Wikidata talk:WikiProject Parliaments
Legislature hierarchy
[edit]The "instance of" for national level legislatures at the moment is a bit of a mess. I've tried to make sure that every country has its legislative body (P194) set, and that every national legislature is correctly marked as a instance of (P31) of unicameral legislature (Q37002670) or bicameral legislature (Q189445) (with the individual houses as upper house (Q637846) or lower house (Q375928) where appropriate — i.e. most countries other than Germany), but we also have quite lot of items that also have a P31 of parliament (Q35749), as well as a sprinkling of rada (Q2323591), majlis (Q10858057), Council of state (Q3687335), etc.
Is there sufficient value in saying that something is both a bicameral legislature (Q189445) and a parliament (Q35749), or do people think instance of (P31): bicameral legislature (Q189445) will suffice? If we want both, should we add instance of (P31): parliament (Q35749) (or a subclass/equivalent) to all the ones missing it? There are quite a few things set on parliament (Q35749) that it seems should be in the tree somewhere, but the legislature (Q11204) hierarchy could do with being cleaned up more generally: https://tools.wmflabs.org/wikidata-todo/tree.html?lang=en&q=Q11204&rp=279
Or perhaps there's a better way of marking whether something is unicameral or bicameral other than via instance of (P31)? There are quite a few cases where Parliaments used to be bicameral, and are now unicameral (and a couple where that has flipped several times), and constructing queries to get a list of all the legislative bodies that actually have members (rather than being the "parent" of individual houses) is surprisingly tricky, with quite a few edge cases. --Oravrattas (talk) 14:01, 22 August 2017 (UTC)
- an interesting edge-case example is New Zealand Parliament (Q1520966). This is currently set as the legislative body (P194) for New Zealand (Q664), but it is a unicameral legislature (Q37002670) (since 1951), and has part(s) (P527): House of Representatives (Q2034424), which is what people have a membership of, etc. In pretty much every other case of a unicameral parliament, this level of indirection doesn't exist: the target of the legislative body (P194) is the body in which people sit. There are various approaches we could take here:
- We could set instead, and leave New Zealand Parliament (Q1520966) as a part of (P361) on that.
- We could harmonise the instance of (P31) tree everywhere so that there is a clearer separation between "legislative body that has members" and "legislative body that only contains other legislatures"
- We could achieve the same result as #2 through use of some other property
- We could accept there can always be this degree of separation and indirection, and document the queries that will be required to get, for example, every national legislature that has members (perhaps by querying for the existence of number of seats (P1342), or has part(s) (P527)|has part(s) of the class (P2670) a subclass of legislator (Q4175034)
- Other options I haven't thought of yet…
- Previously #4 wouldn't have worked, as most national legislatures didn't have those fields set, but over the last few weeks I've ensured that they all do. This makes it at least a viable fallback option, though it doesn't seem ideal.
- Thoughts? Suggestions? Comments? --Oravrattas (talk) 06:52, 23 August 2017 (UTC)
- Maybe 5: create a separate item for pre-1951.
For the other point: supposedly most parliaments are legislatures, but not necessarily all legislatures are parliaments.
--- Jura 07:09, 23 August 2017 (UTC)- About New Zealand: I'd suggest to keep , and to structure New Zealand Parliament (Q1520966) as follows:
- ⟨ subject ⟩ has part(s) (P527) ⟨ New Zealand Legislative Council (Q206569) ⟩
start time (P580) ⟨ 1-7-1841 ⟩
end time (P582) ⟨ 31-12-1951 ⟩
- This might also mean that New Zealand Parliament (Q1520966) was a unicameral legislature (Q37002670) from 1841 to 1854, then became a bicameral legislature (Q189445) that year, but on Jan 1, 1952 it returned to its unicameral legislature (Q37002670) status. (Please, consider that I took this data from en.wp and that something might be missing, but I think you get the point.) Sannita (mySociety) (talk) 15:09, 23 August 2017 (UTC)
- @Jura1: Having a separate item is interesting, and it's what I was already guessing we might need to do for Thailand. The 2014 Thai coup d'état (Q16914766) saw the National Assembly of Thailand (Q1368318) replaced with the National Legislative Assembly of Thailand (2014) (Q25352284), but it looks like they're about to reconstitute the original National Assembly again. National Assembly of Thailand (Q1368318) already has a dissolved, abolished or demolished date (P576) so adding a later inception (P571) seems like it could get confusing. Do you think the right way to handle that is a new item for the new Assembly? Or is there a better way to say that something was temporarily replaced by something else and then came back again?
- On legislature vs parliament, yep, I definitely agree that they're different, but having a parallel "bicameral parliament" vs "unicameral parliament" hierarchy to accompany unicameral legislature (Q37002670) and bicameral legislature (Q189445) doesn't seem like the way to go either. Are we happy that every unicameral parliament essentially needs two instance of (P31) records to say that they're both unicameral legislature (Q37002670) and parliament (Q35749), etc, or is there a better way to arrange the tree?--Oravrattas (talk) 15:56, 23 August 2017 (UTC)
- @Oravrattas: Hi. Something specific, but reading some feedback from people who have the bigger picture in mind can be interesting. Relating the unicamerality/bicamerality and the specific spanish case, I undid Oravrattas in Congress of Deputies (Q539149). There are actual Wikipedia articles about former parliament (Q35749) of Spain in Wikipedia Cortes de la Restauración (Q22284487) and Cortes republicanas (Q21496176) as well as the 2 (assembly (Q1752346)/advisory board (Q4686866)/pseudo-parliament (Q37001057)/governing body (Q895526)) of the two dictatorships: Asamblea Nacional Consultiva (Q8206153)) and Cortes franquistas (Q4893673). With their respective subclasses of Member of the Congress of Deputies (Q18171345) for the former Member of the Cortes republicanas (Q32979666)&member of the Congress of Deputies (Q32963246) (except the senator of Cortes de la Restauración (Q22284487), not modelled) and subclasses of politician (Q82955) for the later (Member of the Consultative National Assembly (Q24044772)&procurador en Cortes (Q27157707)). Data prior to 1876 is not modelled in Wikidata other than after the default standards as far as I know. Regards.--Asqueladd (talk) 16:58, 23 August 2017 (UTC)
- @Asqueladd: — oooh, adding all the historic versions of the Parliament like that is great! I'm really hoping we can do that for a lot more countries. (Perhaps we should create a separate sub-page here to discuss the issues around doing this?) There were no English labels for most of those, so I simply copied the Spanish ones across for now — if you happen to know a more commonly used English version, please do replace them. I'm finding it a little tricky to navigate around them all though — would it make sense to add replaces (P1365) and replaced by (P1366) to some of these? I don't really know enough of the history (or read enough Spanish) to be able to understand the nuances of being a pseudo-parliament (Q37001057) or the like. Or perhaps we could add dated entries to Q29#P194? --Oravrattas (talk) 19:18, 23 August 2017 (UTC)
- @Oravrattas: Hi. Something specific, but reading some feedback from people who have the bigger picture in mind can be interesting. Relating the unicamerality/bicamerality and the specific spanish case, I undid Oravrattas in Congress of Deputies (Q539149). There are actual Wikipedia articles about former parliament (Q35749) of Spain in Wikipedia Cortes de la Restauración (Q22284487) and Cortes republicanas (Q21496176) as well as the 2 (assembly (Q1752346)/advisory board (Q4686866)/pseudo-parliament (Q37001057)/governing body (Q895526)) of the two dictatorships: Asamblea Nacional Consultiva (Q8206153)) and Cortes franquistas (Q4893673). With their respective subclasses of Member of the Congress of Deputies (Q18171345) for the former Member of the Cortes republicanas (Q32979666)&member of the Congress of Deputies (Q32963246) (except the senator of Cortes de la Restauración (Q22284487), not modelled) and subclasses of politician (Q82955) for the later (Member of the Consultative National Assembly (Q24044772)&procurador en Cortes (Q27157707)). Data prior to 1876 is not modelled in Wikidata other than after the default standards as far as I know. Regards.--Asqueladd (talk) 16:58, 23 August 2017 (UTC)
- About New Zealand: I'd suggest to keep , and to structure New Zealand Parliament (Q1520966) as follows:
- Maybe 5: create a separate item for pre-1951.
@Oravrattas: I just noticed on my watchlist that this "position" was added to some people. I can understand that a term based mandate might apply to the UK, but do we have any reference for the existence of such a position? Somehow I share Innocent bystander and Louperivois's concerns expressed at Wikidata_talk:EveryPolitician#too_much_specific_items_for_legislature_members. We do have a qualifier to express the applicable parliamentary term when needed. In general, this is what should be used.
--- Jura 21:26, 23 August 2017 (UTC): Yes: https://www.riigikogu.ee/tutvustus-ja-ajalugu/riigikogu-ajalugu/x-riigikogu-koosseis/juhatus-ja-liikmed/ (this page does not appear in the English version of the site, but Google Translate does a good enough job, other than getting confused by the roman numeral)
- At a purely practical level, I created this to migrate all the existing cases of statements to instead (and to which we can then later add more data relating to dates, elections, etc). This is low-hanging fruit, but I am not aware of any existing tool that would enable us to do this automatically as .
- PetScan (Q23665536), in particular, would simply add the parliamentary term (P2937):10th Riigikogu (Q653547) qualifier to the first existing position held (P39):member of the Riigikogu (Q21100241) it could find for each person, rather than letting us create a new one. --Oravrattas (talk) 22:05, 23 August 2017 (UTC)
- One concern I did not mentioned in the last discussion, is that we in Sweden normally do not use such things as "Nth Parliament", because we do not know how and when to start counting. We have something in Arboga we call "the first Riksdag (1435)", but that is normally described as "not a real Riksdag".
- Neither do we like in frwiki mention Carl XVI Gustav as the Nth monarch of Sweden, since we do not know who the first was. -- Innocent bystander (talk) 06:16, 24 August 2017 (UTC)
- Yes, I guessed that might be why the official Riksdag site tends to simply refer to the terms using dates (2014-2018, etc). There are quite a few countries that do this the same way, especially if the same body has existed for more than around 100 years or so. Countries with much less of an unbroken historical run tend to number them, but not always. There also seem to be a few countries where the terms technically are numbered, and official records of the opening of parliament, or the like, will refer to them as such, but they're generally only referred to using the '2012-2016' form. I think we need to simply decide on a case by case basis whether it makes sense to break things up like this at all, and what terminology to use. My personal preference will generally be for whatever the primary source tends to use in the place where it provides its lists of members, or when detailing the historical membership data for an individual member. That's not always going to be available or appropriate, but it's a useful first approximation. --Oravrattas (talk) 08:32, 24 August 2017 (UTC)
- There may also be some doubts what a "Riksdag-term" is. The terms used to be counted, not between two elections, but between two yearly openings. In a few cases, the summer vacation for MP's were interrupted and an extra Riksdag was called, most of them in times of war. Two extra Riksdags were called 1905 because of dissolution of the union between Norway and Sweden (Q2264910). -- Innocent bystander (talk) 08:44, 24 August 2017 (UTC)
- In normal usage, a legislative term (Q15238777) is usually subdivided into multiple (though occasionally only one) legislative session (Q1812889). (In extreme cases I believe it's also possible that a term might contain zero legislative sessions.) Reality is often messier than theory, and obviously we need to be able to handle edge-cases, but in most countries the distinction is clear most of the time, at least in recent years. Historic political data is always going to raise a lot of interesting questions, particularly where concepts have mutated slowly over time, but if we can at least start making sure that current and recent data is always well-entered, that should give us a much better framework for then going back further in time. --Oravrattas (talk) 09:31, 24 August 2017 (UTC)
- There may also be some doubts what a "Riksdag-term" is. The terms used to be counted, not between two elections, but between two yearly openings. In a few cases, the summer vacation for MP's were interrupted and an extra Riksdag was called, most of them in times of war. Two extra Riksdags were called 1905 because of dissolution of the union between Norway and Sweden (Q2264910). -- Innocent bystander (talk) 08:44, 24 August 2017 (UTC)
- Yes, I guessed that might be why the official Riksdag site tends to simply refer to the terms using dates (2014-2018, etc). There are quite a few countries that do this the same way, especially if the same body has existed for more than around 100 years or so. Countries with much less of an unbroken historical run tend to number them, but not always. There also seem to be a few countries where the terms technically are numbered, and official records of the opening of parliament, or the like, will refer to them as such, but they're generally only referred to using the '2012-2016' form. I think we need to simply decide on a case by case basis whether it makes sense to break things up like this at all, and what terminology to use. My personal preference will generally be for whatever the primary source tends to use in the place where it provides its lists of members, or when detailing the historical membership data for an individual member. That's not always going to be available or appropriate, but it's a useful first approximation. --Oravrattas (talk) 08:32, 24 August 2017 (UTC)
- I don't think I found anything on the website (in English) that indicated that 10th Riigikogu is anything beyond the number of the legislative term. Thus I don't think the page can be used as a valid reference for the existence of a position with that name. You might want to read "structure of the Riigikogu". If it's merely a question about which tool to use, you might want to use Quickstatements instead of PetScan. I can help by converting the statements to the usual format.
--- Jura 03:47, 25 August 2017 (UTC)- On the purely practical matter: QuickStatements has exactly the same problem as PetScan — it can't add more than one P39 with the same target — it will simply add all the qualifiers to the first one it finds.
- On term-specific mandates, the Government site refers to people's legislative positions in this manner: https://www.valitsus.ee/en/minister-education-and-research-mailis-reps refers to the current Minister of Education as having been "a member of the 13th Riigikogu", "a member of the 12th Riigikogu", "a member of the 11th Riigikogu", and "a member of the 10th Riigikogu", for example. --Oravrattas (talk) 08:29, 25 August 2017 (UTC)
- Let's see how it goes how it goes with Quickstatements. It shouldn't be much of an issue as long as there are consecutive mandates only. I think the position is described at structure-of-riigikogu. People seem to be member of the parliament, not of a specific term. In some countries it's frequent that members are identified through their party affiliation so adding this with a qualifier is helpful, but we shouldn't use specific items for "Labour MP" or "Tory MP" in P39.
--- Jura 05:57, 3 September 2017 (UTC)- @Jura1: your assistance here is appreciated, but in this case it has been incredibly disruptive and destructive. As I pointed out before, QuickStatements cannot handle data like this, and your attempt to switch to plain 'Member of the Riigikogu' with qualifiers for the term has been disastrous, taking well formed data and collapsing it into a mess of meaningless and incorrect statements. With a single membership item now having the sum of all previous qualifiers, this means, for example, that queries such as "in which elections were Members of the 11th Riigikogu elected?", or "Which members left the 10th Riigikogu before the term ended?" give completely wrong results. Further, it is now impossible to use QuickStatements or PetScan to add more information about any of this data, and anyone who wants to even manually add richer data for each mandate will first have to undo your merges.
- The consensus across all three current large country-specific subprojects (UK, Germany, and France) for how to model legislative data is to have a separate P39 for each separate mandate. This is what we were also following here — work which you have now broken. Your underlying assumption that your updates "houldn't be much of an issue as long as there are consecutive mandates only" is not only demonstrably wrong about how QuickStatements works, but also reveals a seeming misunderstanding of how the Riigikogu works. As Estonia maintains a strict separation between the Executive and Legislature, anyone who has held Cabinet-level posts will, by necessity, have multiple, non-consecutive, legislative positions.
- On the more philosophical, rather than merely practical front, your repeated insistence that there is really no such thing as a "Member of the 12th Riigikogu", for example, is quite odd. Even if my own opinion from living in Estonia for 10 years counts for nothing in your eyes, I have also provided evidence from official government sources that this is entirely valid, and commonly used, terminology. Modelling the data in this way not only makes conceptual sense, but enables us to create much richer data through existing tooling than is currently possible with your alternative approach.
- I believed I had answered all objections to this way of modelling raised both here and at Wikidata talk:EveryPolitician#too much specific items for legislature members. A lack of further responses certainly doesn't mean that everyone is now in agreement, but I did expect it would mean that no-one would undo all the work to date without further discussion. If you have more issues or suggestions, please do raise them. I'm keen that we end up an approach that people are happy with, and which lets people add much more useful worldwide political information into Wikidata. But I would appreciate it if you avoided making such sweeping breaking changes without discussing what you're going to do first. In the meantime can you please revert these changes?
- --Oravrattas (talk) 07:59, 4 September 2017 (UTC)
- I think you responded to people's comments about replacing mandate items with parliamentary term items, but I don't see much support beyond EveryPolitician for this. I don't think EveryPolitician's undoing of a structure that has grown at Wikidata for years is helpful input at all. If EveryPolitician was paid by WMF to develop tools to improve processing, telling us that one or the other existing tool might not be ideal for EveryPolitician isn't really advancing us much. If you look at one or the other Estonian MP biography, you will notice that the element you insist on being key isn't even included.
--- Jura 08:11, 4 September 2017 (UTC)- I am not sure "this cannot be done with PetScan or QuickStatements" is a good reason to edit in a specific manner. In the early days QuickStatements could not add "Units" to quantity datatype-properties. But it was still wrong to add area (P2046) without unit except for some very unique items. -- Innocent bystander (talk) 08:50, 4 September 2017 (UTC)
- I don't think I understand this point. There is a difference between adding incorrect or meaningless statements that can never be automatically corrected later to the right values (e.g. quantities with no units), vs adding statements that are correct and well formed, e.g. "Enn Eesmaa (Q11857954) has been a member of the Riigikogu (Q33129158) since 30 March 2015, having been elected in 2015 Estonian parliamentary election (Q16412592) in Electoral District 5 (Hiiu-, Lääne- and Saaremaa) (Q2972997) as representative of Estonian Centre Party (Q928652). The only question here, as I understand it, is whether member of the Riigikogu (Q33129158) is a suitable subclass of member of the Riigikogu (Q21100241), or should be expressed instead with a qualifier. That the existing tooling does not make it possible to do the latter is certainly not, by itself, sufficient to say that the first should be preferred. But unless the first approach is obviously incorrect, the ability to use PetScan/QuickStatements for it certainly makes it much more appealing than not having the data at all, especially as it will be relatively straightforward to migrate those statements to a qualifier-based version later, should we decide later to go that route instead. --Oravrattas (talk) 10:19, 4 September 2017 (UTC)
- I am not sure "this cannot be done with PetScan or QuickStatements" is a good reason to edit in a specific manner. In the early days QuickStatements could not add "Units" to quantity datatype-properties. But it was still wrong to add area (P2046) without unit except for some very unique items. -- Innocent bystander (talk) 08:50, 4 September 2017 (UTC)
- I think you responded to people's comments about replacing mandate items with parliamentary term items, but I don't see much support beyond EveryPolitician for this. I don't think EveryPolitician's undoing of a structure that has grown at Wikidata for years is helpful input at all. If EveryPolitician was paid by WMF to develop tools to improve processing, telling us that one or the other existing tool might not be ideal for EveryPolitician isn't really advancing us much. If you look at one or the other Estonian MP biography, you will notice that the element you insist on being key isn't even included.
- Let's see how it goes how it goes with Quickstatements. It shouldn't be much of an issue as long as there are consecutive mandates only. I think the position is described at structure-of-riigikogu. People seem to be member of the parliament, not of a specific term. In some countries it's frequent that members are identified through their party affiliation so adding this with a qualifier is helpful, but we shouldn't use specific items for "Labour MP" or "Tory MP" in P39.
- In terms of this approach being used "beyond EveryPolitician": I first proposed this in Wikidata:WikiProject British Politicians, which is not part of the EveryPolitician project, and it was tested initially with data for Scotland, where it very quickly enabled us to have significantly more, and significantly higher quality data, then existed before. As a result, this was then rolled out for the UK (back to 1997 initially), and Northern Ireland (to the formation of the Assembly).
- I'm not sure what you mean by "undoing of a structure that has grown at Wikidata for years". This is not undoing any existing structure — it is supplementing it. The goal is to move beyond the current situation where the vast majority (over 80% at last count) of P39 statements have no qualifiers at all, and merely record that someone held the position at some unspecified time. We want to move to a situation where we have much, much richer information on all the legislative positions people have held over time.
- On your claim that "telling us that one or the other existing tool might not be ideal for EveryPolitician isn't really advancing us much". That's not what's happening here at all. Not only have we been discussing with Magnus how to make both these tools better, we're also getting on with helping people add much more detailed information in the meantime. In the Estonian case we had already added rich data for the current Riigikogu, but had paused the historic inputs pending this discussion, having done nothing beyond migrate the existing P463 records (which I don't believe anyone is advocating the use of) to P39 records instead. In the middle of that, you have used QuickStatements to migrate all the "one P39 per mandate" records into something more akin "one P39 for a multi-term membership of the legislature" — a model that cannot express all the required data, and which all the major parliamentary data subprojects here have rejected.
- As for "If you look at one or the other Estonian MP biography, you will notice that the element you insist on being key isn't even included". I have already showed you an example from valitsus.ee, but see also http://www.just.ee/en "From 2014–2015, Urmas Reinsalu was a member of the 12th Riigikogu" (if this isn't a concept, why not simply say "was a member of the Riigikogu"?), or https://www.siseministeerium.ee/en/organisation/minister-interior-hanno-pevkur "From 2015–2016, Andres Anvelt was a member of the 13th Riigikogu". On the official Riigikogu website, the history of all MPs are also given on such a term-by-term basis. See, for example, https://www.riigikogu.ee/en/parliament-of-estonia/composition/members-riigikogu/saadik/07132cc4-5afc-4edb-9537-7c721ec39b2d/Enn-Eesmaa Membership which (under the "CV" section) draws a clear distinction between a continuous position ("Deputy Chairman of the Foreign Affairs Committee 2007–2016") and being a member of the "11th, 12th and 13th Riigikogu" over that same period.
- --Oravrattas (talk) 10:05, 4 September 2017 (UTC)
- The current structure has been developed over the years and its qualifiers should generally allow to express the expression you seek. Obviously, if you attempt to include dozens of data-points per term in an item, eventually you wont be able to do that. You will need to create items such as Q27809653.
- In general, the question is not if we can find a page that mentions the term, but if the position is actually defined that way. (As mentioned before) looking at structure-of-riigikogu, this doesn't seem to be the case. E.g. et:Tiit Tammsaar (a somewhat random article) doesn't even bother mentioning it.
- As these people aren't British politicians either, you can't just convert an existing structure to one for British politicians, merely because it may suit EveryPolitician. It seems highly dubious if WMF funds should be used to pay for such edits, especially when it seems to have been granted to pay for development of import tools.
--- Jura 14:10, 4 September 2017 (UTC)- To start with: the work that I am doing on Estonian politicians is not funded by WMF. I have been editing political information on Wikidata, and Estonian political data in specific, for several years now, and will continue to do so entirely independently of the WMF grant. Grant funds are not being used for editing.
- I also don't follow what you're saying about Tiit Tammsaar. The Estonian Wikipage does split his memberships per term like that ("VIII Riigikogu liikmed | IX Riigikogu liikmed | X Riigikogu liikmed | XII Riigikogu liikmed"), and I have already shown you the official parliament and government sites both also do likewise. I do not know why you keep insisting that this is completely invalid. I'm also very confused by the example you've given of presidency of Donald Trump (Q27809653). If it's valid to create that item, what would the equivalent be here? Is your issue that member of the Riigikogu (Q33129158) shouldn't exist at all, or more that you don't think that it's a valid instance of a position (Q4164871)? Would modelling member of the Riigikogu (Q33129158) in a different way be more acceptable to you?
- Can you provide a reference to what you mean by "the existing structure"? I'd like to make sure we're not talking past each other here. The vast majority of P39 items for memberships of legislatures have essentially no structure at all at the minute, as they don't actually include any extra data. But where there has been any discussion around how to model richer legislative information — e.g. at https://www.wikidata.org/wiki/Wikidata:WikiProject_British_Politicians#Properties, https://www.wikidata.org/wiki/Wikidata:WikiProject_France/Politicians/Properties or https://www.wikidata.org/wiki/Wikidata:WikiProject_Heads_of_state_and_government/Germany, the decision has been to have one P39 for each mandate. (In the German case, even each Parliamentary Group is required to have a new Item in each separate term.) Collapsing distinct memberships to a single P39, as you have done this weekend, does not seem to be at all in keeping with any existing structure that I can see, and, as I gave examples of above, now means that queries return incorrect results.
- Your claim that "you can't just convert an existing structure to one for British politicians, merely because it may suit EveryPolitician" is also unfounded. I'm not doing this because it suits EveryPolitician — I'm doing it because it's the best way to add the information about Estonian politicians. Note also that the "existing structure" (i.e. using P463 for this rather than P39) was one that I put in place two years ago (e.g. https://www.wikidata.org/w/index.php?title=Special:Contributions/Oravrattas&offset=20151015190310&limit=500&target=Oravrattas), and is no longer in use for any other national legislature that I know of (as the UK data, which was previously modelling in this manner, has now been migrated to term-specific membership items, just like I have been doing here.) Having lived in both Estonia and the UK for long periods, and having both personal and professional involvement with political data, I am very aware of both the similarities and differences between the two systems, and do not see how this approach is OK for the UK, but not for Estonia.
- --Oravrattas (talk) 15:16, 4 September 2017 (UTC)
- The text of the article at et:Tiit Tammsaar doesn't mention it (last edit: 2015). Not sure about having Q37822184: there may be some use for it, but it's just not a suitable value for position held (P39). --- Jura 06:59, 5 September 2017 (UTC)
- @Jura1: This is getting farcical. I have demonstrated that the Estonian Parliament website uses this terminology; the Estonian Government website uses this terminology; the Wikipedia article you selected also uses it … but you continue to insist that these are all somehow erroneous because a given body of text doesn't include it? What is your suggestion, from the choices below (or an alternative if none of these suit) for how this data should be modelled for the Riigikogu? This weekend you used QuickStatements to migrate most of the legislative data to approach #2. Do you believe that that is the correct modelling (despite the examples I gave of queries that now return the wrong answers)? Or do you accept that you have broken the data, and it should now be fixed? --Oravrattas (talk) 07:58, 5 September 2017 (UTC)
- I might be mistaken, but what you mention from et:Tiit Tammsaar seems to be the categories, not the article text. The text just mentions Riigikogu. Besides, you don't seem to disagree with the description of the position provided by structure-of-riigikogu.
--- Jura 05:35, 6 September 2017 (UTC)- @Jura1: Here are 25 others, where their Wikipedia page describes the people in exactly the formulation you disagree with within the first paragraph (usually the first sentence): et:Kalev Kallemets, et:Martin Kukk, et:Siret Kotka, et:Olga Ivanova, et:Priit Toobal, et:Andrei Novikov, et:Erki Savisaar, et:Urbo Vaarmann, et:Tanel Talve, et: Aivar Kokk, et: Madis Milling, et: Peeter Võsa, et: Kalev Lillo, et: Tatjana Jaanson, et:Kalmer Lain, et:Innar Mäesalu, et:Heidy Purga, et:Neeme Suur, et:Annely Akkermann, et:Juku-Kalle Raid, et:Marko Šorin, et: Heljo Pikhof, et: Neinar Seli, et:Edgar Spriit, et:Eldur Parder.
- Whether you agree that this is how it should be or not, you have been overwhelmingly been shown that this is how these memberships are described in Estonia. As you seem to be unwilling to fix the problems you have created by recklessly collapsing the separate P39s for people into single badly-formed entries (or even engage with any of the questions about this), it looks like other people are going to have to clean up your mess instead. I trust you won't make any further sweeping changes to the data without discussing it first. --Oravrattas (talk) 07:38, 16 September 2017 (UTC)
- I might be mistaken, but what you mention from et:Tiit Tammsaar seems to be the categories, not the article text. The text just mentions Riigikogu. Besides, you don't seem to disagree with the description of the position provided by structure-of-riigikogu.
- @Jura1: This is getting farcical. I have demonstrated that the Estonian Parliament website uses this terminology; the Estonian Government website uses this terminology; the Wikipedia article you selected also uses it … but you continue to insist that these are all somehow erroneous because a given body of text doesn't include it? What is your suggestion, from the choices below (or an alternative if none of these suit) for how this data should be modelled for the Riigikogu? This weekend you used QuickStatements to migrate most of the legislative data to approach #2. Do you believe that that is the correct modelling (despite the examples I gave of queries that now return the wrong answers)? Or do you accept that you have broken the data, and it should now be fixed? --Oravrattas (talk) 07:58, 5 September 2017 (UTC)
- The text of the article at et:Tiit Tammsaar doesn't mention it (last edit: 2015). Not sure about having Q37822184: there may be some use for it, but it's just not a suitable value for position held (P39). --- Jura 06:59, 5 September 2017 (UTC)
- My silence doesn't mean I agree with this approach. It means I'm busy outside Wikidata and I can't follow all the talk pages of each project discussing this issue. I continue to think that third-level tools must not justify the editorial choices of Wikidata. But as those who are absent are always wrong, I have decided to run my bot outside of the scope of your initiative: it will not add or edit MP statements in the UK (national and subnational level) in order to avoid further involving in a debate that is still geographically limited, and to prevent the degradation of the data. Louperivois (talk) 17:03, 4 September 2017 (UTC)
- @Louperivois: thanks for the comments. To be clear, is your objection to having separate P39 memberships for each term-specific mandate? Or simply to those membership items being "Member of the Nth Assembly of Placeistan"? For example if someone was a member of the 3rd, 4th, and 5th Assemblies, from 2001-2014, we currently have several possible approaches:
- a single P39 of "Member of the Assembly of Placeistan", with a start time (P580) of 2001 and end time (P582) of 2014 (and no parliamentary term (P2937) qualifier)
- a single P39 of "Member of the Assembly of Placeistan", with a start time (P580) of 2001 and end time (P582) of 2014 (and three parliamentary term (P2937) qualifiers, one for each term)
- three P39s of "Member of the Assembly of Placeistan", each representing a single-term mandate (and the relevant start time (P580), end time (P582), and parliamentary term (P2937) on each)
- three P39s, one for each of "Member of the 3rd Assembly of Placeistan", "Member of the 4th Assembly…", and "Member of the 5th Assembly…"
- I'm not sure which of these your bot can handle, and which you think are unacceptable, acceptable, or preferable.
- It would be useful to get other people's input specifically on this question too. My view is that #2 is unacceptable; #1 is acceptable only if we don't record any other information about the membership, but once we start recording anything that will vary per term (e.g. elected in (P2715)), we need to move to either #3 or #4. I believe that even aside from third-level tools, #4 has a slight edge as it allows us to store other information about the position during that period, and is slightly easier to query, but those are relatively minor preferences. However, the fact that lots of existing tools (e.g. PetScan, QuickStatements, and Listeria) currently work well with #4 and not at all, or much less well, with #3, without any corresponding downsides, make it the clear choice to me — especially when we also have external partners who want to actively want to add rich political data, and otherwise cannot do so without adding thousands of data records individually by hand or by writing a bot (which is a significantly high barrier).
- Alternatively, it's entirely possible that the lack of agreement on any of these means we simply haven't found the even better approach…
- --Oravrattas (talk) 21:07, 4 September 2017 (UTC)
- @Louperivois: thanks for the comments. To be clear, is your objection to having separate P39 memberships for each term-specific mandate? Or simply to those membership items being "Member of the Nth Assembly of Placeistan"? For example if someone was a member of the 3rd, 4th, and 5th Assemblies, from 2001-2014, we currently have several possible approaches:
- My silence doesn't mean I agree with this approach. It means I'm busy outside Wikidata and I can't follow all the talk pages of each project discussing this issue. I continue to think that third-level tools must not justify the editorial choices of Wikidata. But as those who are absent are always wrong, I have decided to run my bot outside of the scope of your initiative: it will not add or edit MP statements in the UK (national and subnational level) in order to avoid further involving in a debate that is still geographically limited, and to prevent the degradation of the data. Louperivois (talk) 17:03, 4 September 2017 (UTC)
- As mentioned on the project page, most countries have specific identifiers which include constraints that require the use of specific P39 statements, e.g. member of the German Bundestag (Q1939555) on biography at the Bundestag of Germany URL (P1713), member of the Chamber of Deputies of the Italian Republic (Q18558478) (among others) on Italian Chamber of Deputies dati ID (P1341). Over the years, many of us add such statements. Obviously, these shouldn't replaced with idiosyncratic items. Terms can be added with the qualifier created for that purpose: parliamentary term (P2937). Other qualifiers that might be worth adding: oath of office date (P1734), statement is subject of (P805).
--- Jura 06:59, 5 September 2017 (UTC)- I think we're still a long way away from most countries having these. The list of suitable targets on the main Project page only lists 15 countries. (If you know of more it would be good to extend that list.) Creating similar properties for a lot more countries would be very useful, though. I haven't explored the new constraints system in much depth, yet. For the UK, how can we say that an item using P1996 (P1996) must be an instance of a subclass of Member of Parliament (Q16707842)? --Oravrattas (talk) 08:06, 5 September 2017 (UTC)
- If the constraints are new to you, you might want to look into them. There is a general help available at Help:Property_constraints_portal.
--- Jura 05:35, 6 September 2017 (UTC)- The new constraints system apparently can't handle subclasses - see Help talk:Property constraints portal - "I don’t think that kind of combination of type constraint (Q21503250) and item requires statement constraint (Q21503247) is currently possible.". So there's (currently) limits to what we can use a property constraint to define, unfortunately.
- In general, though, one problem with saying "constraints on the identifier define what should be done with the item" is that it only works if everyone uses the same identifier. British MPs might have any combination of P1996 (P1996), Hansard (2006–March 2016) ID (P2170), Hansard (1803–2005) ID (P2015), History of Parliament ID (P1614), or none of the above (for some historic MPs). That implies having to keep four sets of constraints in sync to create a data model. It's also not great if we simply don't have an appropriate property for that country - eg I've done a lot of work on Scottish MSPs and there's no useful identifier property there. I think property-based constraints are a good start, but it shouldn't stop us creating a more sophisticated data model if that's appropriate for a given group. Andrew Gray (talk) 16:37, 6 September 2017 (UTC)
- The constraints aren't new. It's just the documentation and some the implementation has been improved. As British MPs have their own project, you might want to discuss them there.
--- Jura 09:06, 7 September 2017 (UTC)
- The constraints aren't new. It's just the documentation and some the implementation has been improved. As British MPs have their own project, you might want to discuss them there.
- If the constraints are new to you, you might want to look into them. There is a general help available at Help:Property_constraints_portal.
- I think we're still a long way away from most countries having these. The list of suitable targets on the main Project page only lists 15 countries. (If you know of more it would be good to extend that list.) Creating similar properties for a lot more countries would be very useful, though. I haven't explored the new constraints system in much depth, yet. For the UK, how can we say that an item using P1996 (P1996) must be an instance of a subclass of Member of Parliament (Q16707842)? --Oravrattas (talk) 08:06, 5 September 2017 (UTC)
- Hello. Just a quick message to fully support Oravrattas about one statement by mandate. It was discussed (in French, but I can translate if requested) on French politicians project. — Envlh (talk) 08:18, 5 September 2017 (UTC)
- More or less the same here: the option 3 dealing with a statement by mandate (if possible, data does not always exist obviously) can be a bit complex but in the long run, I think this is the best way to go. Cdlt, VIGNERON (talk) 20:56, 5 September 2017 (UTC)
- @VIGNERON, Envlh: thanks for the input. To be clear do you think option 4 above (i.e. the same as option 3, but with term-specific items rather than only as a qualifier) is: (a) never suitable; (b) perhaps suitable in some countries, though not all; (c) a potentially useful interim approach to work around tooling problems, though it would be good to migrate those to option 3 later; (d) some other option…? --Oravrattas (talk) 22:13, 5 September 2017 (UTC)
- @Oravrattas: not sure... I lean more toward c) (maybe b), but I don't know what is the state of available data in each and every country ). One point is not clear to me : in option 4, is there qualifiers or not? (is it a solution for when we don't know and can't put qualifier? or is it with qualifier and purposely a bit redundant for consistency checking, or other usefulness). Anyhow, even if the term-specific items or not used in position held (P39) (or other properties) on politicians, the items still seems useful to me (for structural purposes at least). Cdlt, VIGNERON (talk) 10:45, 6 September 2017 (UTC)
- Ah, yes, the term-specific membership items in option 4 are also used with qualifiers, just like in option 3. For an example, see Q268196#P39. Sorry for not making that clearer. --Oravrattas (talk) 20:08, 6 September 2017 (UTC)
- @Oravrattas: not sure... I lean more toward c) (maybe b), but I don't know what is the state of available data in each and every country ). One point is not clear to me : in option 4, is there qualifiers or not? (is it a solution for when we don't know and can't put qualifier? or is it with qualifier and purposely a bit redundant for consistency checking, or other usefulness). Anyhow, even if the term-specific items or not used in position held (P39) (or other properties) on politicians, the items still seems useful to me (for structural purposes at least). Cdlt, VIGNERON (talk) 10:45, 6 September 2017 (UTC)
- @VIGNERON, Envlh: thanks for the input. To be clear do you think option 4 above (i.e. the same as option 3, but with term-specific items rather than only as a qualifier) is: (a) never suitable; (b) perhaps suitable in some countries, though not all; (c) a potentially useful interim approach to work around tooling problems, though it would be good to migrate those to option 3 later; (d) some other option…? --Oravrattas (talk) 22:13, 5 September 2017 (UTC)
- More or less the same here: the option 3 dealing with a statement by mandate (if possible, data does not always exist obviously) can be a bit complex but in the long run, I think this is the best way to go. Cdlt, VIGNERON (talk) 20:56, 5 September 2017 (UTC)
- It seems we mostly agree on not using member of the Riigikogu (Q37822184). I'm fine with using the French model for Estonia if that suits Oravrattas.
--- Jura 05:35, 6 September 2017 (UTC)
- It seems we mostly agree on not using member of the Riigikogu (Q37822184). I'm fine with using the French model for Estonia if that suits Oravrattas.
See also parallel discussion at EveryPolitician |
- Seems I should have checked on this to make sure that it was followed-up on. Here we go: Wikidata:Requests_for_deletions#Q37822184_etc..
--- Jura 17:06, 27 July 2018 (UTC)
I agree that member of the Riigikogu (Q37822184), member of the Riigikogu (Q61976148) etc. are pseudo-classes and no such positions actually exist. Above you are just misinterpreting Wikipedia category titles and other similar excerpts, non of which explicitly suggest that new parliament member position is created for each parliamentary term. Reasoning above is pretty much the same as is to claim that "42nd president of the United States" and "43rd president of the United States" are different positions if some source uses such phrases, while actually next president is elected to the same position that was hold by previous president. Member of the Riigikogu position is defined by several legislative acts such as Status of Members of the Riigikogu Act. Everything said about this position in legislative acts applies to it regardless of parliamentary term. In legislative acts there is no indication that new position is created when new parliament members are elected and new parliamentary term begins. Maybe in case of some other country there is reason to think that new parliament member position is created after election, but there's no evidence that this would be the case in Estonia. I think misleading pseudo-classes shouldn't be used only in order to make some queries a bit simpler. Classes on Wikidata generally are those that are actually used outside Wikidata. Hence member of the Riigikogu (Q21100241) with parliamentary term (P2937) qualifier would make much more sense. 2001:7D0:81DA:F780:989C:883D:3886:555 09:28, 13 September 2021 (UTC)
What will EveryPolitician provide us with?
[edit]It seems that WMF is paying EveryPolitician to develop some tools which also cover parliaments (through a grant). We currently get complaints from EveryPolitician staff that this or that structure wouldn't be working with existing tools (not some that are being developed). A few projects participants seem to be paid employees of that organization and/or affiliated with it. Specifically, what would Wikidata get in the field of this project and who is doing it?
--- Jura 06:59, 5 September 2017 (UTC)
Bad use of P:P361
[edit]Members of Parliament are currently linked to their Parliament using part of (P361), for example Member of Parliament in the Parliament of England (Q18018860)part of (P361)Member of Parliament in the Parliament of England (Q18018860) That is bad semantics. Per property documentation P361 is supposed to be transitive and equivalent to http://purl.obolibrary.org/obo/BFO_0000050 which is incompatible with that. I think the correct property is member of (P463). --Zolo (talk) 07:36, 7 September 2017 (UTC)
- Your sample mentions Member of Parliament in the Parliament of England (Q18018860) twice. Supposedly you meant part of (P361) House of Commons of England (Q5914849). I think you could easily add has part(s) (P527) on House of Commons of England (Q5914849). Since this was first used has part(s) of the class (P2670) was created. There was some discussion if we should replace P527 with P2670. What do you think?
--- Jura 09:07, 7 September 2017 (UTC)- Yes, I meant House of Commons of England (Q5914849) sorry. Actually, this is not primarily about class vs instances, it is about semantics and transitivity. part of (P361) is supposed to be a transitive property. The example usecase from the doc is "brain: part of the body". So we would have a "brain of MP x"-> part of MP X -> part of House of Commons. That does not make sense, a parliament is not made of body parts. The issue would be similar with has part(s) of the class (P2670) --Zolo (talk) 14:08, 7 September 2017 (UTC)
Members of committees
[edit]Most parliaments have committees (the EP has for example European Parliament Committee on Internal Market and Consumer Protection (Q780106), the Swedish parliament has Committee on Finance (Q5153055)). Should a member of such a committee have member of (P463) or should it be position held (P39) with a value with a new item like 'member of Committee on Finance'? Ainali (talk) 18:52, 17 November 2018 (UTC)
- I favour the second of these. IMO these are positions; the committees are not organizations or clubs. --Tagishsimon (talk) 12:20, 18 November 2018 (UTC)
- Agree - P39 is best for formal committees like this. Andrew Gray (talk) 12:45, 20 November 2018 (UTC)
- Thanks for your feedback. I'll do a test and link here. Ainali (talk) 18:34, 20 November 2018 (UTC)
- Alright, here is a test edit. I think it looks good, do you still agree? Ainali (talk) 20:04, 20 November 2018 (UTC)
- Looks good to me, although one of those has two start dates: I'm assuming 24 September 2018 should be a end time (P582) instead? --Oravrattas (talk) 08:52, 21 November 2018 (UTC)
- Yes, Fixed Ainali (talk) 20:59, 27 November 2018 (UTC)
- Looks good to me, although one of those has two start dates: I'm assuming 24 September 2018 should be a end time (P582) instead? --Oravrattas (talk) 08:52, 21 November 2018 (UTC)
- Alright, here is a test edit. I think it looks good, do you still agree? Ainali (talk) 20:04, 20 November 2018 (UTC)
- Thanks for your feedback. I'll do a test and link here. Ainali (talk) 18:34, 20 November 2018 (UTC)