Wikidata talk:WikiProject Heads of state and government

From Wikidata
Jump to navigation Jump to search

Launch of WikiProject Wikidata for research[edit]

Hi, this is to let you know that we've launched WikiProject Wikidata for research in order to stimulate a closer interaction between Wikidata and research, both on a technical and a community level. As a first activity, we are drafting a research proposal on the matter (cf. blog post). It would be great if you would see room for interaction! Thanks, --Daniel Mietchen (talk) 01:48, 9 December 2014 (UTC)[reply]

Wikimania 2016[edit]

Only this week left for comments: Wikidata:Wikimania 2016 (Thank you for translating this message). --Tobias1984 (talk) 12:03, 25 November 2015 (UTC)[reply]

P31 (instance of) and P279 (subclass of)[edit]

The example shows First Minister of Scotland (Q1362216) as both subclass of (P279) prime minister (Q14212) and instance of (P31) position (Q4164871)

This can't be right. There is only one position "first minister of Scotland", so it's not a subclass of anything.

Additionally, prime minister (Q14212) is a subclass of position (Q4164871) so it would seem the correct relation should be

First Minister of Scotland (Q1362216)instance of (P31)prime minister (Q14212)

(though some might question whether the First Minister is really a kind of prime minister, given that Scotland is not (yet?) an independent sovereign state).

This should be cleaned up. Jheald (talk) 22:02, 14 August 2017 (UTC)[reply]

@Jheald: leaving aside the particularly Scottish aspect of this for now, this model of having both subclass-of (type of position) / instance-of (position/public office) is used on pretty much every similar position worldwide: e.g. Prime Minister of Belgium (Q213107), Prime Minister of Canada (Q839078), President of Argentina (Q12969145), President of South Africa (Q273884), President of the United States (Q11696), etc., so I think it's fine to document that's how this is currently working so that we can at least work towards consistent modelling everywhere. This certainly causes some problems due to an individual person being connected to the position via position held (P39) rather than instance of (P31) (see for example some discussion on when we should be using has part(s) of the class (P2670) vs has part(s) (P527) at User talk:Jura1/5/10/4/1aruJ:klat_resU), but it's not clear to me how we get consensus on how this all should be modelled, so we can change the documentation and update all the uses without half of them being reverted. --Oravrattas (talk) 09:10, 15 August 2017 (UTC)[reply]
Particular people are instances of human (Q5)
Particular posts are instances of items that are subclasses of position (Q4164871)
Particular posts are *not* people. They are jobs that people have.
The statement
⟨ P39 ⟩ subproperty of (P1647) View with SQID ⟨ P31 ⟩
was an error and should be removed.
The items that have a P31 and a P279 statement that are both in the subclass tree of position (Q4164871) are given by the query tinyurl.com/ydho9qk2
Where a post is in the above list and the current P31 is position (Q4164871) or public office (Q294414), we simply remove the current redundant P31 statement, and instead make the current P279 statement a P31 statement instead.
Using QuickStatements2 the whole lot should take no more that three or four hours.
The above simply reflects how we deal with every other property of humans.
A particular person is not an "instance of" painter -- they have occupation: painter.
A particular person is not an "instance of" <member of club X> -- they have membership: <club X>
Jheald (talk) 18:54, 16 August 2017 (UTC)[reply]
Personally, I prefer the current approach. Obviously, feel free to delete the subproperty statement on P39 and create an item for "first ministers".
--- Jura 18:57, 16 August 2017 (UTC)[reply]
But @Jura1: the current state of things is plainly wrong. It's out of step with every other statement about people on Wikidata; it causes people to confuse or misunderstand the nature of items in the position (Q4164871) subtree; it seeds confusion about the proper use of subclass of (P279); and the redundancy is something we eliminate on sight everywhere, because we know it leads to bad practice and causes confusion. Jheald (talk) 19:06, 16 August 2017 (UTC)[reply]
That's just something you are stating. If there are specific areas where people have problems, I'd be happy to help. I can help creating a "first minister" item if you need that. Besides, I think it's consistent with the painter sample you gave: painter is an instance of a profession and a subclass of artist.
--- Jura 19:32, 16 August 2017 (UTC)[reply]
@Jura1: "Painter" is a subclass of artist because the idea of "painter" can be subdivided into different types of painter.
"First minister of Scotland" is not a subclass of "position", "first minister" or anything else because it is a single identifiable post: it is not sub-dividable further.
Similarly Minister for Environment, Planning and Countryside (Q33108632) is not a subclass of environment minister (Q4574084), because it is a single identifiable post, not a class of posts. The difference is important, and worth keeping clean. Jheald (talk) 19:45, 16 August 2017 (UTC)[reply]
I'm not sure where you are getting your sample from, but Wikidata Q1362216#P31 has "instance of" "position".
--- Jura 19:52, 16 August 2017 (UTC)[reply]
@Jura1: What's I'm criticising is that Wikidata Q1362216#P279 has "subclass of" "prime minister" (and "first minister" would be just as bad).
What should be there is for First Minister of Scotland (Q1362216) to be "instance of" "prime minister" or "first minister" or whatever.
And then "instance of" "position" should be removed as redundant, the same way we do everywhere else. On wikidata you don't redundantly give a higher class if it already follows from a more precise class. Jheald (talk) 20:14, 16 August 2017 (UTC)[reply]
@Yair rand: Any comments on this, as the person who added the position held (P39) subproperty of (P1647) instance of (P31) statement? And also pinging @Infovarius: as my understanding of the position held (P39) subproperty of (P1647) instance of (P31) came from the discussion at Topic:Ts00xiu63gohyygu. It would be good to get consensus on this that we can safely apply everywhere. --Oravrattas (talk) 21:05, 16 August 2017 (UTC)[reply]
The use of P279 in these items is correct. A Minister of Defence of Cyprus is a Minister of Defence, and is thereby also a minister. An Associate Justice of the Supreme Court of the United States is a Justice of the Supreme Court of the US is a Justice of a Supreme Court. A Romanian Senator is a Senator and a Romanian legislator and a legislator. Someone wanting to query all ministers, or all ministers of a certain type, or all ministers of a certain country, should be able to do so using P39/P279*. If all X are Y, X P279 Y. That's what P279 means.
I'm not completely certain about the subproperty of (P1647) statement. Whether someone holding the office of Prime Minister "is" a prime minister is ambiguous. However, it is not ambiguous that if someone holds the office of Prime Minister of the United Kingdom, it follows that they hold the position of Prime Minister. --Yair rand (talk) 02:17, 17 August 2017 (UTC)[reply]
@Yair rand: There is only one position that is Minister of Defence of Cyprus.
Therefore Minister of Defence (Q25907547) is not a subclass of anything.
If there is only one X, then X P31 Y. That's what P31 means. Minister of Defence (Q25907547) instance of (P31) defence minister (Q2518691).
If somebody wants to query all ministers, or all ministers of a certain type, etc, the appropriate relation they should be using is P39/P31?/P279*
In natural language, the use of the term "prime minister" (or "painter") is ambiguous. Does it mean a role, or does it mean a person? On Wikidata, this can of potential difficulties has been recognised, and a decision has been taken to avoid it. On Wikidata, prime minister (Q14212) is a role. A person is not a role, so X instance of (P31) prime minister (Q14212) is not valid. That invalidity means the statement position held (P39) subproperty of (P1647) instance of (P31) is also not valid. Jheald (talk) 10:32, 17 August 2017 (UTC)[reply]

┌────────────────────────────────────────────────────────────────────────────────────────────────────┘

  • It seems that there are different way of seeing this, but are there specific items where the current approach causes you some problems ?
    --- Jura 11:03, 17 August 2017 (UTC)[reply]
I object to all 11,000 of them.
I'm concerned that this error should be corrected as soon as possible, because I think it instills wrong ideas: wrong ideas about how Wikidata deals with people and roles -- people should never be instances of positions, professions, occupations; and wrong notions about redundancy -- items should never additionally be made instances of a top-level class if they are instances of a subclass. Both of those are important principles for new users to become familiar with on Wikidata, and in my view in this respect, ie with regard to new-user understanding of Wikidata data organisation, these 11,000 items are currently not just unhelpful, they are actively harmful.
The sooner this is fixed, the sooner we will stop propagating this bit of bad understanding and bad structure, and the less code will get written assuming it.
I particularly want to see this fixed by Saturday morning, when there is to be a significant workshop on Wikidata in London, co-hosted with a major external partner in the area of political data.
I want any participants at that meeting, that may have scripts already based on P39/P279*, to realise the need to migrate them, and to be facilitated to migrate them to P39/P31?/P279* in the easy supportive environment of the workshop. And for (probably larger number) of participants experiencing political data on Wikidata for the first time, I want them never to be led astray in the first place, but to see and learn from the data as it should be right from the start.
As I say, that meeting starts on Saturday morning. So I would want to see the QS2 run well underway by early Friday night at the latest. Jheald (talk) 14:54, 17 August 2017 (UTC)[reply]
For what it's worth, the Saturday event is looking like it's going to be mostly focused on the UK context and (looking at the suggested topics) mostly on parliamentarians, so we're not likely to do any irretrievable harm if we don't get this sorted out before then.
I'm finding it quite hard to get my head around the distinction here. @Jheald:, I agree that we should never say a person is instance of (P31) prime minister (Q14212), but I don't quite see the semantic step where our use of position held (P39) is equivalent to saying that. If position held (P39) subproperty of (P1647) instance of (P31) is making that assertion, I would agree it should be removed, but I'm a bit lost as to how it links to the subclass of (P279)/instance of (P31) question.
In terms of subclass of (P279)/instance of (P31) on the position items, one possible issue is that we're focusing on using the "specific" items in position held (P39), ones that don't have any subdivisions. However, there will always be items that have to use position held (P39):mayor (Q30185) rather than a more specific version such as Mayor of New York City (Q785304), so we'll always have cases of items with P39 using an "intermediate" item with subclass of (P279) as a target. Will this cause problems for your model? Andrew Gray (talk) 23:13, 17 August 2017 (UTC)[reply]
Hi @Andrew Gray: In answer to your last para, the intermediate thing is taken care of by the ? (which makes it optional) after the P31 in the suggested P39/P31?/P279*chain.
So it would match somebody that was position held (P39) Minister of the Environment (a specific position, indicated by it being an instance of (P31) a more general class -- and by the capital 'M' in "Minister")
Or it would match somebody that was position held (P39) a minister in the Department of the Environment (a class of positions, indicated by the class being subclass of (P279) the more general class -- and by the small 'm' in "Minister").
In response to your first para, using subclass of (P279) for the capital-M Minister suggests that that item is a class with multiple instances -- leading people to the error of thinking that making <person> instance of (P31) Minister of the Environment is acceptable -- which, as I think you agree, it absolutely isn't. Jheald (talk) 12:41, 18 August 2017 (UTC)[reply]
(edit conflict) Re "people should never be instances of positions", P31 isn't transitive. X P31/P31 Y does not imply X P31 Y. A person being a prime minister and prime minister being a position doesn't mean that the person is an instance of a position.
I'm also confused about what confusion you think will be caused with regards to extra instance-of relations of humans. The P1647 statement isn't exactly visible to the average editor changing any given item. Positions can be subclasses.
IIRC, using P279 for position items has been discussed quite extensively in the past. I understand that you feel it's important to change this before the workshop in London, but the odds of running a broad discussion to establish sufficient consensus to overturn years of common practice within the next 24 hours is essentially nil.
X P279 Y means that instances of X are instances of Y. If prime minister (Q14212) or First Minister of Scotland (Q1362216) had statements indicating that they were subclasses of position (as you said in your first post), then we'd have a problem, as it would imply that people are positions. But there are no such statements. If we used P31 to connect certain positions to more general positions, we'd have quite a few consistency problems and a very broken ontology, and probably a few users getting confused into thinking that P31 is a general "tagging" system.
--Yair rand (talk) 23:20, 17 August 2017 (UTC)[reply]
@Yair rand: So, working through the series of points above:
(i) My key objection to the proposition Prime Minister of the United Kingdom (Q14211) instance of (P31) position (Q4164871) (or in this case instance of (P31) public office (Q294414)) is that it is leapfrogging directly to the top of the tree. This is massively redundant, and is not how we do things on Wikidata -- anywhere. Instead, the statement should be Prime Minister of the United Kingdom (Q14211) instance of (P31) prime minister (Q14212).
From this, since A P31 X P279 Y P279 Z implies that A is also implicitly an instance of the super-classes Y and Z, then by tracing up the chain of subclass of (P279) statements from prime minister (Q14212), this is enough to infer that Prime Minister of the United Kingdom (Q14211) is an instance of "position", so the statement Prime Minister of the United Kingdom (Q14211) instance of (P31) position (Q4164871) is not necessary, and per conformity with standard Wikidata practice, it should be removed.
(ii) Now looking at the P279 statements on Prime Minister of the United Kingdom (Q14211): if we accept that A P31 X P279 Y P279 Z implies that A is also implicitly an instance of the super-classes Y and Z, then this means that making the statement <person> instance of (P31) Prime Minister of the United Kingdom (Q14211) is implying that <person> is an instance of position, because Prime Minister of the United Kingdom (Q14211) P279+ position (Q4164871).
This is one reason for why we don't make statements of the form <person> instance of (P31) Prime Minister of the United Kingdom (Q14211) -- instead we use a specific property position held (P39) to link a person to any particular role they may have, where position held (P39) is not a subproperty of instance of (P31).
By very strong convention on Wikidata, the only instance of (P31) statement that should exist on a person is <person> instance of (P31) human (Q5).
(iii) You say that "there would be a problem if prime minister (Q14212) or First Minister of Scotland (Q1362216) had statements indicating that they were subclasses of position". But the chain prime minister (Q14212) -> subclass of (P279) minister (Q83307) -> subclass of (P279) public office (Q294414) -> subclass of (P279) position (Q4164871) implies exactly this.
(iv) The purpose of P31 is to identify that an item represents a specific identifiable non-subdivide-able concept or entity. It allows the useful distinction to be made between prime minister (Q14212) subclass of (P279) minister (Q83307) (ie "prime minister" is a class of positions with many instances) and Prime Minister of the United Kingdom (Q14211) instance of (P31) prime minister (Q14212) ("Prime Minister" is a unique distinct position).
(v) You make the claim that what I am presenting would lead to "consistency problems" and "broken ontology" -- but you have yet to produce any specific examples of either, that would need to be considered.
In contrast, I have presented examples of consistency problems, broken ontology, and practice strongly at variance with the norms of the project -- which I advocate fixing, and fixing as soon as possible. Jheald (talk) 22:23, 18 August 2017 (UTC)[reply]
@Jheald: minister (Q83307) subclass of (P279) public office (Q294414) is an error, as individual ministers are not public offices.
Re (i): We're not leapfrogging to the top of the tree, because position is on a completely different tree, and is a metaclass. Nothing which is an instance of (P31) position should be P279+ position.
Re specific consistency problems: We have over 15000 items using P39 mayor (Q30185). That's a position, not a class of positions. Having both it and instances of it be accepted P39 values would break consistent meaning of the property's use. --Yair rand (talk) 02:17, 21 August 2017 (UTC)[reply]
@Yair rand: "minister" is a class of posts. On Wikidata individual persons are not ministers, they are humans (Q5), who hold posts (P39).
Similarly mayor (Q30185) is a subclass of public office (Q294414), because there are instances of distinct posts that are members of that subclass: eg Mayor of London (Q38931), mayor of Paris (Q19822699), Mayor of New York City (Q785304).
As you say, there are currently just over 15,000 persons with position held (P39) mayor (Q30185) (query), with the qualifier of (P642) in just under 4,000 cases. (query).
I don't see any problem with this, one simply makes sure that the constraints on the property allow subclasses to ba acceptable for its values, not just instances. (diff). Jheald (talk) 09:50, 21 August 2017 (UTC)[reply]
Please don't make radical changes to widely-used properties without consensus. P39 can not currently be used for classes of positions, only for positions (such as minister, mayor, and mayor of London, according to the current framework).
The problem isn't just the constraints, it's the conceptual meaning of the property. Splitting one property into two uses across class/non-class lines breaks the consistent meaning of the property. --Yair rand (talk) 21:58, 21 August 2017 (UTC)[reply]
Obviously, if it's merely focused on British parliamentarians, people might get confused if you change the existing model overnight. The structure there seems to have been the same for ages. If there are items or queries you built that don't work, we can look into them now.
About the P31/P279 question: I think there are at least three aspects: (a) the problem with mayors (even if I think we should avoid items like Q30185), (b) the question if any office-holder is an instance of a class of office-holders (someone once wrote a lengthy paper on that question, at least about P106), (c) the P31 value on positions. If we replace that with items "position with a single simultaneous office-holder", "position with ..", we could probably simplify your tree view.
--- Jura 23:28, 17 August 2017 (UTC)[reply]
@Jura1: The outline programme for the workshop can be found here: Wikidata:WikiProject British Politicians/Wikifying Westminster. Yes, there is a focus in the examples there on British parliamentarians. But this extends also to what government roles they have. But furthermore, our co-partners for the workshop "MySociety", while originating in the UK, are particularly interested in the availability of information about executive and legislative roles and incumbents worldwide -- as reflected in this Wikidata:EveryPolitician project they are spearheading, and the objectives of their Wikimedia/MySociety funding collaboration: m:Grants:Project/mySociety/EveryPolitician.
So when they are bringing coders who are interested in giving support to their work to an event of Wikidata, when groups will be encouraged to break into project teams and follow their own interests, the data modelling we use for executive government roles is likely to be of interest; may well feed into the project work people undertake; and is likely to form part of what they learn about Wikidata, and the understanding they gain of typical data modelling in Wikidata. So consistency problems, broken ontology, and practice strongly at variance with the norms of the rest of the project are decidedly unfortunate.
To move to some of your more specific points. (a) What problem with mayors? A mayoralty, eg Mayor of London, or Mayor of Paris is an instance of a type (subclass) of positions. More generally, "mayor" (small m) is a class of positions, of which specific mayor positions are specific instances. I don't see anything difficult or controversial or problematic in this. (b) I don't know of any paper in the past about P106, but the position now is quite clear. occupation (P106) is not a subproperty of P31. P31 should not be used to relate persons to occupations. The value of P31 for persons should always be human (Q5) and should only be Q5. (c) I'm not sure that I see why what you are suggesting would particularly help with anything I have been talking about. But if you feel that adding that distinction into the subtree of position would be a useful distinction, it could certainly be considered. I think we would still typically need subclasses "minister in the UK Department of X", which particular ministerial positions would be instances of. And I think we would still need to allow the possibility of a position held (P39) statement having one of those subclasses for its value, rather than a particular named ministerial position, to allow for cases where we don't have the data of the specific position, just the class of position.  ::::But I do see that e.g. there are presently two ministers with the title "Minister of State" in the current UK Department of Education (see en:Department_for_Education#Ministers), so if we have an item for the role "Minister of State, UK Department of Education", it probably is useful to be able to indicate that more than one person can hold this role concurrently. My instinct is that this would be better done via a property on the post, rather than representing the fact via an additional subclass. But I'd certainly be open to discussion on that point. Jheald (talk) 23:20, 18 August 2017 (UTC)[reply]
@Jheald: I don't understand why A particular person is not an "instance of" painter. It is logical to name someone who paint a painter. I think it is possible to be a painter... --Infovarius (talk) 10:59, 18 August 2017 (UTC)[reply]
@Infovarius: See my response (ii) to Yair Rand at 22:23 for an example of one of the difficulties that this can create.
If "A is an instance of X" then the statement "A is an instance of Y" should also make sense for all super-classes Y of X. If it doesn't, then that's an indication of a "bad smell" around the modelling.
We try to avoid this problem by making the choice that painter (Q1028181) specifically is an item for the occupation, not for a group of people. Jheald (talk) 23:38, 18 August 2017 (UTC)[reply]
@Jheald: You have fallacy in (ii). If Prime Minister of the United Kingdom (Q14211) instance of (P31) position (Q4164871) then it doesn't implicate Prime Minister of the United Kingdom (Q14211) P279+ position (Q4164871) and so from some person having instance of (P31) Prime Minister of the United Kingdom (Q14211) it doesn't mean that this person instance of (P31) position (Q4164871). So no difficulties here. --Infovarius (talk) 21:30, 20 August 2017 (UTC)[reply]
@Infovarius: Please read again my response to Yair Rand at 23:20, 17 August 2017 (UTC).[reply]
My objection to Prime Minister of the United Kingdom (Q14211) instance of (P31) position (Q4164871) is that the value of the P31 ought to be something much more precise, eg Prime Minister of the United Kingdom (Q14211) instance of (P31) prime minister (Q14212), as I set out in point (i) of my answer to Yair.
If you read point (ii) of my answer to Yair again, you will see that I am quite clear that the "person is an instance of position" inference would not be drawn from the P31 claim on Q14211, but from the P279 chain. Specifically, it follows via chains of statements like Prime Minister of the United Kingdom (Q14211) subclass of (P279) prime minister (Q14212) ... subclass of (P279) minister (Q83307) ... subclass of (P279) public office (Q294414) ... subclass of (P279) position (Q4164871)
So <person> P31 Prime Minister of the United Kingdom (Q14211) is an implicit claim that <person> instance of (P31) position (Q4164871).
On Wikidata we avoid this problem by insisting that a <person> is only to be P31 human (Q5); and that items like painter (Q1028181) or Prime Minister of the United Kingdom (Q14211) refer specifically to the occupation or role, not to any set of persons who may have carried it out. Jheald (talk) 22:36, 20 August 2017 (UTC)[reply]
I think we noticed this assumption in your explanation.
--- Jura 10:01, 21 August 2017 (UTC)[reply]

Similar P31/P279 issue with executive bodies[edit]

I've run into a similar issue to the above discussion about positions in the case of Finnish Government (Q2366737) - the executive body in Finland. Normally in other countries we have designated these as "instances" (P31) of governments or whatever the relevant class is (probably cabinet (Q640506) in this case). However, in wikidata currently it has been stated to be a "subclass of" (P279) cabinet (Q640506), because the separate instantiations of this governing body have been entered as its instances - for example Matti Vanhanen's first cabinet (Q81988) instance of (P31) Finnish Government (Q2366737). I think both sides of this are arguably correct, just different points of view (i.e. you can see Finnish Government (Q2366737) as an individual thing which changes over time, or you can see it as the class whose instances are its state at various points in time). But allowing both within wikidata is definitely confusing - we should probably pick one and be consistent about it across the project, just as we have picked one way of doing P31 for people. I think the best way here is to view a government body (or any other continuing identifiable single entity) as a single thing that may change over time, and use some other property (perhaps facet of (P1269) ?) to link the state at specific points in time to it. To be consistent this would suggest Mayor of London (Q38931) should be an instance of, and not a subclass of, mayor, position, etc. ArthurPSmith (talk) 15:16, 23 August 2017 (UTC)[reply]

I'm not sure this is really equivalent. If we were talking about whether a theoretical item for "Fourth president of the United States" would be an instance, subclass, or facet of "President of the United States", that would be closer to equivalent to the issue with cabinets.
A more similar situation is, in my opinion, occupation (P106)s. A person can have occupation (P106) biologist (Q864503), which is subclass of (P279) scientist (Q901), but there are also subclasses of it such as zoologist (Q350979). All three items are profession (Q28640)s, but should there be a "bottom level", which is simply an instance of the level above it? --Yair rand (talk) 06:42, 24 August 2017 (UTC)[reply]
well I said they were "similar", not "equivalent" - your "Fourth president of the United States" case is definitely much closer and a good analogy. On occupation (P106), I would say that's essentially the same as the situation for position held (P39) - any generic (lower-case) occupation can be subdivided into different groups and so should be treated as a class (and so subclass of other types up to "profession"), but if there are (and I'm not sure anything that would be in wikidata qualifies) "occupations" that are so specific that only one person in the world can ever hold them at a time, I would argue that should be a instance of (P31) case - anything I can think of along those lines seems more like position held (P39) than occupation (P106) though. ArthurPSmith (talk) 14:23, 24 August 2017 (UTC)[reply]

Changes in head of government roles…[edit]

I'd like to start an open conversation about to handle changes to the role of head of government itself. These changes come with constitutional changes and coups d'etat. They explain why Aung San Suu Kyi (Myanmar), Abdullah Abdullah (Afghanistan), and perhaps Fayez al-Sarraj (Libya) have no predecessor. For a particularly complex example, see en:President_of_Bolivia: President alternates with Liberator of Bolivia, Chief, Junta of Commanders of the Armed Forces (held by several people), Chairman of the Military Junta, and Co-Chairman of the Military Junta (two people held that title at once). If we can come up with a clear way to code that case, I think we'll have a framework that works in general. (Actually, we would also have to figure out how to handle overlapping chains of sucession a la en:Antipope, but that's more complicated.)--Carwil (talk) 13:36, 25 November 2017 (UTC)[reply]

  1. So, I'm a regular Wikipedia contributor who is also an academic researcher (principally on Bolivia). (I got here and jumped into the conversation basically because I'm a "domain expert" as described in this talk.) In the course of my work, I come across factual information and add it to English Wikipedia, and I sometimes scrape data and run basic analysis on the scraped results. Since a lot of what I add is in tabular form, and my working language (English) differs from the main languages of the country I work on (Spanish, Quechua, Aymara), I would ideally like to add material in Wikidata form. But I'm completely unwilling to do line-by-line data entry to do so. So maybe the project isn't ready for users like me to put our energy here.
  2. Regardless, I'd like to help make the data structures that describe things work for the cases I work on. In the simplest form, this means that a Wikidata Query should be able to generate the table on en:President_of_Bolivia in any language. Right now, however, the data is very incomplete (see User:Carwil/BoliviaPresidents) and there is no clear model for generating a list that includes multiple offices (and hence multiple Q codes) for the head of government over time. Or is there?
  3. Conversely, what I would like the dataset to facilitate is simple queries about Bolivia's leadership, such as: How many days has Bolivia been ruled by an elected head of government since 1936? (I've data scraped and run an analysis on that question actually, but Wikidata is useless for addressing it unless and until political offices include information about how people came into them.) Or, which heads of government left office early due to popular unrest? Or, which Bolivian heads of government came to power as representatives of the military and which as representatives of political parties, and were any both or neither?
  4. Now, could I just make up properties that might enable queries on these questions? Of course. (There would be a SPARQL learning curve to climb first, but that's my problem.) But unless and until a WikiProject like this one agrees upon and standardizes such information, there can be no expectation that the datasets about politicians (or even this narrow case of heads of state and government) will enable such queries.
  5. One more point as a "Wikidata Muggle": Unless and until Wikidata facilitates some kind of key properties template (that is, when users add "President of Bolivia" to a person, they are prompted to qualify that with start and end dates, predecessor and successor, political affiliation, cause of assuming office, etc.), the dataset will be filled with blank spaces and inconsistencies. I know that problems 1 and 5 are beyond the scope of this WikiProject, so let's pretend I'm just asking for a discussion about 2, 3 and 4.--Carwil (talk) 03:08, 26 November 2017 (UTC)[reply]
If a position has a replaced by (P1366) reference to its successor position, and there's a replaces (P1365) back again, then it's possible to construct a fairly simple query that finds all the holders of an office, even across such changes. (This obviously gets complex if two offices are merged into one, but that's going to be more unusual for heads of government than for, say, ministerial positions). In terms of how people came to office, the position held (P39) statements can include elected in (P2715) (as I've added to Evo Morales (Q42410)), or, where not elected could include a different has cause (P828) — though without some consistency around that it's going to be difficult to write queries like the ones you suggest above. Similarly, we can add end cause (P1534) qualifiers explaining why they leave office (we're using that on British MPs at the moment; see Wikidata:WikiProject British Politicians#Core_properties_and_qualifiers). I haven't seen any approach yet for saying that someone is a military representative, though I agree it would be good to find one. I'm happy to help out trying to make the data for president of Bolivia (Q373548) good enough to be able to answer these sorts of queries. --Oravrattas (talk) 07:58, 26 November 2017 (UTC)[reply]
Okay, so walk me through this with Junta of Commanders of the Armed Forces 1971 (Bolivia) (Q16911605), the Military Junta that took power on August 21, 1971, before handing power to one of its members, Hugo Banzer (Q269039). They need to be an instance or a subclass (both P values, but which do I use) of a "Military junta of Bolivia" (not in Wikidata yet), right?
Once that's sorted out "Military junta of Bolivia" should be office held by head of government (P1313) of Bolivia, with a long series of start and end dates, and a subclass of military junta (Q28418808), right? But does Junta of Commanders of the Armed Forces 1971 (Bolivia) (Q16911605) replace Juan José Torres (Q195762), the president they overthrew? That would make sense, but "replacing" is a property of a position and the Junta of Commanders of the Armed Forces 1971 (Bolivia) (Q16911605) and Juan José Torres (Q195762) don't share any position (other than Head of government of Bolivia, which isn't a Q code at all, but instead office held by head of government (P1313) of Bolivia).
It should go without saying, but once there are answers to these questions, there needs to be a domain-specific help page that walks people through these things when entering data for new countries. Who's working on that?--Carwil (talk) 16:54, 27 November 2017 (UTC)[reply]
office held by head of government (P1313) or office held by head of state (P1906) just lists offices .. If there are several, just add start/end dates: Q2007#P1313. Not sure how a help page for that would look like. It's a fairly basic Wikidata structure. Personally, I don't think replace qualifier are of much as use, as it's should be possible to get the sequence from the dates. Once the office defined, one just needs to query its holders. The problem with a collective office slightly complicates that. Either you list all simultaneously directly in P39 (as https://tools.wmflabs.org/reasonator/?q=Q258045&lang=en ) or you create a "member of" position (e.g. Q11811941). Both slightly complicate the query if you actually want to list its holders.
--- Jura 18:21, 27 November 2017 (UTC)[reply]
Okay, so I'm creating Q43893109 as a public office and a military junta (subclass of each). So is Junta of Commanders of the Armed Forces 1970 (Q6314366) position held (P39) Q43893109, or is Junta of Commanders of the Armed Forces 1970 (Q6314366) instance of (P31) Q43893109?--Carwil (talk) 20:58, 27 November 2017 (UTC)[reply]
For the presidents part, the list at "position held" on https://tools.wmflabs.org/reasonator/?q=Q373548&lang=en is a good progress indicator.
--- Jura 18:53, 27 November 2017 (UTC)[reply]
Yes, but that returns us to the original problem: we have a list of presidents, but not of heads of government. Does creating "Head of government of Bolivia," with "President of Bolivia" and "Military junta of Bolivia" as subclasses solve that problem, or not? (It could if timelines include inherited subclasses, but it could not if the timeline only shows the null set of Q items that P39 to "Head of government of Bolivia.")
P.S. To say that a manual isn't needed is to say that every new user will make the same guesses on these questions. Which is false. I wouldn't even make the same guesses twice and I'm one person.--Carwil (talk) 21:03, 27 November 2017 (UTC)[reply]
I don't quite see what you are trying to go.
The basic structure can be much simpler: compare with Q238 that has Q238#P1313 with Captain Regent of San Marino (Q258045) as value. Then each holder (person) has position held (P39) with Captain Regent of San Marino (Q258045) as value, sample: Q16189465#P39. This can be used to generate a full list of holders.
If there are several different offices that need to be in P1313, just add all of them (e.g. Q2007#P1313). Then each holder (person) has position held (P39) with the applicable offices as value.
P31 has no impact. It can work the same way with office held by head of state (P1906).
--- Jura 21:26, 27 November 2017 (UTC)[reply]
But that produces an incorrect, or at least an inadequate, list. This is the correct one: en:President of Bolivia#List of Presidents of Bolivia. There shouldn't be gaps of time when the head of state had a different title. Moreover, the replaces/replaced by chain ought to connect people even when the title changes.
Also, how can I indicate the start and end times of say "President of Bolivia" as office held by head of government (P1313) of Bolivia? President starts and stops being the Head of government numerous times.--Carwil (talk) 21:42, 27 November 2017 (UTC)[reply]
Did you have a look at Northwest Territories?
--- Jura 21:46, 27 November 2017 (UTC)[reply]
For the chronological retrieval of the actual holders, it's important that dates are set on the P39-statements for each person. Currently, this is incomplete for Presidents of Bolivia, check "position held" on https://tools.wmflabs.org/reasonator/?q=Q373548&lang=en
--- Jura 21:57, 27 November 2017 (UTC)[reply]
Yes, I saw Northwest Territories. But I did not see a way to generate a list of heads of government of Northwest Territories that includes all three positions. Moreover, the titles of heads of government in NWT are sequential. In Bolivia, they alternate. That's why I asked about multiple end dates above.
Also, Jura, you keep responding as if I'm trying to generate a list of presidents of Bolivia, but what I want is a list of all heads of government of Bolivia, which is what exists here: en:President of Bolivia#List of Presidents of Bolivia. That includes people designated, Liberator of Bolivia, Provisional President, Dictator for Life, and groups of people designated Co-President, Government Junta Military Junta, Junta of the Armed Forces. It also includes people who held one of those offices and then changed their title to another. What data needs to be entered to generate a common chronological list of those people? What query would produce it?
To repeat the questions I still have unanswered…
  1. Does Junta of Commanders of the Armed Forces 1971 (Bolivia) (Q16911605) "replace" Juan José Torres (Q195762), the president they overthrew? That would make sense, but "replacing" is a property of a position and the Junta of Commanders of the Armed Forces 1971 (Bolivia) (Q16911605) and Juan José Torres (Q195762) don't share any position (other than Head of government of Bolivia, which isn't a Q code at all, but instead office held by head of government (P1313) of Bolivia).
  2. Is Junta of Commanders of the Armed Forces 1970 (Q6314366) position held (P39) Q43893109, or is Junta of Commanders of the Armed Forces 1970 (Q6314366) instance of (P31) Q43893109?
  3. How do I code the members of a Military Junta?
  4. How do I code co-presidents? Do I have to create a "co-presidency of X and Y" Q-item? How do I handle "replaces" and "replaced by' for them? (There was only one co-presidency, followed by singular presidents.)
  5. I understand how to code elections, but how do I code fact that someone came to power in military coup?
--Carwil (talk) 15:38, 28 November 2017 (UTC)[reply]

coup?

@Carwil: I'm travelling at the minute so can't reply in as much depth as I'd like yet, but this is a great example, and I agree that if we can work out the correct approach to this (and document it well!), it will be very useful in other situations too. One question I have that could be important: for any of the positions that have been head of government (e.g. President), were there also holders of the position who were not, at that time, head of government? In other words, did the position continue to exist, but was not head of government for a while, or did the position cease, and then perhaps get re-instituted later? --Oravrattas (talk) 08:24, 29 November 2017 (UTC)[reply]
I think if you start with the Northern Territory sample it should work out. Complete Q750, the for each holder, add P39 with the appropriate office (with start/end date qualifiers). Sometimes it's hard to visualize before actually retrieving the data.
--- Jura 10:45, 29 November 2017 (UTC)[reply]

Current consensus about ambassadors structure[edit]

Hello there, I have been away from wikidata for a while and I was wandering what is the current consensus about ambassadors structure.

In 2017, I proposed a structure and but there was some disagreements about whereas to put "ambassador of X" and "ambassador to Y" as instance of (P31) or as subclass of (P279).

I advocated in favor of using instance of (P31) because in my opinion:

  1. there should be no instance of "ambassador of X to Y"; the only correct use beeing as target of position held (P39), with items having instance of (P31)>human (Q5)
  2. with this method, each "ambassador of X to Y" item is instance of (P31)>function (Q11348) by transitivity of instance of (P31)+subclass of (P279)
  3. with this method, each "ambassador of X to Y" item is also instance of (P31)>ambassador (Q121998) and so on (per the same mechanisme), allowing to perform requests in a easy and precise fashion.

Of course, I am open to other suggestions as long as you explain their advantages and consistency.

The current state of the database is a mix of the 2 solutions (some item having instance of (P31), some subclass of (P279)), and this made queries very unpractical. If we could find a consensus, maybe we could add some constraints to avoid having items used with the wrong property (and if the solution I suggested is adopted to forbid these items to have any subclass of (P279) defined), and in any case having both "ambassador of X" and "ambassador to Y" set but never one alone.

I have seen some recent discussions about this topic, and I would like to centralize this here to let everyone able to be involved about this:

Let me know if there is some other discussion I missed about this topic.

I tried to notify every people involved in these previous discussions. If by mistake, I forgot someone, please notify them.

— Metamorforme42 (talk) 18:36, 27 May 2022 (UTC)[reply]

I add Topic:Wxanljqt5oakzibz to the list. I also notified WikiProject Ontology — Metamorforme42 (talk) 08:43, 12 June 2022 (UTC)[reply]
Thanks for linking to previous discussions. Though, you sort of overlook suggestions and explanations that there are already. There's not much to reply that I or other users already didn't say in these previous discussions.
No other item uses "ambassador of X to Y" as P31 value, but this isn't relevant really. See Help:Basic membership properties where it says at the top that "Classes are abstractions (that might or might not have instances at Wikidata)". "Ambassador of X to Y" position is an abstract object (class) the same way as ambassador (Q121998) or diplomat (Q193391) is. "Ambassador of X to Y" is just more specific than "ambassador", the same way as "ambassador" is more specific than "diplomat", and in Wikidata "subclass of" relation is used to state that one class is more specific than another. "Ambassador of X to Y" class as an abstract object doesn't concretize on which individual exactly occupies certain ambassador position at some point. In person items we don't use "ambassador of X to Y" as P31 value under our common schema (and I do not suggest we should), but nonetheless semantically it would be correct to do so.
As for consistency, your schema is inconsistent to how most (or maybe all) other position items are modelled. Take for example Mayor of London (Q38931), mentioned in a previous discussion that you linked above, which isn't set an individual mayor.
As for queries, as far I can see, you can perform these at least as reliably if P279="ambassador of X" statement is properly used in "ambassador of X to Y" item. For example, you can query all classes that are subclasses of subclasses of "ambassador of a country" (classes that are two levels below the latter), or query all subclasses of "ambassador" (at all levels) that also have P17 statement and/or P31=position statement. 2001:7D0:81E6:EF80:44E0:95:E9F2:1936 08:10, 13 June 2022 (UTC)[reply]
In the help page you quoted, abstract object is not defined by "if you cannot touch it, it is abstract", but by "represents a set of items, called its instances". I think this is the main source of confusion. See in the example section "2016 Taiwanese presidential election (Q20683626) instance of (P31) public election (Q40231)" (this is simplified because in reality it is instance of a subclass of this item).
You said "In person items we don't use "ambassador of X to Y" as P31 value under our common schema (and I do not suggest we should), semantically it would be correct to do so", and I do not agree because it would not even make sense semantically for an human (Q5) to be instance of (P31) position (Q4164871) (by transitivity of subclass of (P279)): position (Q4164871) are not living people.
IMO, an ambassador of X to Y is not an abstract object. Take for example ambassador of France to Germany (Q20942020): it is clearly a position of ambassador. And this position is unique at a certain point in time (this is why we use replaced by (P1366) and not followed by (P156)), therefore it is not a class.
Also, Mayor of London (Q38931) is both a instance of (P31) public office (Q294414) and a subclass of (P279) public office (Q294414) (strategic regional authority mayor (Q110418288) < Directly elected mayors in England and Wales (Q5280485) < mayor (Q30185) < public office (Q294414)), and this is inconsistent. I want to avoid the same kind of inconsistency on ambassadors. By the way, Mayor of London (Q38931) is set to an individual mayor at a certain point in time.
I agree for queries. One can always create a query matching a specific model.
I am looking forward for project Ontology to confirm or infirm this. They are used to this kind of questions. — Metamorforme42 (talk) 09:10, 13 June 2022 (UTC)[reply]
I didn't suggest that instances (concrete objects) can be only physical objects. 2016 Taiwan presidential election being a concrete election is of course correct. The same way, say, ambassador of Iceland to Sweden can have P31=position statement as it is a concrete position. But ambassador of Iceland to Sweden nonetheless is not a concrete ambassador, unlike a concrete person holding this position. On a higher level, "ambassador" is also a concrete diplomatic rank, but in our classification scheme primarily is a diplomat subclass that in turn has subclasses.
Any class item, having a P279 statement, can be also an instance of some metaclass (class for classes), and in this case "position" can be considered a metaclass. It isn't mandatory to use metaclasses, but sometimes it's useful.
it would not even make sense semantically for an human to be instance of position – Yes, but this would be indicated only if "ambassador of X to Y" is set as a subclass of position, instead of instance of position.
Yes, this is true that an item should not be set as an instance of and a subclass of the same item, forming class circles. Class trees around public office (Q294414) appear to be somewhat messed up indeed and should be fixed. This sort of mistakes can easily occur on higher levels of class tree as well, but they are fixable and shouldn't be a reason to avoid classes altogether, though.
Mayor of London is set to an individual mayor at a certain point in time – do you mean that Mayor of London item lists individual mayors and their terms? Well, yes, it can link related items, but these mayor instaces nonetheless aren't inherent to Mayor of London class that can exists even if no concrete mayor (person) yet hasn't or currently doesn't occupy this position. 2001:7D0:81E6:EF80:41AC:C9A7:FFE8:7A39 13:00, 13 June 2022 (UTC)[reply]
I attach aside a visualization of the ontology I propose. The goal of using classes of positions, is to avoid setting direct instance of positions and use "P31/P279*" chain instead.
do you mean that Mayor of London item lists individual mayors and their terms? – by "is set to", I mean officeholder (P1308)/position held (P39) does that on this item, but this item is of course not a list by itself. This comment was only to highlight the fact that there is only a single Mayor of London position, and I can't see what would be instance of Mayor of London position since to me it is already a concrete item and not a class.
What do you mean by But ambassador of Iceland to Sweden nonetheless is not a concrete ambassador, unlike a concrete person holding this position.? What is for you a concrete ambassador, for me this term is a shortcut for "concrete ambassador position" (and this specific position would be an instance of it), not to be confused with the "office holder" who is also concrete but cannot be position since it is human (only hold a position).
With the model you propose, could you give an example (even a fictional one) of an item with statement instance of (P31)ambassador of Iceland to Sweden (Q38351320) to clarify? I mean per Help:Basic membership properties even if The complete set of instances implied by a class might or might not exist as items within Wikidata, subclass of (P279) is used to state that all the instances of one class are instances of another, so this should not be difficult to create a fictional one if the model is consistent.
Also it would really help if a visualization like aside were drawn for your model, to better understand what is an instance, a class, a metaclass and so on. From what you said, I understand that:
  • position moved to the metaclass column so there is no subclass of position, only instances
  • ambassador is no longer a subclass of position
  • ambassador of France to Germany is moved to the class column, and is subclass of French ambassador and subclass of ambassador to Germany
  • ambassador of France to Germany is an instance of position
I think one of main disagreements is about whether ambassador (Q121998) should be or should not be a subclass of position (Q4164871). For me it should for you it should not.
Please correct me if I misunderstood any point. — Metamorforme42 (talk) 16:13, 13 June 2022 (UTC)[reply]
avoid setting direct instance of positions – You mean in item of an individual person otherwise P31 would be used instead of P39 to link to a position? This is common that P31 subproperties instead of P31 itself are used to link to certain kinds of classes, e.g. also occupation (P106) or heritage designation (P1435) is used to link to something that the item subject is. It shouldn't be necessary to manipulate the classes to force this.
it [Mayor of London] is already a concrete item and not a class – As far as I can see, there isn't a contradiction really. As already pointed out, as a class it can have both P279 and P31 statements. It is a concrete position, but also a kind of mayor (abstract object) corresponding to concrete persons that have been mayors of London.
Yes, I'm referring to a person that is a concrete ambassador. This is to follow the natural language where it's normal to say that Benedikt Sigurðsson Gröndal (Q712686) is an ambassador. If position is meant, then one would normally specify, e.g. we'd say that ambassador position is vacant, not ambassador is vacant.
Note that ambassador (Q121998) is also set as a subclass head of mission (Q1251687), which in turn is in subclass tree of person (Q215627). This contradicts the P279=position statement in the same item. This contradiction should be solved somehow, but it's probably better to discuss this separately somewhere else. In any case ambassador (Q121998) is a class that can have subclasses.
In your drawing, the line between instances and classes seems arbitrary. This is for the reasons outlined above. In other aspects I don't have a problem with this model.
I propose the same model you see in other ambassador items that don't follow your schema, and the same model you see is used for most other positions in Wikidata. E.g. there's class ambassador of Romania to the Gambia (Q64698625), set as subclass of ambassador to the Gambia (Q30249893) and ambassador of Romania (Q30158711). Though, I don't feel strongly about P31=position (Q4164871) statements. It probably wouldn't matter a lot if the latter is just omitted, or possibly some other (new) metaclass could be linked instead from items of ambassador positions. 2001:7D0:81E6:EF80:8100:9793:D88B:DEEE 09:17, 14 June 2022 (UTC)[reply]
You mean in item of an individual person otherwise P31 would be used instead of P39 to link to a position? – No, I mean on the position itself.
This is common that P31 subproperties instead of P31 itself are used to link to certain kinds of classes – See on section Wikidata_talk:WikiProject_Heads_of_state_and_government#P31_(instance of)_and_P279_(subclass_of) above position held (P39)subproperty of (P1647)instance of (P31) is no longer true.
a class it can have both P279 and P31 statement – Yes, but then the instance of (P31) value must be instance of metaclass.
we'd say that ambassador position is vacant, not ambassador is vacant – this may be true in English but not in every language, for example in French we say "Le poste d'ambassadeur est vacant." (meaning "The ambassador position is currently not hold"). Items on wikidata are about concepts, not language; if you want to link natural language senses of words to concepts you should use lexemes instead.
it's probably better to discuss this separately somewhere else – I agree.
Though, I don't feel strongly about P31=position (Q4164871) statements – The instance of (P31)position (Q4164871) (or a subclass of it) is required by constraints on position held (P39).
possibly some other (new) metaclass could be linked instead from items of ambassador positions – Then how should we call it? maybe "class of ambassadors between two countries"? (by opposition to "class of ambassadors from a country to an international organization" (like OTAN)).
I will try to draw a visualization of your proposal. Your model may be better after all if it allows us to specify the exact diplomatic rank (Q303618) a position is related using subclass of (P279)Ambassador extraordinary and plenipotentiary (Q20068736) or subclass of (P279)high commissioner (Q123695) in addition to the countries (I discovered these two items very recently and don't know yet the exact nuance in meaning). — Metamorforme42 (talk) 11:47, 14 June 2022 (UTC)[reply]
Here it is. Is it what you expected? If you want I can still modify it. If it matches to what you propose, I can create a new Wikidata:Requests for comment, in hope to have more opinions than just ours. — Metamorforme42 (talk) 12:34, 14 June 2022 (UTC)[reply]
Honestly, this image looks better for me. --Infovarius (talk) 20:00, 14 June 2022 (UTC)[reply]
These remarks on whether P39 is subproperty of P31 (formally or not) again comes to the question whether position items are (also) about types of persons. Given P39 constraint is of "instance or subclass of" relation (indicates inconsistent schemas), and so omitting the P31=position statement alone wouldn't violate the constraint. But yes, it's probably safer to keep it, or use P31 to link to some other more clear-cut metaclass.
Strictly speaking, ambassador as a type of person and ambassador as a type of position, are different. So, if new "ambassador position" metaclass is created then this would be the item for ambassador as a type of position, and ambassador (Q121998) could be kept as the item for ambassador as a type of person. P279=position statement could be moved from the latter to the metaclass item then. To me simply "ambassador position" would seem fine as metaclass label. Another option is to just live without an item that is for ambassador as a type of position, and keep P31=position (Q4164871) statements.
I contrasted these language examples only to illustrate the difference between ambassador in these two different senses that correspond to concepts that we deal with here.
Yes, this image looks better. Though, really it feels to me that these schemas sort of distract us from the main issue. It shouldn't be necessary to decide on the whole schema (metaclasses and all) in order to fix/harmonize P279/P31 usage for ambassador positions in particular. Metaclasses are generally built around classes, they are not supposed to dictate relations between classes. 2001:7D0:81E6:EF80:1D3C:EB34:31A2:369C 09:37, 15 June 2022 (UTC)[reply]
@Metamorforme42: I'm afraid I find this model very confusing. Given the current set-up, how do I generate the list of all the senior diplomat positions for a given country — e.g. the UK's list at Heads of Diplomatic Missions of the United Kingdom? (NB: I'm only talking about the positions themselves here — not who is holding them) --Oravrattas (talk) 08:56, 4 July 2022 (UTC)[reply]
I would say, by querying every ?itemP279*/P31ambassador of the United Kingdom (Q18115939) with the first model of the discussion or ?itemP279*ambassador of the United Kingdom (Q18115939) with the second model. Currently only a few are linked for United Kingdom… Does it seems correct? — Metamorforme42 (talk) 18:01, 4 July 2022 (UTC)[reply]
@Metamorforme42: I think I'm still missing something fundamental here. Where does something like High Commissioner of the United Kingdom to Bangladesh (Q56760944) fit in with this? There are currently somewhere in the region of 200 relevant UK position items, but I can't see how to get them all from a single query without enumeration of disconnected trees. --Oravrattas (talk) 20:25, 4 July 2022 (UTC)[reply]
IMO a query like https://w.wiki/5Q4m ought to include returns for all of the High Commissioners of the UK. At the moment it doesn't, because the statement
High Commissioner of the United Kingdom (Q56760832)subclass of (P279)ambassador of the United Kingdom (Q18115939)
is currently missing; IMO that statement should be added.
I also agree that a specific post that is unique, like Mayor of London (Q38931) should not have any subclass of (P279) statements, only instance of (P31) statements. Same for specific ambassador posts. Jheald (talk) 21:50, 4 July 2022 (UTC)[reply]
@Jheald: I think those trees do need connected up somehow, but I don't really know enough to know whether this is best modelled such that High Commissioners and Permanent Representatives etc all have a direct subclass of (P279) relationship to Ambassador, or whether these should all be a subclass of (P279) of slightly more abstract item. That said, I'm very much in favour of making the data as usable as possible as quickly as possible, and then refining the model later. --Oravrattas (talk) 05:04, 5 July 2022 (UTC)[reply]
@Jheald: it is a specific post, but this in itself isn't reason to omit P279 statements really. This was discussed above. Class (item with P279) can have P31 statements as well. In addition to being a specific post it is also a class of specific persons that have hold this post. The same way as parent class "ambassador" is a specific diplomatic rank, and also a type of person, type of civil servant etc. 2001:7D0:81E6:EF80:C93B:58CB:BACF:7637 08:40, 5 July 2022 (UTC)[reply]

@Metamorforme42: what do you see as the way forward then? Can we now finally switch back to normal use of classes/metaclasses in items of ambassador positions, i.e. link parent classes like Q30249893 using P279? Do you insist that new "ambassador position" metaclass should be created, or can we keep using more general position (Q4164871) as P31 value, or is it perhaps fine to just omit P31 statement in items of ambassador positions? 2001:7D0:81FD:BC80:14F8:9D55:2B1C:1789 06:53, 20 May 2023 (UTC)[reply]

Heads of state and government for Indigenous Nations/First Nations/etc[edit]

I'm working on organizing a slate of work across the Wikiverse on better representation for sovereign governments ("states") for indigenous groups. My focus is on Native American Tribes in the United States because that's where I am based and am a citizen of one of those Tribes (Cherokee Nation). The two core concepts here, head of state and head of government, apply to these sovereign nations that pre-date the imperial nation(s) whose jurisdictions they now lie within. It is important to nail down the details about these roles and their history through time as much of this is often obfuscated within the predominant colonial context of the "conquerors."

I don't currently see any major issues with how this project has played out and the schemas/property constraints that have resulted. I will be working to test and apply the principles to the Native Nations context for at least the 574 Federally recognized Tribes along with the State recognized Tribes in the U.S., which will result in some interesting new additions to Wikidata as we look at the unique titles and specific roles that the constitutions or other formational documents have put in place. I am also looking to extend beyond these two concepts to a full representation of Tribal Governments as they exist today, including legislative and judiciary roles where those apply. I invite anyone else with an interest in this particular context for head of state/government concepts to join in and contribute/discuss. Skybristol (talk) 16:48, 6 October 2023 (UTC)[reply]

Guidance on references[edit]

I do not see much if any current guidance on references for the claims being addressed through this project. In examining this aspect of better representing Tribal Governments in the U.S., I am finding it necessary to also work on better characterization of the statutory authorities and foundational documents that should serve as references here (IMO). For instance, as I work to document the history of Principal Chief of the Cherokee Nation, it is necessary to document the distinct Cherokee Constitution history from 1999, 1977, and 1839. The named role has existed since at least the 1839 constitution, but there are a number of important dynamics that have occurred over this time as a result of U.S. Indian policies. Part of this is captured already in Wikipedia, WikiSource, Wikidata, and elsewhere, but it needs work.

This is different, of course, all over the world in various sovereign nations where these concepts apply. However, I would think we would want to provide guidance on the key claims that are part of this schema that should really have appropriate references backing the claims. If we say that a term is 4 years, where is that legislated or documented in an enforceable way? Skybristol (talk) 16:57, 6 October 2023 (UTC)[reply]

Retrieved from "https://www.wikidata.org/w/index.php?title=Wikidata_talk:WikiProject_Heads_of_state_and_government&oldid=1987373433"