Wikidata:Property proposal/ontological level of Wikidata item

From Wikidata
Jump to navigation Jump to search

ontological level of Wikidata item[edit]

Originally proposed at Wikidata:Property proposal/Generic

   Not done
DescriptionA meta property that describes the level of the Wikidata item. Its objective is to make explicit which items refer to an individual (for example, Immanuel Kant, which is a singular entity), which ones represent first-order types (for example, philosopher), which ones represent second-order types (i. e. profession) and so on.
Data typeItem
Domainitem
Example 1Immanuel Kant (Q9312)individual (multi level modelling) (Q83978504)
Example 2philosopher (Q4964182)first-order type (multi level modelling) (Q83978609)
Example 3Laika (Q53662)individual (multi level modelling) (Q83978504)
Example 4dog (Q144)first-order type (multi level modelling) (Q83978609)
Planned useAdd manually the property to items within my domain of expertise (biology). Perhaps auto-add classification for some entities.
Expected completenesseventually complete

Motivation[edit]

This property is motivated by readings of multilevel theory and the assessment of Guizzardi et al of ontological anti-patterns in Wikidata items (referenced below).

By making it explicit the ontological level of the Wikidata item, it might be easy to identify anti-patterns, improve interpretation of the differences between "subclass of" and "instance of". For non-ontologists, it might sound intuitive that dog (Q144) would be a physical entity, even though there are actually only instances of dog (Q144) in the real world.


  • Giancarlo Guizzardi; Freddy Brasileiro; João Paulo A. Almeida; Victorio A. Carvalho (April 2016), "Applying a Multi-Level Modeling Theory to Assess Taxonomic Hierarchies in Wikidata", Proceedings of the 25th International Conference Companion on World Wide Web: 975–980, doi:10.1145/2872518.2891117 Wikidata Q27037396 .

TiagoLubiana (talk) 20:19, 29 January 2020 (UTC)

Discussion[edit]

I would argue that the meanings of individual (Q23958946) and {Q|Q83978504} are different in that the first refers to the physical concept described by the item and the second to a property of the item itself. So Douglas Adams (Q42) would be an instance of individual (Q23958946) and a Q42 (Q42395533) would be an instance of individual (multi level modelling) (Q83978504) The idea of a class (Q5127848) is different of first-order type (multi level modelling) (Q83978609) too, as first-order type (multi level modelling) (Q83978609) is more precise in its application to ontologies, not applying to general to philosophical considerations. I see that things could be classified by the class tree, though. But individuals and classes seldom are explicit, and this is confusing for beginners. Wikidata is specialized, but it is still public, so maybe this could help people to get used to it. But I agree, there are probably better ways to model this, and this was simply what I was able to come up with. Eager to read your proposal. TiagoLubiana (talk) 14:57, 4 February 2020 (UTC)
  • Symbol oppose vote.svg Oppose I agree with Yair rand. This solution has the problem that it increases the effort to create a new items by requiring an additional statement to be filled for every item. Every instance of (P31) of dog (Q144) should have the same ontological level and a solution in which different instances of the same class have the possibility to be in different ontological levels seems to me problematic.
The example where Laika (Q53662) and dog (Q144) have the same ontological level seems wrong to me. ChristianKl❫ 09:43, 4 February 2020 (UTC)
  • Yeah, I already don't get the samples ;) --- Jura 09:47, 4 February 2020 (UTC)
Oh my, the examples are not the same level, you are correct, sorry. It was kind of a copo (a typo of copying and pasting). I will correct it. TiagoLubiana (talk) 14:57, 4 February 2020 (UTC)
Additionally, I agree that it seems redundant to every instance of (P31) of dog (Q144) to have the same level, but maybe redundancy makes it clearer. In the source article referenced above, they saw that there are many cases in which things have relations that break the logical model of subclasses in the same level, and instances of in hierarchically related levels. Maybe this could be enforced with a property similar to this one. For example, all instances of dog (Q144) (which is a -->first-order type (multi level modelling) (Q83978609)) would have to be --> individual (multi level modelling) (Q83978504). This could even be done automatically to identify conflicts. TiagoLubiana (talk) 14:57, 4 February 2020 (UTC)
  • I do agree that abominations like anonymous (Q4233718) instance of (P31) human (Q5) that break ontological assumptions shouldn't exist. While there are benefits to making thinks clearer the cost of this proposal are two high. I think the approach of defining necesseary properties for instances of a class would also achieve the goal of improving clarity. We might also need other policy. ChristianKl❫ 21:43, 4 February 2020 (UTC)
Agreed. I would remove the proposal, but I believe that some people might throw in useful opinions about the subject. Is it okay for you to leave it here? TiagoLubiana (talk) 18:51, 6 February 2020 (UTC)

Pictogram voting comment.svg Comment Would you argue that the concept of a dog is a class, in the ontological sense, or that ontologies model the concept of dog as a class? Rephrasing: are classes real things, that have to be agreed upon logically by anyone that follows formal logic, or are they restricted to Wikidata (ex: Wikidata deals with profession (Q28640) as a fixed order metaclass first-order metaclass (Q24017414) but other ontologies could model the same idea in a different, but equally valid way).  – The preceding unsigned comment was added by TiagoLubiana (talk • contribs).

  •  Not done no support for creation.--- Jura 09:41, 22 February 2020 (UTC)