Wikidata:Property proposal/est/a été en conflit avec

is/has been in war with edit

Originally proposed at Wikidata:Property proposal/Generic

Not done

Description	countries which have been/are in plain war with that country
Represents	state (Q7275)
Data type	Item
Domain	state (Q7275)
Allowed values	state (Q7275) , political organization (Q7210356)
Example 1	Ethiopia (Q115) → Eritrea (Q986) with statement is subject of (P805)→Eritrean–Ethiopian War (Q832273) and start time (P580)→6 mai 1998 and end time (P582)→25 mai 2000
Example 2	Nazi Germany (Q7318) → French Third Republic (Q70802) with statement is subject of (P805)→Armistice of 22 June 1940 (Q170360) and start time (P580)→3 septembre 1939 and end time (P582)→22 juin 1940
Example 3	French Third Republic (Q70802) → Qing Dynasty (Q8733) with statement is subject of (P805)→Sino-French War (Q706183) and start time (P580)→1884 and end time (P582)→1885
Example 4	French First Republic (Q58296) → Kingdom of Bohemia (Q42585) with statement is subject of (P805)→French declaration of war against Austria (1792) (Q3042700) and start time (P580)→ 20 avril 1792 and end time (P582)→25 mai 1802
Expected completeness	always incomplete (Q21873886)
See also	conflict (P607)

Motivation edit

To make possible to directly query which country is experiencing existing or old wars. Bouzinac (talk) 16:09, 6 August 2020 (UTC)[reply]

Discussion edit

Comment I like the idea of being able to query for this. But wouldn't it be better to improve the modelling of the war items instead? Country items are already notoriously cluttered and this would add dozens of statements to those. Belteshassar (talk) 20:18, 7 August 2020 (UTC)[reply]

Well, the current existing countries / regimes are much big, true, but they have had few wars, or be them countable on fingers of a hand :) French Fifth Republic (Q200686) has only had fought (to my knowledge) two wars : Algerian War (Q200790) and Gulf War (Q37643). Bouzinac (talk) 06:02, 8 August 2020 (UTC)[reply]

I was thinking of Sweden (Q34) and Denmark (Q35), but I now realize that they are very much outliers Belteshassar (talk) 15:44, 8 August 2020 (UTC)[reply]

Strong oppose Belteshassar is right that this would add many statements to country items. Looking at the list of wars the USA has fought, there are dozens of wars and many of the wars have multiple opponents. Additionally, since many wars involve more than two countries, using the proposed property would lead to an extreme amount of redundancy as each combatant would need to be duplicated for each country on the opposing side. If four countries are allied against three countries, then this property would require 24 statements (4x3+4x3), whereas putting this information all on one page should only need seven statements. — The Erinaceous One 🦔 04:37, 23 September 2020 (UTC)[reply]

~~Oppose~~ Wait I agree with Belteshassar, lets rather improve the war items. I dont think the motivation I have read is sufficient for creating a new property, it is already possible to query the information using the inverse property Eritrean–Ethiopian War (Q832273)participant (P710)Ethiopia (Q115) for the example item Eritrean–Ethiopian War (Q832273) which should allow the creation of a SPARQL query:

SELECT ?item ?itemLabel 
WHERE 
{
  ?item wdt:P31/wdt:P279* wd:Q198.
  ?item wdt:P710 wd:Q986.
  SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
}

Try it!

Also I think each war should be able to get its own item so I think listing them on the country page does not really help, it actually makes it more difficult to maintain since somebody may forget to add it to the country page. Overall we are talking about 3799 entries (as of yet) that need to be kept up to date with the entries of instance war. However, in this case where the data is already modelled and present in the database, adding inverse properties is really more convenience and does not help much in terms of modelling, but it leads to a lot of maintenance effort without really providing anything of use. And yes, some countries have a lot of wars, eg. with Ottoman Empire (Q12560) 70 followed by United States of America (Q30) with 61 -- you need a lot of fingers on your hand for that (see https://w.wiki/Zkg ). --Hannes Röst (talk) 16:29, 17 August 2020 (UTC)[reply]

Hi Hannes Röst (talk • contribs • logs), well, https://en.wikipedia.org/wiki/List_of_wars_involving_the_United_States has it that there would have been 119 conflicts involving the USA. So OK, my argument, the one stating there would be few statements, was surely a bit thoughtless. But I still think that property would be useful because with your sparql example, you cannot clearly state if Eritrea (Q986) is ongoing a war or not, without a very complicated sparql query. Shouldn't wikidata be simple to query, even if it takes a bit of maintenance. There is not so much war that have been fought between states. That Quora] replies there would have been a total of 10,000 wars in the whole history. It's quite a big number but Wikidata can handle that, I'm sure. There are many list of wars in wikidata that would help populate that property. Bouzinac (talk) 19:20, 17 August 2020 (UTC)[reply]

I think its actually very easy and I am happy to help you look for that since you can look for a property

SELECT ?item ?itemLabel 
WHERE 
{
  ?item wdt:P31/wdt:P279* wd:Q198.
  ?item wdt:P710 wd:Q30.
  FILTER NOT EXISTS { ?item wdt:P582 ?endtime}
  SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
}

Try it!

while in your proposal you would have to look for a qualifier which is actually harder to do in SPARQL. The query above shows 4 ongoing wars for the USA (one is a false positive), but also highlights the dangers of duplicating data as you suggest: it will be very easy for a war using your new property to not have an end time (P582) but have it in the item or the other way around -- data consistency suffers from duplication. I agree that Wikidata can handle to enter the data twice, but the danger is that it takes humans to maintain these entries. Secondly, I think wars are so important that they should be their own item and not properties. --Hannes Röst (talk) 00:56, 18 August 2020 (UTC)[reply]

@Hannes Röst: One thing I struggle with currently is how to model if countries were adversaries or allies in a war. Take Six-Day War (Q49077) as an example, How do I express that Israel (Q801) was at war with Egypt (Q79), but Egypt (Q79) was not at war with Jordan (Q810)? Belteshassar (talk) 06:02, 18 August 2020 (UTC)[reply]

@Belteshassar: very good point, this is currently lacking. I see two solutions here: if its a "normal" war with 2 sides then we could simply use something more descriptive than participant (P710) with a list of countries, either there is a new property for this or we use qualifiers. In this case we could for example group the participants as such: Six-Day War (Q49077)participant (P710)Israel (Q801)series ordinal (P1545)"1", Six-Day War (Q49077)participant (P710)Egypt (Q79)series ordinal (P1545)"2", Six-Day War (Q49077)participant (P710)Jordan (Q810)series ordinal (P1545)"2" and maybe we can come up with a better property than series ordinal (P1545) (maybe "allied faction"). That seems to work for wars with N faction where each faction is behaving the same way. Maybe there are more complex cases where country A in faction 1 is at war with country B in faction 2 but not country C in faction 2? In that case you would need proper modelling of bilateral behaviour and then you probably need a property as described above. For example, there were 3 days when the US was at war with Japan but not Germany and Italy (see https://en.wikipedia.org/wiki/Italian_declaration_of_war_on_the_United_States and https://en.wikipedia.org/wiki/United_States_declaration_of_war_on_Japan) so to properly model that you would need bilateral modelling, I agree. This is a good point and if the intent is to model on this level of detail I am happy to remove my objection. --Hannes Röst (talk) 14:15, 18 August 2020 (UTC)[reply]

Hi all, happy to have started an interesting topic. What about modelling Kingdom of Italy (Q172579)+Finland (Q33) in the context of World War II (Q362)? It was first against Allies of the Second World War (Q329888) and then fought with Allies of the Second World War (Q329888). Simply having participant (P710) might be not enough specific to help describe which country was at war with which country.

Another example is this list https://en.wikipedia.org/wiki/List_of_wars_extended_by_diplomatic_irregularity : there is countries that are "technically" currently ongoing a war : they did not formalize a peace truce/treaty. Bouzinac (talk) 20:34, 18 August 2020 (UTC)[reply]

I think that the property needs some modification to really model what we want, but we are moving in the right direction. I think with my proposal we could still model Kingdom of Italy (Q172579)+Finland (Q33) for example: World War II (Q362)participant (P710)Finland (Q33)series ordinal (P1545)"1"start time (P580)"start" using start time (P580) and end time (P582) for the time it was in faction 1 and then World War II (Q362)participant (P710)Finland (Q33)series ordinal (P1545)"2"start time (P580)"start" using start time (P580) and end time (P582) for the time it was in faction 2. So it would be possible. However, I had another idea: we can use existing items of the type Egypt–Jordan relations (Q21686822) and Egypt–Israel relations (Q574821) to model wars. It seems like the actual correct place for a m:n relationship to me and being at war is a type of diplomatic relation between two countries. I have tried this here: https://www.wikidata.org/wiki/Q574821#P793 and here https://www.wikidata.org/wiki/Q21686822#P793

Secondly, I realized that there are actually 2 distinct concepts: (i) the diplomatic state between two countries (when they did declare war against each other and when they signed a peace treaty / armistice), (ii) participation in actual campaigns where they fought against other countries in a coalition with allies and adversaries. those may not overlap as Bouzinac has shown. In the case of Egypt-Israel it seems that the state of war existed from 1948 to 1979 while during that time they fought multiple wars. So the state of war/peace should be distinct from any particular war they fought in (and we should model both). So I think we could model this as Egypt–Israel relations (Q574821)significant event (P793)State of war (Q942343) but it would be better to have a new property instead of significant event (P793) whould would apply to instances of bilateral relation (Q15221623)} and would be called "state of bilateral relationship" with a few allowed instances (armistice, peace, war etc) and qualifiers could be used. On top of that we can model participation in *specific* wars through Egypt–Israel relations (Q574821)significant event (P793)Yom Kippur War (Q49100) and use qualifiers for that. Unfortunately there would be some duplication with whatever is in the war item Yom Kippur War (Q49100) so we need to make sure that this is consistent. Any thoughts? @Bouzinac, Belteshassar: --Hannes Röst (talk) 14:23, 20 August 2020 (UTC)[reply]

Property population edit

I would perhaps use this dataset to help populate that property, if this creation is accepted. https://correlatesofwar.org/data-sets/MIDs (not sure yet about the copyright issue). Bouzinac (talk) 12:39, 19 August 2020 (UTC)[reply]

Oppose We already have enough properties that get used on countries so that the loading time is often inhibited. The information can be well stored in items about the individual wars. ChristianKl ❪✉❫ 15:56, 24 August 2020 (UTC)[reply]

I don't agree with your arguments. The size / load issue should not be used as an argument to deny eligibility for a property (= do you mean you wish to ban new properties usable on countries items?) Bouzinac (talk) 19:31, 24 August 2020 (UTC)[reply]

I don't think that we should create properties to store information that can easily stored outside of country items in country items. ChristianKl ❪✉❫ 20:53, 24 August 2020 (UTC)[reply]

I agree especially since we already have items for all / most bilateral relations. --Hannes Röst (talk) 18:01, 3 September 2020 (UTC)[reply]

As I pointed out above some countries have more than 60/70 wars of which each one has presumeably one or more participant, so we are easily talking about adding hundreds of elements to some items. --Hannes Röst (talk) 18:02, 3 September 2020 (UTC)[reply]

I can understand your oppositions. I just find it a (small) shame because, for a said country X, it's different between querying whom has been in war with X // and reading in the Q of X whom has he been in war with. Query has the drawback of relying on data quality of many war and just looking to this query, supposed to give all current war, gives way much false positives. Whatever. Bouzinac ^{💬●✒️●💛} 19:27, 23 September 2020 (UTC)[reply]

Not done Clear opposition. JesseW (talk) 02:25, 12 March 2021 (UTC)[reply]