Talk:Q3965271

subset (Q177646)

Merge? --Fractaler (talk) 07:43, 20 October 2016 (UTC)

Nope. See Von Neumann–Bernays–Gödel set theory (Q278770)    for example. author  TomT0m / talk page 17:25, 20 October 2016 (UTC)

"In set theory and its applications throughout mathematics, a subclass is a class contained in some other class in the same way that a subset is a set contained in some other set" --Fractaler (talk) 10:00, 21 October 2016 (UTC)

Yep, but as not every class is a set, this means that not every subclass is a subset. A set is always a subclass of some class - at least itself - but a proper class is nether a subset of another proper class. The subset relationship is the subclass relationship specialized for sets - that said, most of the classes we have in Wikidata are essentially sets author  TomT0m / talk page 10:19, 21 October 2016 (UTC)

For example? What we can not express, using a set? --Fractaler (talk) 12:26, 21 October 2016 (UTC)

set (Q36161)    per Russell's paradox (Q33401)    and definition of the class of all sets. author  TomT0m / talk page 12:47, 21 October 2016 (UTC)

Paradoxes are due to polysemy (Q191928). All sets (Russell's paradox (Q33401)) - homonym (Q160843). Parts: "all sets at the time t1", "all sets at the time t2", ..., etc.--Fractaler (talk) 15:07, 21 October 2016 (UTC)

This does not make any sense. author  TomT0m / talk page 15:16, 21 October 2016 (UTC)

polysemy (Q191928) ("All sets") does not make any sense. We need +time (arrow of time (Q186559)) --Fractaler (talk) 15:35, 21 October 2016 (UTC)

Mmm I think I'm beginning to understand, are you trolling about putting date on instance of (P31) statement ? Time is not really relevant for maths. Maths sets are not a dynamical system. Classes in semweb are name that way, for example in OWL, because a class can be defined by intension by a predicate in OWL language - I don't know if OWL class expression (predicates) only defines maths sets in some flavors - probably, that's probably useful for decidability. That said OWL does not have a real time notion as far as I understand coded in its logic rules. Wikidata has no logic at all atm.

The example of hurricane can be interesting - hurricanes could be classified differently in the course of their existence. They can belong to a high-speed hurricane in the beginning then be classified in low-speed one. You and me will be classified in the class of dead people sometime, we don't at that point. So, for dynamic stuffs, (assuming the philosophical position that I'm the same person as yesterday, some people argues that you're actually someone else over time), they can belong to some class at some point in time because they meets the criteria that qualifies to belong to that class, then fail to meet this later in time. Say someone win the lottery he will qualify for begin classified in the upper class, then he gives all its money to charity, he will no more - but he may qualifies the criteria to be classified as ruined person.

Talking about death, we'll be dead before our corpse will be totally decomposed ... we rarely qualifies a body as a human though.

All of this to say that temporal entity evolves in time and that it's certainly useful to be able to classify at some point in time - that's where temporal qualifiers are useful. And they can be used for classes as well imho if we choose the good meaning : between 2013 and 2014 he meat all the criteria to be rich could be modelled by

⟨ the lottery winner ⟩ instance of (P31) ⟨ rich people ⟩
start time (P580) ⟨ 2013 ⟩
end time (P582) ⟨ 2014 ⟩

. I know people thinks we can define an essence of something that does not change over time - a human stays a human, but it's philosophically difficult to argue its corpse is him imho, or that he's his immortal soul that would require taking a religious position which is far from being neutral.

So ... I think we can distinguish two kind of object : temporal one and platonical ones - those who exist in a non-material world. For temporal one, they can evolve, and their properties over time changes, something wikidata can totally handle. I think this definitely also works for class membership with no real problems. For atemporal classes like the mathematical ones it's not meaningful to put times constraint. They are discovered and/or described at some point in time, like biological taxons however, but that's an extrinsic property, a species - more precisely its instances did start to exist before it was described by humans.

Another interesting case is the case of classes whose criterias for membership are different over time - for example the definition of a poor depends on the income of people of a country at some point in time.

Another interesting limit temporal case is the one of projects, stuffs that are planned but do not exists yet and may not exist at all in the future.

Also relevant for the arrow of time are has cause (P828) and https://www.wikidata.org/wiki/Property:P1478 Search for example, but it's largely orthogonal to classification issues.

As a bird eye view, without over-thinking it, I think we have mostly what we need with when relevant timed-qualified instance of (P31) statements. Non temporal classes won't have such qualifiers. Over-thinking it : I think key key question is to deal with the question of whether or not time properties can be part of the definitions of a class - I think the answer is sometime "yes". For example the middle ages battle class or category where the classification has time criteria. Sometimes it is not, say the class of all people who ever lived is defined atemporally. But its extension grow all the time - say it's a dynamic class, or a time opened class ? by opposition the middle ages battle would be "time closed" as its extension can not grow now-days. A third type like the poor people class is probably more complex as the criterias to belong or not to the class depends on properties of the contemporary world - they are time dependant themselves ...

The question on whether or not these time dependent criteria can be modelled using a set-theoretic class framework where time is just another predicate is opened for me :) author  TomT0m / talk page 10:26, 22 October 2016 (UTC)

1) are you trolling about putting date on instance of (P31) statement ? - I use classification structure (Q6671777) subset→set→superset. Time is not really relevant for maths. Maths sets are not a dynamical system. Classes in semweb are name that way - if in mathematics (Q395) (and only), then: class (Q217594) (collection of sets in mathematics ...) ⊂ collection (a set of items or amount of material procured or gathered together) ⊂ set (Q36161). Wikidata has no logic at all atm - Wikipedia (tree of categories), Wikidata use the classification structure (Q6671777) subset→set→superset. "Set at the time t1" - subset, "set at the time t2" - set, "set at the time t3" - superset. No dynamical system, only sets. --Fractaler (talk) 08:14, 24 October 2016 (UTC)

2) I think we can distinguish two kind of object : temporal one and platonical ones - if you mean worldview (Q19618723)/worldview (Q2030462) (picture/model of the world, lat. forma ideaque mundi), then 2 part: I. dynamic, kinetic, kinetic energy (Q46276) →☯→ II. static, potential, potential energy (Q155640) (by yin and yang (Q62744), Georg Wilhelm Friedrich Hegel (Q9235), Q4184872 and other). For example: static, level №1 - languages (ab.wikipedia.org+ady.wikipedia.org+zu.wikipedia.org) → dynamic (conversion) ☯ → static of next level (level №2) - number (wikidata.org) --Fractaler (talk) 09:14, 24 October 2016 (UTC)

3) medieval battle (Q15991159) (subset) ⊂ "middle ages military operation" (set) --Fractaler (talk) 12:03, 24 October 2016 (UTC)