classifiable as concepts and thus not fundamental. Another way to put it, a concept is not an object. A concept is not an entity. They are both concepts however. Also, an entity implies existence and realness where all concepts are not real. Not all concepts exist or can exist. For example an infinity cannot exist in reality, but it is a mathematical concept. Slight0 (talk) 09:16, 30 April 2021 (UTC)

Instances of something else than entity

@Emw:

Query	Count	Description	Link
claim[31]	9,966,008	instance of something	[1]
claim[31:(tree[35120][][279])]	8,138,188	instance of entity or of one of its subclasses	[2]
claim[31] and noclaim[31:(tree[35120][][279])]	Autolist error; no value provided	instance of something but not of entity or one of its subclasses	[3]
claim[279]	40,741	subclass of something	[4]
claim[279:(tree[35120][][279])]	32,608	subclass of entity or one of its subclasses	[5]
claim[279] and noclaim[279:(tree[35120][][279])]	8,133	subclass of something but not of entity or one of its subclasses	[6]
claim[31] and claim[279]	8,505	subclass and instance of something	[7]

I would think in a proper system the first two quantities should be equal. Tamawashi (talk) 12:55, 5 July 2014 (UTC)

Date	WDQ1	WDQ2	WDQ3	WDQ4	WDQ5	WDQ6	WDQ7	User
2014-07-06 10:08 UTC	9,739,500	7,922,543	-	39,356	31,659	7,697	8,239	Tamawashi (talk) 10:09, 6 July 2014 (UTC)
2014-07-07 11:59 UTC	9,981,472	9,937,031	44,441	41,189	33,839	7,350	8,548	Tamawashi (talk) 12:01, 7 July 2014 (UTC)

Tamawashi, thanks, interesting analysis. I expanded the table to include the same data for classes. There are 8,133 unrooted classes and 1,827,820 unrooted instances. Rooting those classes would root their instances. Emw (talk) 16:59, 5 July 2014 (UTC)

Emw - thanks, interesting too. When looking into the unrooted classes I found some that are also instances. Therefore I added "claim[31] and claim[279] || 8,505". Does W3C recommend against having items in the latter set? Tamawashi (talk) 17:22, 5 July 2014 (UTC)

Tamawashi, no, W3C actually made the ability to model something as both a class and instance a new feature in OWL 2 DL. Modeling a class as an instance is called metamodeling, and is enabled by a technique called punning. Major ontologies I've seen, e.g. ChEBI, do metamodeling implicitly -- that is, they only use subclass of to classify things like hydrogen (e.g. hydrogen subclass of atom), and let reasoners like those in Protege implicitly convert classes to instances as needed for inference.

A common argument for explicit metamodeling on Wikidata is that it enables us to determine what type of type something is in a concept hierarchy. The canonical example is biological taxonomy, where we could say "human subclass of Homo", "human instance of taxon". However, in other domains, metamodeling easily leads to awkward constructs like "atom type class" when classifying chemical elements (more here). So while it might make sense for certain domains like biological taxonomy, I think explicit metamodeling is often unhelpful. Emw (talk) 18:25, 5 July 2014 (UTC)

Or even 'Human' 'subclass of:Homo'; 'Human' 'instance of:Species'; 'Species' 'subclass of:Taxon' (or is 'Species' an 'instance of:Taxon'?).

Tamawashi: There has been some discussion on wikidata about High Level Ontologies and I don't think wikidata has a consensus that entity (Q35120)    is the root of our ontology. Personally I can't see why we should only have one High Level Ontology. We could reproduce all the various High Level Ontologies (with qualifiers to say that A is subclass of B according to Ontology C but is subclass of D according to ontology E) all leading up to 'subclass of:High level ontology root item'. Remember that the english wikipedia category system has three different root categories and the english wikipedia portals have a a root portal which is different from the categories.

While we are waiting for our high level ontology to be resolved we can carry on building the low level ontologies upwards. From your statistics it looks like we are making good progress with this. Filceolaire (talk) 19:54, 5 July 2014 (UTC)

By the way - One of the problems with using any of the existing high level ontologies is that, as far as I could see when I looked into this, they all have some copyright restrictions. None are CC-0 so there could be problems reusing any of them here. Filceolaire (talk)

Filceolaire, the top item for all the external top items still would be Q35120. Tamawashi (talk) 20:34, 5 July 2014 (UTC)

@Emw: - OK, I see Albert Einstein (Q937) is/was an instance of human (Q5) and of all upward classes, but not of taxon, since that is connected to human via "instance of". Looks useful if one wants to make some statements about classes. Another special case for which I know no query, are items where subject of and instance of have the same values, e.g. Wikisource is instance of A and B and subclass of A and B. Tamawashi (talk) 20:40, 5 July 2014 (UTC)

Tamawashi: Yes. In fact we don't, in general, use 'instance of' and 'subclass of' for taxons. We use the specialist properties taxon rank (P105) and parent taxon (P171) instead though 'instance of:taxon' can be used with taxon rank (P105) so that items have an 'instance of' statement for use by tools that depend on that property. I should have been more specific in my statement above. Sorry. Filceolaire (talk) 21:42, 5 July 2014 (UTC)

Note that while autolist2 times out on the third query, the actual query works, it just takes a minute. Need to improve that. Meanwhile, 1827524 items that are an instance of something that's not an entity. --Magnus Manske (talk) 22:12, 5 July 2014 (UTC)

@Magnus Manske: - Numbers went down, see new bullet point below the table. Any change in autolist2 that could have caused this? Tamawashi (talk) 10:12, 6 July 2014 (UTC)

@Tamawashi: according to [8] the data set of 25th June was restored for wikidataquery, autolist and autolist2. --Pasleim (talk) 13:37, 6 July 2014 (UTC)

@Magnus Manske:, @Pasleim: this can lead to creation of false claims via Widar. An item may have changed in WD, but Autolist still selects it. Tamawashi (talk) 13:44, 6 July 2014 (UTC)

WDQ seems to be back. The numbers of items that are an instance of something that's not an entity went down from 1,827,524 to 93,118 44,773!

It is important to understand what happened.. The *replication* did not work for some time and this results in "issues" where WD and the results from a query using Autolist or Autolist2 differ. Given that replication was restarted and is again up to more or less 15 minutes of difference, it explains how these numbers can be so much out of kilter. Thanks, GerardM (talk) 20:33, 6 July 2014 (UTC)

@Emw: Is there a wdq for instance of something that has no subclass statement? That would be those items that even after rooting all subclasses would not be instances of entity. Tamawashi (talk) 12:04, 7 July 2014 (UTC)

New page: Wikidata:Subclasses and instances. Tamawashi (talk) 06:31, 9 July 2014 (UTC)

The descriptions and the Wikipedia sitelinks should be changed for this item

The Wikipedia sitelink refers to an "entity" as an individual, rather than the ontological class, which is what this item is clearly defined as in Wikidata.

The class should be different from the entities that the class contains. There is no Wikipedia page for Entity_class, and for (en) it redirects to Entity–relationship_model:

Perhaps, a new Wikipedia page should be created for Wikidata_Entity? This is the top class, so it should have a real and accurate ontological definition. I think that E1_CRM Entity in CIDOC is comparable.

Furthermore, the descriptions that refer to "something" need to be changed to say "class of somethings". I changed the English description based on this reasoning.

I have also reasoned that since http://www.wikidata.org/entity/Q35120 is also an rdfs:Resource and an rdfs:Class, "Entity" is an subclass of owl:Thing (it cannot be equivalent to owl:Thing).

It is actually equivalent to an owl:Class.

Chjohnson39 (talk) 15:24, 1 May 2016 (UTC)

@Chjohnson39: This seems totally wrong. The class should be different from the entities that the class contains : I can't assign a meaning to that sentence. This is just meaningless. The definition of the enwiki article is An entity is something that exists as itself, as a subject or as an object, actually or potentially, concretely or abstractly, physically or not. which seems to be exactly what we're talking about. author  TomT0m / talk page 16:32, 1 May 2016 (UTC)

@TomT0m: What I am trying to explain is entirely clear to me, but I admit that this is a very confusing subject for many people. An "entity" is an resource, a class is a group of resources. The "entity class", defined by this item, is a class group (certainly not an individual "entity"), it is at the top of the Wikidata class hierarchy. This absolutely is a different concept than an "entity" as a "member individual" or "subclass" resource of the "entity class". All other Wikidata items are therefore members (entities) of this class

What you're trying to say is that everything in wikidata is an instance of this class. Including itself of course, overwise there would be something in wikidata that is not an instance of this class. Any class or metaclass is a subclass of this. This is a "variable order class" (see w:metaclass (Semantic Web)). A class is an abstract concept, therefore an abstract entity, therefore it fulfills the english definition. author  TomT0m / talk page 17:28, 1 May 2016 (UTC)

I agree that the entity class can also be an instance of itself. But conceptually, the identity reference of the class and the identity reference of the instance are not equivalent. They are different concepts and from Java, when a class instantiates itself it is called a clone. A clone (or copy) is not the same (because it has a different reference) as the class from which it is derived, even though it may have exactly the same properties i.e. a deep copy. Really there should be another item for the entity class that distinguishes it from the instance of the entity class. This item serves both identities, and this is incorrect.Chjohnson39 (talk) 19:35, 1 May 2016 (UTC)

@Chjohnson39: I really don't get your point. First: what's the problem you are trying to solve and why should it be solved ? Once again this is definitely not a programming language here. Second : what's wrong with the Cyc ontology which does something kind of similar ?

My point is that I would like to be able to reason decidable paths through data with inference queries. Something that is nearly impossible with the undefined (and primarily wrong) structure that is represented currently. The principle purpose of ontology and attempting to make these distinctions between individuals and classes is for queries. I am a programmer and actually am trying to use the data for my project, this is my point. As far as a conceptual reference model, I would recommend CIDOC, which is an ISO standard, and has formed the basis for many domain ontologies, like the one that Wikidata is trying to represent. Cyc is not a conceptual reference model, it is a domain ontology.Chjohnson39 (talk) 11:27, 5 May 2016 (UTC)

But Cyc totally makes a difference beetween classes and instances. See http://sw.opencyc.org/2012/05/10/concept/en/Individual and http://sw.opencyc.org/2012/05/10/concept/en/Collection . I still can't make any sense in your comments. Not a surprise actually ... author  TomT0m / talk page 17:39, 5 May 2016 (UTC)

Entity instance of entity

@Nurni: I assumed entity was the root of our class tree until now : anything is an entity, including itself, including abstract objects. Did you merge because you think that this item is not this object ? The class of all concrete object or events is individual entity (Q23958946)   . Any objection that I revert your revert ? author  TomT0m / talk page 17:10, 28 June 2016 (UTC)

Entity is the root of the class tree. As such, it should be the only item in the tree that is not a subclass of anything. It is the highest-level class. Like all other classes, it is an instance of a class (or it is an instance of a subclass of class). Class itself is a subclass of entity; classes are entities like everything else. --Yair rand (talk) 17:37, 28 June 2016 (UTC)

@Yair rand: I don't understand, the removed statement was not a subclass statement but an instance of one. author  TomT0m / talk page 17:48, 28 June 2016 (UTC)

@TomT0m: Everything is an instance of (a subclass of) entity, including entity itself. Specifically, entity is an instance of a class (or more specifically, a "variable-order metaclass", I guess? I don't know what that is, but presumably it's supposed to be a subclass of metaclass and thus class), which is a subclass of entity. (This doesn't involve any loops, even though it kind of sounds like it does.) --Yair rand (talk) 18:12, 28 June 2016 (UTC)

@Yair rand: It's easier to define "fixed order metaclass" : a class whose all instances are individuals (a type in the type–token distinction (Q175928)    sense) is a 0-order metaclass. A metaclass whom all instances are 0-order metaclass is a first order metaclass, a second order metaclass instances are all first order metaclass and so on. "fixed order metaclass" is the class of all "n-th order metaclass". Variable order metaclass is the class of all classes who are not fixer order. author  TomT0m / talk page 18:24, 28 June 2016 (UTC)

Cyclic classification

@Yair rand: I don't need to read the talk page to know that cyclic classification is one of the bad things to avoid in any ontologies or classification schemes. So please explain this:

• entity is instance of variable-order metaclass
• variable-order metaclass is instance of variable-order metaclass AND variable-order metaclass is subclass of metaclass
• metaclass is instance of concept
• concept is subclass of abstract object
• abtract object is subclass of object
• object is subclass of entity

So instead of reverting, please clean your stuff. Snipre (talk) 11:47, 12 July 2016 (UTC)

@Snipre: I'm not very knowledgeable about data ontologies, but I don't really see how it could be any other way. The top level class, which everything else is an instance of > subclass of* or just subclass of*, must still be a class. "Class" itself is also an instance of a (subclass of*) class, it would be inaccurate to say otherwise. If class it's to be an instance of anything, that must be a subclass of* the root item or else it's no longer a root item. There are no subclass cycles involved here. If "instance of" loops are prohibited, I don't see how "class" or its subclasses can work. --Yair rand (talk) 14:17, 12 July 2016 (UTC)

@Yair rand: You are completely lost in your classification scheme because you mix classification rules and real classification. Explanation: the rule is a subclass of a class is a class. That's a classification rule. So when I say organic compound subclass of chemical compound (real classification), I didn't add the fact organic compound instance of class (inferred from classification rules) even if this is true according to the classification rule.

This is the same problem here: entity is according to your classification rule an instance of something, but in the real classification, as it is the root, entity can't be an instance or a subclass of anything else.

If you want to let the instance of for entity, you have to be coherent and add in all items defined as subclass of something the statement instance of class. Snipre (talk) 15:55, 12 July 2016 (UTC)

No you don't. but in the real classification wtf is the real classification ??? author  TomT0m / talk page 17:08, 12 July 2016 (UTC)

@TomT0m: The real classification is what everyone can do without the need of concepts like class, metaclass, variable order metaclass,... That the result of some absolute analysis. Example organic compound subclass of chemical compound. You can replace subclass by category of, group of, part of or whatever you want, the name of the relation doesn't matter as the relation is independent of it. The class or metaclass concepts are just abstract things we use to define some rules and relations but they are subjective: in another ontology they can used other terms with different definitions. Examples ? Metaclass concept doesn't exist in some ontologies.

Introducing concepts like metaclass or variable order class creates a matrix classification, with vertical relations and horizontal relations (real classification vs. classification organization):

organic compound subclass of chemical compound, vertical relation

organic compound instance of class, horizontal relation

ethanol instance of organic compound , vertical relation

ethanol instance of instance, horizontal relation

Same with entity where no vertical relation can exists as it is the root/top of the classification, but we can still create some horizontal relation like the one here entity instance of metaclass. But using the statements system to describe this relation is wrong because it disturbs the classification tree by creating cyclic relations.

But the terms or the definition are not so important: just try to fix the cyclic relation whatever the system, the classification theory you want to use, because cyclic relations can't exist in an ontology. Snipre (talk) 13:17, 13 July 2016 (UTC)

This is nonsensical. Cyc is an ontology and has this exact statements. That proves your entire comment unfounded. An original work. What's your purpose ? author  TomT0m / talk page 13:20, 13 July 2016 (UTC)

@Snipre: "subclass by category of, group of, part of or whatever you want" = subset (Q177646) (hierarchy (Q188619),directed graph (Q1137726), transitive relation (Q64861)): organic compound (Q174211) (subset (Q177646)) → chemical compound (Q11173) (set (Q36161)), ethanol (Q153) (subset (Q177646)) → organic compound (Q174211) (set (Q36161)). --Fractaler (talk) 09:40, 17 October 2016 (UTC)

@snipre: it was again cyclic https://www.wikidata.org/w/index.php?title=Q35120&diff=492505030&oldid=489242254 77.180.69.66 16:15, 30 May 2017 (UTC)

subsets

molecular entity (Q2393187), former entity (Q15893266), temporal entity (Q26907166), fictional entity (Q14897293), type of business entity (Q1269299) --Fractaler (talk) 18:08, 16 October 2016 (UTC)

synonym (Q42106)

+object (Q488383)--Fractaler (talk) 08:38, 23 May 2017 (UTC)

Unit for the quantity number of entities

The ISQ defines a fundamental (not in the sense of base quantity, but rather "important") quantity called "number of entities": As the name suggests, it counts the number of entities of a specified type. The statement entity (Q35120)measured physical quantity (P111)number of entities (Q614112) seems to fit quite naturally. But then this item needs to be in the class unit of measurement (Q47574) (or one of its subclasses). Would such a use conflict with the "top level" nature of this item? Toni 001 (talk) 15:55, 30 March 2021 (UTC)

Part of psychology terminology

Currently entity (Q35120) is part of (P361) psychology terminology (Q77468620). This seems not to be correct. While the word "entity" may appear as part of psychology terminology, entity-as-a-concept seems to me not to be a part of any terminology. — Finn Årup Nielsen (fnielsen) (talk) 22:10, 28 October 2021 (UTC)

I have now removed it. I am wondering if other editors agree with me here? — Finn Årup Nielsen (fnielsen) (talk) 22:29, 28 October 2021 (UTC)

Existence is not a quality of entity

Otherwise, the ontology would be inconsistent. Since, there are imaginary entities and these do not exist or at least the do not have reality. One could actually remove the statement but it is probably better to keep it for documentation purposes and deprecate it. On the other hand, one could have that kind of documentation also on the talk page. I don't know which is really better, so I will go with deprecation since it is less intrusive.

In any case, as long as one posits the entity of "non-existent entity", existence cannot be a quality of entity. --Dan Polansky (talk) 11:32, 29 December 2022 (UTC)

Sources not requiring existence as quality/property of entity

• https://www.macmillandictionary.com/dictionary/american/entity
• WordNet

--Dan Polansky (talk) 11:41, 29 December 2022 (UTC)

Definition

The following definition seems correct to me and making almost no ontological commitments; perhaps some, I don't know:

• the root of ontology

Of course, one should not think that it is therefore true that entity is-instance root. The above is a description on the meta-level. But that is not a definition for mere mortals. Hence, here is the 2nd approximation:

• thing

But that is ambiguous since things are often distinguished from persons. Let's try better:

• thing, animate or inanimate, real or unreal, abstract or concrete

That seems to be the best current approximation. Why do we use the or-phrasing? It is to make clear the range of things involved under the concept, including the world, humans, persons, non-human animals, trees, numbers, Bilbo Baggins, etc. But we can do even better:

• thing, living or non-living, real or unreal, abstract or concrete

Since, animate is ambiguous to say the least and points via etymology to breath, which does not seem to help anything: not all living things are breathing. The current definition is this:

• anything that can be considered, discussed, or observed

That is probably fine. What is important is "can". --Dan Polansky (talk) 13:22, 29 December 2022 (UTC)

Later: One requirement one can make on entities is that they can in principle exist even if there is no-one to consider, discuss or observe them. Thus, stars existed long before humans did; and stars were entities. One can then start nitpicking on the use of the word "can" in the last definition, and wonder whether stars really could, had the capacity, to be considered, discussed or observed, given there was no-one to do these things. One may respond by saying that "can" means something like "can in principle" or "can provided we stipulate a mind, whether real or imaginary" or something of the sort. That could work. --Dan Polansky (talk) 13:52, 26 January 2023 (UTC)

entity subclass-of class

Untrue since some entities are not classes, namely instances, and probably many concepts. However, if entity is not a class, how can entity act as object in subclass-of statements? It probably can, since the 2nd argument of subclass-of possibly does not need to be a class, only the first one. Needs deeper clarification. --Dan Polansky (talk) 18:21, 29 December 2022 (UTC)

No opposite of entity

"Second, any fully general category lacks a contrast or complement", per SE:Object[9]. --Dan Polansky (talk) 13:56, 30 December 2022 (UTC)

What is entity in relation to arbitrary slices

• Defining axiom: Any referent of a definite description is an entity.
• Corollary: The referent of "the merely mentally delineated arbitrary slice or subset of the apple before me" is an entity.

Therefore, things that we possibly would not think of as entities are ranked as entities, as long as we are able to construct definite descriptions referring to them. This gives us motivation for distinguishing entity from unit: of unit we may require more than just being a referent of a definite description. Thus, in military, a particular organizationally delineated group of humans is a "unit", whereas an arbitrary subgroup of that group is an "entity" but not "unit". Tracing to authoritative sources missing. Dan Polansky (talk) 13:49, 25 January 2023 (UTC)

"the merely mentally delineated arbitrary slice or subset of the apple before me" is quite an entity for me. --Infovarius (talk) 12:34, 27 January 2023 (UTC)

We can consider wilder entities: "the merely mentally delineated discontiguous arbitrary object resulting from joing an arbitrary slice of a human's heart and an arbitrary slice of a human's liver". We can in fact consider any set of atoms or elementary particles or quarks and take it to correspond to an entity. There is something bizarre about considering these kinds of referents to be entities, but as long as all referents are entities, we have to do that.

We can probably mentally join a positive and a negative entity into a compound entity; thus, "the compound entity that results from joining this apple and the hole in that cup's handle". There seems to be no limit to this craziness.

This deliberation offers the possibility that object is distinct from entity/any referent, depending on the definition of object. --Dan Polansky (talk) 07:55, 1 February 2023 (UTC)

bad disjoint union

There are many items that are instances of more than one of former entity, current entity, non-existent entity, hypothetical entity, and missing entity so I'm deprecating this disjoint union. Peter F. Patel-Schneider (talk) 20:23, 13 December 2023 (UTC)

May be to change to simple union? --Infovarius (talk) 20:55, 17 December 2023 (UTC)

Link broken

The link Wikidata talk:WikiProject Ontology#Should we have an "entity of unknown existence" as top node of the subclass tree? is broken.

The discussion is now found at:

Wikidata talk:WikiProject Ontology/Archive for 2022#Should we have an "entity of unknown existence" as top node of the subclass tree?

Can someone with the proper edit rights update this? TimBorgNetzWerk (talk) 10:55, 30 January 2024 (UTC)