Wikidata talk:WikiProject every politician

Active discussions

Wikidata query: occupations of parlamentariansEdit

You might like this:

BubbleCharts! --Atlasowa (talk) 08:57, 8 December 2016 (UTC)

  • Not really. At some level, I suppose it could be good to hear that people assume coverage of Wikidata is that good, but on some other level it really isn't. There might just be gaps in professions other than journalism.
    --- Jura 09:01, 8 December 2016 (UTC)


For people, reference URL (P854), start time (P580) and end time (P582) must be mandatory, without this, it is useless. --ValterVB (talk) 20:30, 14 March 2017 (UTC)

@ValterVB: I'm presuming you're talking about the position held (P39) membership information here? --Oravrattas (talk) 09:40, 15 March 2017 (UTC)
Sorry, I read here in section Mandatory: "preferably with a reference URL (P854) (ideally to the official site), and with modifiers (where known/appropriate): parliamentary term (P2937), start time (P580), end time (P582), electoral district (P768), elected in (P2715)". It's need to know (mandatorily) the source of the data and the period of validity of the position held (P39). --ValterVB (talk) 12:32, 15 March 2017 (UTC)
We can certainly firm up the wording to explain that when someone is adding a P39 membership, they MUST add a reference, but the much bigger problem is that out of the 375,000 current uses of P39, only 71,000 have references at all, with over 300,000 without. --Oravrattas (talk) 15:23, 15 March 2017 (UTC)
In incipit of Wikidata:EveryPolitician I read "This is a tracking page for an ongoing project to provide comprehensive coverage of politicians in Wikidata" Can yo u explain what is the source? --ValterVB (talk) 20:40, 15 March 2017 (UTC)
I'm not sure I understand the question. There isn't one source for this — there are thousands of sources: official parliamentary sites, monitoring organisations, press articles, elections results sites, etc etc. --Oravrattas (talk) 08:37, 16 March 2017 (UTC)
Perfect, it's exactly what need. What I ask is: this source will be added to the data? Because add other a lot of other data without source isn't acceptable (wikipedia is not a source).. --ValterVB (talk) 17:44, 16 March 2017 (UTC)
At the moment a lot of these are imported from Wikipedia and so the dates are not yet in place but we do want them to be there. I agree it's not very useful without them. Andrew Gray (talk) 14:01, 16 March 2017 (UTC)


m:Grants:Project/EveryPolitician mentions there is a shortlist of countries somewhere on this page but I didn't see one. Anyway, in case you work on Italy, do you know about ? The have arrangements with the Italian Parliament and help publish a lot of open data about parliamentary activities. --Nemo 15:30, 10 June 2017 (UTC)

@Nemo_bis: — Yes, we know Openpolis well, and they are currently one of our main sources of data for Italy. The Italian Parliament itself is already one of the best in the world for publishing open data (and even has a SPARQL interface to it) — --Oravrattas (talk) 06:04, 11 June 2017 (UTC)
Indeed, Openpolis have worked many years on this. I'm glad you build on their work. --Nemo 07:06, 11 June 2017 (UTC)

Useful QueryEdit

This is a useful SPARQL query showing how many people hold a given position by date. (The link goes to one for Israeli parliamentarians, but it can work for any legislature by replacing the relevant Q-ids.) Can be useful for finding gaps and inconsistencies in the data. --Yair rand (talk) 21:18, 3 July 2017 (UTC)

Project UpdatesEdit

Project updates will henceforth feature on: the timeline page for the associated grant. --Lucyfediachambers (talk) 13:37, 20 July 2017 (UTC)

Member of the Swedish parlamentEdit

Many of the member of the Swedish Riksdag (Q10655178) should probably be migrated to "Member of the Swedish first/second chamber". Many of these predates the single chamber parlament which was established in 1971. Bertil Johansson (Q4895688) for example was member of the Second chamber 1965-1970 and the Riksdag from 1971. -- Innocent bystander (talk) 18:21, 5 August 2017 (UTC)

@Innocent bystander: we could definitely help with getting the data a lot better for all the members since 1976 (using the official API, which makes structured data available with no licensing restrictions), ensuring everyone has dates of service, party groups, electoral districts etc. Is that something you'd be willing to help out with, or do you know other people we should talk to about that? 1971-1976 seems a little bit harder, and going back even further than that looks like it would get much harder very quickly — or do you know any good sources for older data? --Oravrattas (talk) 05:32, 6 August 2017 (UTC)
There are good sources for the Second chamber many decades before 1976, but I seriously doubt that they are machine-readable. Also the First chamber has sources, but they are trickier, since you have to be at least boyscout to interpret them. (They were not elected in general elections, but by a sort of electoral collage, containing of councillors in some citys and county counties.) The users with most experience of those are IP's. I definitely do not have the time, but maybe André Costa or Ainali know SomebodyTM. -- Innocent bystander (talk) 06:47, 6 August 2017 (UTC)
This is now done :) Popperipopp (talk) 09:13, 15 June 2020 (UTC)

Lists and P360Edit

For lists of parliamentarians, the preferred syntax for property and qualifiers should probably be

list of MPs elected in the United Kingdom general election, 2010 (Q4158639) is a list of (P360) human (Q5)
qualifier: position held (P39) Member of Parliament (Q16707842)
qualifier: parliamentary term (P2937) 55th United Kingdom Parliament (Q21084472)

rather than

list of MPs elected in the United Kingdom general election, 2010 (Q4158639) is a list of (P360) Member of Parliament (Q16707842)
qualifier: parliamentary term (P2937) 55th United Kingdom Parliament (Q21084472)

The first form (with the position held put into a qualifier statement) gives the additional information that each member of the list will be instance of (P31) human (Q5), and will be related to Member of Parliament (Q16707842) by position held (P39)

This makes it more possible to auto-generate or check such lists (as e.g. Reasonator tries to do, link ), and also to analyse them at scale. Jheald (talk) 21:02, 14 August 2017 (UTC)

Unfortuately there's a problem -- there's no understood way to signal in a is a list of (P360) that membership relies on a qualifier value, rather than a property value. (The P360 statement is assumed to give a list of properties and their required values, with no provision for qualifiers and values).
There's no easy way I can see around this. But there is a different model of semaphore that's been suggested in connection with category combines topics (P971), so maybe something like that could be a possibility. Jheald (talk) 07:53, 19 August 2017 (UTC)

too much specific items for legislature membersEdit

I noticed that very specific position items related to each legislature of one assembly have been created and massively used in position held (P39), for instance Member of the 52nd Parliament of the United Kingdom (Q36634044). User:Oravrattas is pursuing the creation of similar items for each imaginable parliament. In my opinion, these items are too much specific and could be avoided simply by using parliamentary term (P2937) as a qualifier of Member of Parliament (Q16707842). It is incorrect that being a member of a given parliamentary term (P2937) is a semantically different position that another parliamentary term (P2937). Moreover, several problems are emerging right now: The translation of these items allowing their use (for instance in infoboxes) will always be unsatisfactory; and the mass import of these claims is redundant with more general claims related to the same charge. I therefore ask User:Oravrattas to discuss here before resuming his work. Louperivois (talk) 22:34, 20 August 2017 (UTC)

@Louperivois: the previous discussion for this is at Wikidata:EveryPolitician/Proposal:Term Membership Items. Perhaps it would be best for us to continue the conversation over there? --Oravrattas (talk) 06:29, 21 August 2017 (UTC)
  • It might be worth doing a specific project for parliaments. The "how to import from the EveryPolician website"-project seems unsuitable to coordinate with the Wikidata community. For the UK, there is already a specific project.
    --- Jura 08:25, 21 August 2017 (UTC)
    • Agreed; Sannita (MySociety) (talkcontribslogs) was discussing a possible WikiProject:Politics at Wikimania last week. We currently know of country-specific projects around politics for the UK, France, Germany, and Finland. If anyone knows of any others, pointers would be very useful. The specific discussion on per-term "member of" positions arose out of the UK case, and is where we have been trialling it — see discussion at Wikidata talk:WikiProject British Politicians/Archive 1#One_P39_per_parliament_versus_one_P39_per_held_seat and following.--Oravrattas (talk) 10:48, 21 August 2017 (UTC)
      • Most countries have specific properties which generally include constraints defining the datamodel for the relevant MPs, at least what's needed beyond P39.
        --- Jura 11:00, 21 August 2017 (UTC)
        • Here is one for parliaments: Wikidata:WikiProject Parliaments‎
          --- Jura 11:47, 21 August 2017 (UTC)
          • I was about to say "wow, how on earth did I miss that!" but I see you've just put it together. Thanks, this is useful :-) Andrew Gray (talk) 12:24, 21 August 2017 (UTC)
@Louperivois:. I was originally quite uncomfortable with the idea of term-based membership items for the UK, but after Oravrattas and I did some work on it, it became clear it was the most practical option for the time being. Because we can't assign start-end dates to specific qualifiers, we're always going to need multiple P39s for many politicians (sat for a different seat? elected more than once? changed party?). We tried having lots of position held (P39):Member of Parliament (Q16707842) with qualifiers, but a lot of external tools don't work very nicely with this approach - eg QuickStatements will try and put all the qualifiers for every entry onto a single entry.
What I've done at Wikidata:WikiProject British Politicians is try and formalise a "level 1 to level 4" approach here - the most basic, a single position held (P39):Member of Parliament (Q16707842) with no qualifiers is level 1; and the "full" system with Member of the 52nd Parliament of the United Kingdom (Q36634044) and qualifiers is the most detailed, at level 4. A well-constructed query should in theory work for all levels, although of course you'll only get the full information once the qualifiers are added. I think this might be a good approach to use when thinking about this project - most countries will sit at Level 1 with perhaps Level 2 for current politicians (so we can identify them as current), and that's fine, if we don't have anyone willing and able to improve the data. But where we have the data and someone able to do the work (eg the UK), we can push on to everyone being Level 2, Level 3, or Level 4. Andrew Gray (talk) 12:24, 21 August 2017 (UTC)
The model with "123rd Parliament" works poorly for some chambers. The Swedish first chamber had elections every year for 1/8 of the seats. (They were normally elected for eight years.) It is even possible that there were more than one election some years. -- Innocent bystander (talk) 12:30, 21 August 2017 (UTC)
Yeah, I don't think it'll work in all cases - in the UK I don't currently think we'll use it for the House of Lords, for example. But for the ones where a legislative term is a coherent concept, it seems to makes sense.
I've been thinking about the translation issues mentioned earlier and I agree that's a concern. I'll try to think of a way to make sure this is handled systematically. Andrew Gray (talk) 12:39, 21 August 2017 (UTC)
I am used to some complex constructions of templates on svwiki, and from that point of view I think there are no large problems, as long as there is a simple and straight relation between "Member of the 52nd Parliament of the United Kingdom" and "member of the House of Commons of the United Kingdom". The subclass of (P279)-relation is technically correct here, but is it enough simple and enough straight everywhere? It also have to be simple and straight up to next level, "member of parliament", which we can imagine everybody have a translation of. If there are too many levels between "member of 53rd Parliament" and "member of parliament" the template soon becomes to complex to be displayed on Wikipedia. -- Innocent bystander (talk) 12:55, 21 August 2017 (UTC)
At the moment the hierarchy is a bit of a mess: --Oravrattas (talk) 13:17, 21 August 2017 (UTC)
I guess it's just UK politics that is complicated to understand for other countries. In most countries, people just get elected and then sit forever ..
--- Jura 12:47, 21 August 2017 (UTC)
Not here  . On local level, it is not unusual that people got tired after half a term and leave the seat to somebody else. And on national level, they have parental leave, get elected for EU-Parliament etc etc. -- Innocent bystander (talk) 12:55, 21 August 2017 (UTC)
Well, forever is relative .. Nothing that couldn't really be solved with a single start and end date for consecutive time in office ..
--- Jura 13:08, 21 August 2017 (UTC)
Almost all of the primary chambers of national legislatures (with a handful of notable exceptions) nowadays follow a fairly standard term-based model, where all (or almost all) the seats are re-elected in a single general election, members are often commonly referred to as having been a member during the Nth Period, etc. In most cases official parliamentary websites that actually include historical data include it in that format (rather than as a single start date to single end date model). It is also very common on Wikipedias for members to be in categories such as "Member of the 8th Legislature of Country" (and thus very suitable for quick imports to suitable P39 targets). Using Andrew's ranking system, Level 2 data is certainly a lot better than anything before it, but it starts to break down quite quickly when you actually want to have comprehensive data, including tying each mandate to a specific election, tracking party changes accurately, etc. I suspect this will be even more so in countries where legislators need to give up their seats (often temporarily) if they are then appointed to a Cabinet-level position, etc, though we haven't modelled any of those sufficiently yet to see it in action. --Oravrattas (talk) 13:31, 21 August 2017 (UTC)
@Oravrattas: "Nth Period" is never used here. First of all, you can as a member of parliament have several terms between two elections. A replaces B from October 2016- April 2017 while B has a parental leave. C gets a long time sick period from May 2017 to February 2018 and A replaces her then. In December 2019 D gets promoted to Government and A take Ds seat until next election. Members of the Government have a seat for 2-3 days until they are promoted and replaced. -- Innocent bystander (talk) 14:45, 21 August 2017 (UTC)
@Innocent bystander: "Here" is Sweden, yes? From what I can see the Riksdagen website does break up what would otherwise be a continuous membership into periods that coincide with the parliamentary terms/elections — e.g. — but in common with quite a few other countries simply refers to those terms by dates (e.g. "the 2014-2018 electoral period") rather than numbering them. And yes, Sweden definitely tends to have a lot more individual "membership periods" within a term than most other countries, in large part I suspect because it's one of the few countries where a member is actually replaced by their deputy when absent. Do you think that each entry of that table from data.riksdagen should have a separate position held (P39) entry in Wikidata, or do you think there's a better way to model the data here? --Oravrattas (talk) 15:24, 21 August 2017 (UTC)
No, I do not think such things as sick-periods or parental leaves should influence how we use P39. But if they are replaced because they are promoted to Government or as Chairman after only a few days, they probably shouldn't be mentioned as MP's at all. But you still find them listed in Such complicated situations as Mrs As above mainly affects deputies in larger party groups and constituencies. From what I have learned deputies are normally restricted to one single party and one single constituency. Most deputies in small constituencies as Gotland (2 seats) probably never see Stockholm other than on a postcard. (~25 constituencies for 349 seats and 8 political parties) -- Innocent bystander (talk) 16:14, 21 August 2017 (UTC)
I think we have a qualifier to include the information. The problem with attempting to follow Wikipedia categories is that some are based on the assumption that one can't have more than 199 category members. So these get broken down by party, electoral district, year of election, seniority, etc. All objective criteria, but in most countries not part of the description of the office itself.
--- Jura 15:34, 21 August 2017 (UTC)
I read the work that you did and I respect it. But even with the existence of the WikiProjects, we should follow the principle of least astonishment (Q22668). It is not obvious at all that it is a mistake (according to your proposal) to add position held (P39):Member of Parliament (Q16707842). Moreover, the failure of third-party tools to evolve with the reality of multiple terms and qualifiers reflecting them should not determine the intrinsic structuring of the data. I run a bot that is fulfilling position held (P39). I stopped it during more than one year to find a solution about handling multiple terms and the bot is now taking all qualifiers before editing a statement. I expect not less from other software. Right now, the proposal will result to a harder maintenance and consistency of data, especially on about internationalization. Louperivois (talk) 23:52, 23 August 2017 (UTC)
@Louperivois: there are a few different points wrapped up there, so let me try to separate them out a little:
  1. Distinct items such as Member of the 3rd Northern Ireland Assembly (Q37303713) cause translation difficulties. I'm afraid I don't follow this. Wikipedias often have separate items for the list of all such members in a term, and thus we also end up with Wikidata pages for each of those: e.g. Members of the 3rd Northern Ireland Assembly (Q6814606). Why is the first more problematic than the second?
  2.  A member of one term of a legislature is not semantically different from being a member of a different term. Again, I'm not sure I follow. We already often have separate items for the individual legislative terms themselves. We also often have separate items for the list of members in those terms (as above). Why is it then bad to also having the individual position of "member of term X" as well. Why can we have "List of ZZZs" but not "ZZZ" itself? This is a well-defined conceptual class, with multiple pieces of data that can be usefully applied to it, and which is very useful in constructing queries to get the data back out again in meaningful ways.
  3. Tools should be shaped by modelling decisions, not the other way around. In theory I agree with this. However, we also have the contrasting goal of wanting to have useful information available. A great many of the national legislatures of the world already have lots of data available to us that is structured in exactly this form (a person being a member of a specific term). If tools like Petscan and QuickStatements made it possible to import that data with the correct qualifiers, I am convinced that we would have a much higher amount of higher quality political information in Wikidata than we do at the minute (where in a great many cases all we know is that someone was, at some unspecified point, a member of their country's parliament). Should we decide later that per-term membership items were a mistake, then a single bot could migrate all P39s using them to point at their parent instead, with the suitable qualifier. But at the moment, if we do not allow such items, the only way to enter all this data is manually, or by writing a bespoke bot (or persuading someone to write one for you). In practice that means we will not get this data in a great many cases, or at least not any time soon.
  4. According to my proposed version, would a non-term-specific membership item a mistake? No, I don't think so. There are lots of scenarios where a "Member of Country's National Legislature" item will be sufficient. But I believe the value in also being able to use a more specific "Member of Term X" item where appropriate is sufficiently large, makes conceptual sense, and is easily migrated later if we do choose a different approach.
For me a key question not only in this case, but in general, is simply whether an approach makes our data better or worse. Switching to this way of working for UK data enabled us to enter complete and comprehensive membership data for the House of Commons (Q11005) (since 1997), the Scottish Parliament (Q206171) (since inception), and the Northern Ireland Assembly (Q285714) (since inception) in a matter of weeks, in a way that simply was not feasible before. I also have a personal interest in ensuring that data for the Riigikogu (Q217799) is equally comprehensive, and a professional (Wikimedia Foundation (Q180)-funded) one in helping people in other countries do likewise. The Estonian data is an awkward mismatch of position held (P39) and member of (P463) data at the moment, but this approach makes it really easy to migrate all those P463s very quickly, so that we at least have consistently entered data as a base for deepening it. For many other countries we have lots of useful data already available to us through, for example, Wikipedia categories (e.g. the term-by-term subcategories ofКатегорія:Народні_депутати_України), that currently could only be easily imported as a plain People's Deputy of Ukraine (Q12132454), thus losing lots of information, versus importing to term-by-term membership items as a much stronger basis for later improvement.
In my view this is all very low-hanging fruit, enabling us to greatly improve our political data very quickly. --Oravrattas (talk) 07:03, 24 August 2017 (UTC)
We anyhow need a model that allow us to do either way. We do not have items for the 3rd parliament of Skåne Regional Council (Q3232565). We do not even have an item for this specific parliament at all. We could create these items, but that would soon fill our days with only the task of creating such items. The model have to allow people to add statements like "Position held:Member of legislative body" "from:2018" "to:2022" of:"the Region of Scania". When we have items like "legislative body of Region of Scania" we can exclude the "of:Region of Scania"-qualifier, since it should be added to the "legislative body of Region of Scania"-item. And if we have somebody who stays in office the whole term, we maybe do not have to add "from/to" either, since such information should be included in the item about the "2018-2012"-item. And in the case of UK, the number (53rd) could be some kind of statement too. -- Innocent bystander (talk) 08:57, 24 August 2017 (UTC)
I agree wholeheartedly. No-one is claiming that this is the only suitable way for entering the membership data. It should always be as easy as possible to capture whatever information is actually available with the structure that already exists. Over time people might come along and create all the intermediate items that will allow us to be more precise in the modelling, but none of that should be required to simply capture the data initially.
And yes, the "53rd", should absolutely be in data. Usually we do that using a series ordinal (P1545) qualifier, though that's currently missing from most of the UK entries. I've added it to 53rd United Kingdom Parliament (Q21084470) and a few others now, and we should make sure it's on all the others. --Oravrattas (talk) 09:39, 24 August 2017 (UTC)

This discussion is very long and I'm not sure to understand what would be the problem (or at least, as the matter is quite complex, I fail to see the real problem). Can someone give a clear example of what of what was done wrong and how it should be corrected? Cdlt, VIGNERON (talk) 07:11, 6 September 2017 (UTC)

@VIGNERON: this is largely a parallel discussion to the one at Project:Parliaments. The issue here surrounds items like Member of the 54th Parliament of the United Kingdom (Q35647955), which are then used as a target for position held (P39) statements, e.g. on Caroline Flint (Q251095). (That is quite a good example as, due to the process of cleaning up old data still being in progress, she still has the data from the previous modelling — i.e. of a single Member of Parliament (Q16707842) membership — as well as the much richer term-by-term claims.)
I do not believe there are any serious objections to switching to separate P39s for each mandate/term like this (apologies if I am misrepresenting anyone) — or, at least, that does not seem to be the subject of this discussion.
Rather, the complaint is that these are entered using term-specific items: Member of the 52nd Parliament of the United Kingdom (Q36634044), Member of the 53rd Parliament of the United Kingdom (Q35921591), etc. instead of using Member of Parliament (Q16707842) with parliamentary term (P2937) qualifiers for 52nd United Kingdom Parliament (Q21084469), 53rd United Kingdom Parliament (Q21084470) etc.
Several arguments have been advanced for why this is bad:
  1. We already have an adequate way of entering this data (i.e. with parliamentary term (P2937) qualifiers)
  2. This is a significant and surprising change to the data model, with implications for people querying the data; using it infoboxes; etc.
  3. This requires a new "member of…" item for each legislative term, causing difficulties with translating into multiple languages.
  4. These items are an artificial construct, seemingly created solely to have been able to enter the data using QuickStatements: the positions do not actually exist.
  5. There are countries/legislatures for which this approach will not work.
(Although I am a proponent, not an opponent of the model, I hope I have fairly summarised the problems raised. If I have omitted any, or mis-represented, or weakened the force of any of them, I am happy to edit this into the strongest case. I am also happy to summarise the responses to these, if it is useful, but to keep this uncluttered for now, I will leave this as a statement of the objection first.)
If we were to decide that this approach was a mistake, the correction would be to migrate all P39s with these term-specific positions (e.g. Member of the 5th Northern Ireland Assembly (Q37279037)) to the parent class instead (Member of the Legislative Assembly of Northern Ireland (Q3272410)), and add a qualifier to each for the relevant parliamentary term (P2937) instead. I am not aware of any existing tools that would let us achieve that, but I do not believe it would be particularly difficult to write a bot to do so.
--Oravrattas (talk) 07:02, 6 September 2017 (UTC)
@VIGNERON, Oravrattas: So, I took last week off to go hiking, did very little Wikidata work, and it was great for clearing the mind :-). I'm very sympathetic to all these points (particularly #2 & #4). However, we're still left with the fact that we can use term-based memberships to do things (importing large amounts of data) that aren't otherwise very easy to do with existing tools. I wonder if a solution here would be to say "ultimately, this is a transitional model" - in the fullness of time we can aim to transition back to a more elegant and simple "basic item + P2937" model, and we should try and ensure that term-based items and standard queries are constructed in such a way as to make this possible.
For example, we could -
  • ensure that all term-based items also have an appropriate parliamentary term (P2937) qualifier so that queries using this format can pick it up
  • standardise on "instance or subclass of" queries to ensure that they pick up both models
  • develop a simple bot script to "upgrade" all "member of the Xth Y" memberships to "member of the Y" as and when completed
  • keep a clear central list so it's clear where any given group of politicians stands in terms of imported data (which fits very nicely with what EveryPolitician are trying to do anyway...)
  • run reports to check that "member of the Xth Y" items aren't being used once that set of data has been properly updated
Or would this approach just cause more confusion? Andrew Gray (talk) 16:29, 6 September 2017 (UTC)
  Support there are definitely some interesting ideas in there, particularly also adding parliamentary term (P2937) qualifiers even when it's seemingly redundant. I do suspect there will be some countries where it will end up making sense to keep the per-term membership items, but we can revisit that later to see if it does turn out to be so. But in principle I'm certainly happy to class this as a transitional model that enables people to get a lot more data in place, and then have a bot that can convert to the 'plain' membership items. --Oravrattas (talk) 20:22, 7 September 2017 (UTC)

Joint office holdersEdit

Not sure where is the best place to raise this, but User:Llywrch pointed out [4] that some positions have joint holders and it is useful to be able to identify the other joint holder -- for example the two consuls of ancient Rome, or MPs for many seats in England in the 19th century. The en-wiki template en:Template:Succession box has slots to show them, as can be seen on eg en:Tiberius Sempronius Gracchus (consul 215 BC) or en:Richard Cobden.

How should this relationship be entered on Wikidata? For the easiest retrieval by the template, would a new qualifier "office held with" be appropriate, to go on position held (P39) statements?

If so, does it matter if that appears to leave no way to indicate which dates apply, if the other incumbent changed part-way through the term (eg as in the Cobden example) -- could a template get that information from the item for the other incumbent? Jheald (talk) 21:43, 12 September 2017 (UTC)

together with (P1706)   tends to be used for this at the moment:
SELECT ?position ?positionLabel ?person ?personLabel ?with ?withLabel WHERE  {
  ?person p:P39 [ ps:P39 ?position ; pq:P1706 ?with ].
  SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
Try it!
We tend to already split position held (P39) statements in two if any of the qualifiers on that no longer hold (e.g. if someone changes party), so I'd suggest we do the same again here — i.e. one property to say that Cobden was MP with George Howard from 1847–1848, and another to say he was MP with Edmund Denison from 1848 on. --Oravrattas (talk) 07:27, 13 September 2017 (UTC)

confusing nameEdit

Can we rename this, e.g. "WikiProject Politicians"? WikiProjects are generally named after what they cover not who created them. --
--- Jura 08:57, 14 November 2017 (UTC)

Actually, it duplicates the existing Wikidata:WikiProject Politicians. I think it should be merged there.--
--- Jura 20:58, 15 November 2017 (UTC)

I think we have four things that overlap to some degree.

  1. Wikidata:EveryPolitician - a tracking page I set up way back in 2015 when EP wasn't linked to WD at all, then used by this project for a while in mid-2017 before it shifted here, now no activity.
  2. Wikidata:WikiProject every politician - where the EP work seems to sit, some activity
  3. Wikidata:WikiProject Politicians - set up by Lucy in June as part of this project, no sign of any activity (I work on politicians and honestly never knew it was there)
  4. Wikidata:WikiProject Parliaments - set up by Jura in August, some activity

Of these, #1 should probably just be redirected here (since this is the place where the actual work is being done). #3 is pretty much dead and could be redirected either here or to #4. I wouldn't say that #4 is "the initial name", though - it was created well after the EP work was underway. Andrew Gray (talk) 21:50, 16 November 2017 (UTC)

  • Wikidata:EveryPolitician was an import page, later used in relation to the WMF grant (to do further imports of data from there). Parliaments and Heads of Government are just related projects. In any case, I think we should make sure to aovid confusion of Wikimedia Foundation websites with MySociety's EveryPolitician website.
    --- Jura 22:00, 16 November 2017 (UTC)
    • The original EP page was a tracker for our politician properties to help with synchronising, not one intended for automated import - if anything, it helped them import our data, not the other way around. I know that, because, as I said, I was the one who set it up :-). On your second point, fair enough, but I don't think anyone is actually confused. Andrew Gray (talk) 22:16, 16 November 2017 (UTC)
      • It did read "This is a tracking page for an ongoing project to provide comprehensive coverage of politicians in Wikidata. This is currently a recipient of a Wikimedia Foundation (Q180) grant." and include series of import lists. Somehow the associated discussion is now disconnected.
        --- Jura 22:55, 16 November 2017 (UTC)
  • I agree with Andrew Gray. #1 should be redirected to #2 (including its talk page to avoid lost discussions). #3, which was never active, should be redirected to #2. We should also consider to merge #4 into #2, as it could scatter efforts made in the same topic. — Envlh (talk) 00:18, 17 November 2017 (UTC)
    • I agree with that. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 21:26, 20 November 2017 (UTC)
      •   Support for merging #1 and #3 into #2. #1 is essentially a redirect now anyway, and I previously tried to redirect #3. I'd be OK with keeping #4 as a separate project to track information about the legislatures themselves, distinct from the legislators within them, if people prefer that. --Oravrattas (talk) 08:41, 21 November 2017 (UTC)
        •   Done: #1 was merged in #2 quite a while ago. I've migrated the historic discussion from there to here as well, and set up a redirect on the Talk page. For now #4 lives on, as it has slightly tighter focus (similar to Wikidata:WikiProject Heads of state and government) — though we can revisit that later if there's ongoing consensus to merge them. #3 is still outstanding. --Oravrattas (talk) 11:00, 5 March 2018 (UTC)
  • I agree this should be merged into WikiProject Politicians, especially if the name "Every Politician" is being used by an external organization for a similar/related effort. The WikiProject should not portray itself as affiliated with an external organization. --Yair rand (talk) 18:04, 30 December 2018 (UTC)
"Merge into WP:Politicians" would really just be "rename this as WP:Politicians"; there's no activity there to merge to. (No objection to doing it, assuming it doesn't break links/maintenance tools etc - would need to spend a bit of time checking that before doing anything) Andrew Gray (talk) 14:02, 31 December 2018 (UTC)
At a purely practical level, there are close to 800 sub-pages which would presumably also need to be moved, and a variety of templates with calls looking to see if things exist at named locations, which would also need to be adapted. None of this is insurmountable if we choose to go down this route, but it's not as simple as a single page move. (Can admins move all the sub-pages en masse?) On the question of whether we should do this, I'm fairly relaxed either way. I had certainly hoped that by this stage the external EveryPolitican project hosted by mySociety would have essentially closed down as a distinct project in its own right, or be little more than a tailored front end to Wikidata's political information, and there would be very little scope for confusion or this project being seen as an outshoot of that one. As that didn't happen, it's definitely worth revisiting. (FWIW, I haven't really seen much confusion, or people thinking they're connected, other than from mySociety staff.) I do still think "WikiProject every politician" is a more evocative name (akin to WikiProject sum of all paintings) than a plain "WikiProject Politicians", so if we want to distance this from the external EveryPolitician, perhaps there's a better name we could use? --Oravrattas (talk) 15:30, 31 December 2018 (UTC)

Alaskan Legislators digressionEdit

I was curious about the name and scope of this project myself. Near the top of this discussion, I see the statement "it's named for what it is, and what it covers — i.e. every politician in the world." Well, my main focus is on politicians in Alaska. Looking at the "What links here" links pertaining to our state legislature, it appears that this site is only copying the limited subset of topics that Wikipedia has covered and is not even trying to encompass everyone who has served in our legislature. I would be quite happy to help, but have noticed that this site makes adding offline sources very burdensome, and it's going to take such sources to adequately cover certain individuals. Little wonder that I've come across so many "sourced" statements which are actually sourced to patently unreliable sources. If you want to help me out with this, feel free to get back to me.RadioKAOS (talk) 02:45, 17 January 2019 (UTC)

@RadioKAOS: I think you maybe had better explain a little more what you see the problem as being, RadioKAOS. For instance "but have noticed that this site makes adding offline sources very burdensome"? What? Adding offline sources is not hard; for instance, create an item for the source, and then use the stated in (P248) property reference. Or "it appears that this site is only copying the limited subset of topics that Wikipedia has covered"? What? This project is trying to add items for all politicians. Clearly what is added depends on who turns up to do the work necessary; and those that have turned up have been most interested in national legislatures and current politicians. But there's nothing to prevent users adding State legislature office-holders. The detail of work here far exceeds that done on any language wikipedia, and is certainly not constrained to or by any language wikipedia. Or "Little wonder that I've come across so many "sourced" statements which are actually sourced to patently unreliable sources."? What? By all means point them out to us, but my experience is of good sources, including, for instance, the websites of national legislatures. You say "If you want to help me out with this, feel free to get back to me." Well, okay, we'd love you to help. The floor is yours and there's nothing to stop you making a big impact by way of sorting out Alaskan legislators. There are quite a few people here who have done likewise with other legislatures, and would be willing to act as sounding-boards for your work should you need.
I cannot help but feel that you have somehow grasped the wrong end of the stick and are waving it about for reasons known only to yourself; I find your evalution based on "Looking at the 'What links here' links pertaining to our state legislature", in particular, confusing. Clearly if no-one has yet worked on Alaskan legislators, there will not be many 'What links here' links pertaining to the Alaskan legislature: but that is in the nature of things. Had work been undertaken on Alaskan legislators, there would be no need for acceptance of your kind offer of input. So, in closing, colour me a little puzzled. --Tagishsimon (talk) 03:23, 17 January 2019 (UTC)
@RadioKAOS: yes, please do join in. We'd love to have great data on the Alaskan state legislators! There are a lot of people here who have had lots of experience with different ways of adding lots of data like this, especially when there isn't very much already, so if there are specific questions you have (even as vague as "Where do I start?") please do let us know a little bit more about specific tasks you'd like to achieve, and I'm sure people will help out. --Oravrattas (talk) 09:45, 17 January 2019 (UTC)
In other news, more useful 'what links here' starting points might be found attached to member of the Alaska House of Representatives (Q26265831) and member of the State Senate of Alaska (Q21360107), neither of which look all that unhealthy to me, and both of which probably will belie your initial impression. --Tagishsimon (talk) 11:21, 17 January 2019 (UTC)


It's all gone a bit Pete Tong on the Zambia Positions front - why is this report - Wikidata:WikiProject every politician/Zambia/Q45398490/positions kaput? (Was copied from Wikidata:WikiProject every politician/Wales/Q3112646/positions. There are 30 cabinet minister items populated - WhatLinksHere/Q45398490.

  • You were missing the row template, so it didn't know how to display each row. I've added that now.--Oravrattas (talk) 09:55, 14 December 2017 (UTC)

More generally, I've given Zambia a good going-over in the last few hours, concentrating on the current government; there's some more work to be done on Cabinet Positions, and there are iirc 10 or 14 appointed rather than elected Assembly members ... right now they have inaccurate qualifiers as having been 'elected in' a certain general election.

Any 'you got that wrong' feedback welcome. thx. --Tagishsimon (talk) 07:57, 14 December 2017 (UTC)

Oh, the source of all the Zambia data is but I've not got around to working out how to use quickstatements to add a URL stylee reference. Any clues gratefully received. --Tagishsimon (talk) 08:41, 14 December 2017 (UTC)
How add reference -> reference URL (P854)? --Fractaler (talk) 08:56, 14 December 2017 (UTC)
In QuickStatements you start Source properties with an S instead of a P so they get added in the correct manner — see the section at “Each statement can be followed by an unlimited number of "source pairs"” at
@Tagishsimon: That's superb, thanks! I've added a couple of reports so we can see better what data we have or don't have: members and bios for the current term. Those highlight a few things that should be cleaned up:
  • The description for everyone is in quotation marks: ”Zambian politician”
  • Lots of constituency labels include a disambiguation phrase — e.g. "Kapoche (Zambian National Assembly constituency)". In Wikidata that should be on the description, but not the label itself: Help:Label#Labels_can_be_ambiguous
  • It looks like you've created new items for people, even if they already existed in Wikidata: e.g. Brian Mushimba (Q45382629) vs Brian Mushimba (Q27662555) — we'll need to merge all the people who now have two records.
  • Most of the people have no external identifier fields, and no source information on any claims, and don't link to any Wikipedia pages, so are liable to simply be deleted. Your suggestion to add reference information to all of these will help greatly with that.
Thanks for all your work on this! --Oravrattas (talk) 10:11, 14 December 2017 (UTC)
Thanks for your framework for this work! I'm afraid it all got a bit zzz, this morning, after an all-nighter. ”Zambian politician” and the label dismbiguators are now fixed; I spotted a couple of pre-existing MPs, but admit I could have done more checking ... I'll look at this. And then, yes, references for all of the assertions. DoBs also are low-hanging fruit, but I need to up my scraping game to get them efficiently.
I've also not given much thought to working out QIDs for the given name & family name of our MPs ... is there prior art here?
Best way I've found is to use OpenRefine. So, for example, if you have all the data in a Google Spreadsheet, and can generate 'given name' and 'family name' fields in that, you can then use the built in reconciliation tool to look each of those up and generate a new column full of Wikidata IDs. Happy to talk through that approach in more detail if you get stuck somewhere. --Oravrattas (talk) 22:41, 14 December 2017 (UTC)
I'll get around to addressing blanks & other QA issues surfaced by reports shortly: some advice on the 12th report constituency gaps, please: if we take a politician such as Davies Chama (Q45382648), he was appointed MP and has no constituency & thus was not elected. So the question here: what is the formulation for specifying he's an appointed MP, and how, if at all, do we deal with the observation that he has no Constituency (e.g. in a way that prevents our successors repeatedly trying to 'fix' the absence of constituency in the 12th report). --Tagishsimon (talk) 15:59, 14 December 2017 (UTC)
Best way to explicitly record having no constituency is to set a electoral district (P768) qualifier of 'novalue'. I've set that on Davies Chama (Q45382648) as an example. We haven't had to do that anywhere on a large scale yet, so we might need to tweak some of the queries to cope better with it if they don't do the right thing. --Oravrattas (talk) 22:41, 14 December 2017 (UTC)
The 12th Assembly report and the P39 no district query both hold their hands up for tweaking, although not perhaps with any urgency. 10 or so of our Zim MPs now have an electoral district qualifier of 'no value'. --Tagishsimon (talk) 01:00, 15 December 2017 (UTC)
@Tagishsimon: Excellent work on pulling all the Cabinet details together too. I've added a report to show all the current members (they're all showing an end date of when the table was last updated because of the way I'm constructing the SPARQL query. Suggestions for improvements there very welcome, which ideally don't require a different query for the current cabinet than historic ones). --Oravrattas (talk) 08:01, 18 December 2017 (UTC)

Czech Republic MPsEdit

At cswiki cs:Seznam členů Poslanecké sněmovny Parlamentu České republiky there is an exhaustive list of members of the Chamber of Deputies of the Parliament of the Czech Republic with its parliamentary terms. Most of these data are missing on Wikidata (but we have items for all MPs and parliamentary terms). Any idea (or volunteer) how to scrap these data? --Jklamo (talk) 20:01, 17 December 2017 (UTC)

Hi Jklamo. I have, for an exercise, scraped the QIDs of all of the politicians in the cs:Seznam členů Poslanecké sněmovny Parlamentu České republiky table, and together with the table data can, for each member, create the position held (P39) values of Member of the Chamber of Deputies of the Parliament of the Czech Republic (Q19803234) for each parliamentary term (P2937). Right now, though, we would lack electoral district (P768) & parliamentary group (P4100). We can infer start time (P580) and end time (P582), but we won't be taking into account those who left office early or joined a term late, such as via a by-election. There is electoral district (P768) & parliamentary group (P4100) on - at least for the current legislative term; and iirc some pages on show parliamentary group (P4100) for prior elections, albeit in the plain text of article lists. is not the most reliable source, I guess. I've not looked hard for, but think we might lack some electoral district (P768) & parliamentary group (P4100) for earlier legistalive terms.
I'll pause at this point to play some other games; we can discuss whether we wish to pursue an approach based on --Tagishsimon (talk) 00:37, 18 December 2017 (UTC)


Outstanding issues on Malawi...

New property: Hungarian MP identifierEdit

Please support the introduction of the identifier property Wikidata:Property_proposal/Hungarian_MP_identifier for members of the Hungarian Parliament.--Tdombos (talk) 14:24, 8 January 2018 (UTC)


Oravrattas Tagishsimon Jacksonj04 Owenpatel Markcridge Louisecrow Nomen ad hoc Tubezlob Siwhitehouse Mhl20 Alexsdutton Danadl Teester Zache a_ka_es Hasive Nat965 masti Papuass Jklamo ProtoplasmaKid Jmmuguerza Graemebp Pete Forsyth Jelabra Rfitzel Davidpar Canley Bodhisattwa CYAN Masssly MJL tdombos salgo60 Daniel Mietchen Lefcentreright Pedropaulovc Shahadusadik ミラP Xaris333 BrokenSegue M2545 Gnoeee DrThneed

  Notified participants of WikiProject every politician This property proposal has been hanging around a while, could you vote for it so it can move forward? Thanks. --Tdombos (talk) 15:16, 16 March 2018 (UTC)

Modelling of 'reserved' seatsEdit

There are cases in several legislatures where a subset of seats are reserved for representatives of a specific group. For example Kenya has representatives for women, Pakistan for women and religious minorities, and Uganda for a range including the army, youth, workers and persons with disabilities.

In Kenya, this is currently modelled using subject has role (P2868) qualifiers on the position held (P39) statements, pointing to an instance of a position (Q4164871). For example, Women's Representative (Q47484213) as seen on Fatuma Ibrahim Ali (Q17411041)

In Pakistan, this is currently modelled by setting the electoral district (P768) qualifier on the position held (P39) to a non-geographic instance of constituency of the National Assembly of Pakistan (Q33513247). Examples of this are reserved seat for Women (Q33513293) and reserved seats for minorities (Q33513308), as seen on Fais Azeem (Q18764047).

It makes sense to me for this to be more consistent, and the approach I prefer is that of the former - using subject has role (P2868) qualifiers pointing to an instance of a position (Q4164871). This neatly avoids two problems, one that some of these 'reserved' seats also belong to a geographical constituency (eg in Kenya, Women's Representatives are still elected by a district), and that electoral district (Q192611) is explicitly described as geographical in nature which would seem to preclude its use for these groups.

--jacksonj04 (talk) 12:38, 15 March 2018 (UTC)

Hmm, I would disagree - I think in most cases these can be handled using the existing constituency model (though we should find a way of marking a constituency as reserved, either by electorate or by candidate). In the case of Pakistan, for example, reserved seat for Women (Q33513293) is effectively a single sixty-member electoral region, covering the whole country, and elected by proportional representation. Listing the members with P768 to that constituency is consistent with how we handle other jurisdictions where both region and constituency seats are represented in the Parliament.
(A geographic basis is normal but not an absolute requirement for a constituency - Ireland still has two university-graduate seats, India has Anglo-Indian reserved seats in the Lok Sabha (Q48724360), Hong Kong has the remarkable functional constituency of Hong Kong (Q5508804) system, and various jurisdictions still have "at-large" seats covering the entire area.)
In the specific case of Kenya, as far as I can see, they have electoral districts with no restrictions on who can stand, then an overlapping but distinct group of reserved regional seats. In this case, it seems to me that this is effectively another 47 geographical constituencies, with restrictions, in addition to the existing ones - Fatuma Ibrahim Ali is the member for Wajir District, which is a constituency with one seat, that is reserved. Andrew Gray (talk) 23:32, 15 March 2018 (UTC)
We have right to vote (P2964) which could be extended to apply to restricted constituencies as well as elections, and used as per 8th Constituency of Regional Councilors of Nantou County (Q49934086). I do worry slightly, however, that that would only work in Kenya because the higher-level constituencies to which the Women's Representatives get elected don't also have other types of representatives. I can easily imagine a scenario where person X is the Women's Representative for the Eastern District, whilst person Y is the Youth Representative for that same district. Making two separate items for such a constituency, each of which has all the same data, other than the electorate doesn't seem right to me, but if we only had a single item, which knows that it has both types of electorate, I can't currently think of a way to answer questions like "Who are the current Youth Representatives", unless we also had some way to record that on the position held (P39) (as per the subject has role (P2868) example above) --Oravrattas (talk) 07:59, 16 March 2018 (UTC)
right to vote (P2964) looks to be a useful property, but doesn't really apply to Women's Representatives since any member of the electorate can vote for them, not just woman. Other systems may well make this distinction, but in Kenya anyone can cast a vote, and in Pakistan the reserved seats are filled by party lists and aren't directly voted for at all. We'd probably need a 'right to stand' property.
In Kenya there is definitely a good argument to be made that the Women's Reps do sit in an Assembly constituency with one seat which is logically distinct but physically identical to the Senate constituency, but Independent Electoral and Boundaries Commission (Q6016573) definitely considers them to be the same thing, just returning one result for Women's Rep and one for Senator. This is further muddied by the fact that the Senate also includes 16 seats for women which are allocated by party list, so we'd need to have another country-level constituency for those. --jacksonj04 (talk) 14:10, 19 March 2018 (UTC)
Ah, my mistake. I misunderstood who could vote for the Women's Representatives. right to vote (P2964) is definitely a red herring here then! --Oravrattas (talk) 21:44, 19 March 2018 (UTC)

Constituency subclassing in Hong KongEdit

In Hong Kong for the Legislative Council there are two types of constituency: geographical constituency (Q2973947) and functional constituency of Hong Kong (Q5508804). These are both subclassed from Hong Kong Legislative Council constituency (Q19999028). In addition functional constituency of Hong Kong (Q5508804) is subclassed from functional constituency (Q2973945). In turn both Hong Kong Legislative Council constituency (Q19999028) and functional constituency (Q2973945) are subclassed from electoral district (Q192611). This leads to functional constituency of Hong Kong (Q5508804) being subclassed twice in the same hierarchy - as show:

I feel that Hong Kong Legislative Council constituency (Q19999028) could be removed, then subclassing geographical constituency (Q2973947) directly from electoral district (Q192611) and functional constituency of Hong Kong (Q5508804) from functional constituency (Q2973945) alone - leading to this:

I think this simplifies the model without losing any information. Would others agree with this? -- Owenpatel (talk) 16:13, 25 May 2018 (UTC)

Multiple meanings of IntendantEdit

intendant (Q6491450) mixes the meaning of Intendant as a public administrator with that of Intendant as the manager in an opera/theatre company. I've just posted on Q6491450 talk page looking for a way of disentangling these meanings. Comments from members of the every politician project would be welcome. Owenpatel (talk) 15:34, 11 June 2018 (UTC)

Tying a legislative term to its election?Edit

At the moment there doesn't seem to be a commonly adopted approach for going from a legislative term to the election leading to it — currently that needs to happen backwards from something like:

I've seen a few examples where the link is made in the opposite direction, using significant event (P793):

Do people think that this is a sensible approach? Are there other ways that these have been (or could/should be) connected? Should we add this to the data model? --Oravrattas (talk) 16:29, 6 July 2018 (UTC)

For the second method, there's determination method (P459) ... not sure if that flies better than P793. --Tagishsimon (talk) 22:24, 8 July 2018 (UTC)
@Tagishsimon: I'm not so sure about determination method (P459). The discussion on that has a vaguely similar scenario which wasn't deemed appropriate, and it seems like a bit of a stretch to here too. Do you think this is significantly different enough? --Oravrattas (talk) 07:06, 10 July 2018 (UTC)
Wasn't deemed appropriate by a single response. I appreciate that the election-term relation was not anticipated by the proposer of P459, but we really do not stretch the definition of the property - "Qualifier to indicate method used to get the value that appears in the statement" - if we say that the method of arriving at the 57th parliament was the 2017 general election. Equally, I understand & see that there is stretch in terms of understanding that "determination method" means "the election which brought about the parliament". Neither is ideal, and I'm not advocating one over the other ... it's probably more important that we establish & document a pattern ... thus I'm as happy to go with significant event. --Tagishsimon (talk) 11:53, 10 July 2018 (UTC)

Number of seats?Edit

What property (if any) should we be using to represent the number of seats/members returned by a given constituency/district? At the moment number of representatives in an organization/legislature or won in elections (P1410) seems to be constrained for use with political parties/groupings ("Labour has 302 seats"), and number of seats (P1342) for use on legislative terms or on the parliament itself ("The Assembly has 270 seats" or "In the 1949-50 term the Parliament had 103 seats"). We don't seem to have a clear property to define the number of seats returned by a multi-member constituency. There's some discussion on Property talk:P1410 which suggests it was intended to be used for this case but I don't know if that was ever formalised. Andrew Gray (talk) 12:03, 28 July 2018 (UTC)

@Andrew Gray: some of the discussion around number of representatives in an organization/legislature or won in elections (P1410) is a little odd (including at the initial property proposal discussion), as some of the participants appear to have been unaware that number of seats (P1342) already existed. I don't think we'd need to stretch the definition too far to make it work for this case, though, if we class the "group" in question as the constituents in a particular place. This doesn't feel completely satisfactory, though it seems a little better than expanding number of seats (P1342). I don't think I have a strong opinion either way at the moment.
In terms of current usage, there appear to be currently 35 constituencies using number of seats (P1342) already:
  ?item wdt:P31/wdt:P279* wd:Q192611 ; wdt:P1342 [].
Try it!
and 47 using number of representatives in an organization/legislature or won in elections (P1410):
  ?item wdt:P31/wdt:P279* wd:Q192611 ; wdt:P1410 [].
Try it!
So nothing particularly persuasive there either. --Oravrattas (talk) 07:56, 30 July 2018 (UTC)

Geographic areas as constituenciesEdit

I'm currently researching legislatures in Australia. I'm doing this work in my job as a Political Researcher at mySociety (Q10851773) and it is part of a larger project to collect information on elected representatives across all countries.

At the moment, members of the Australian Senate have electoral district (P768) statements that point to items for the States of Australia. For example, Ian Macdonald (Q5982140) from Queensland (Q36074).

In Mexico, constituency of the senate of Mexico (Q53537091) has been created as a subclass of electoral district (Q192611) and then constituency items such as Puebla (Q53539319) have been created to keep the two concepts of "area as constituency" and "area as geo/political entity" separate.

What approaches have people taken elsewhere? What do you think best practice looks like? Ideally, it would be good to update the everypolitician documentation giving best practice. @Andrew Gray: @Jheald: as I see you both contributed towards a similar discussion Siwhitehouse (talk) 16:03, 7 September 2018 (UTC)

@Siwhitehouse: Both approaches are quite common: for example, United States Senate (Q66096) also tends to take the same approach as Senate of Australia (Q783330), whereas European Parliament (Q8889) has distinct items for the constituencies, even where those are whole countries (e.g. Germany (Q1350565)). I think it's generally accepted that ideally there would be distinct items for the place as electoral district, even where that maps exactly to somewhere that exists as a different type of administrative geography (e.g. a state), but pragmatically, unless / until there's actually a need to separate them out, if no such constituency items already exist, there's no real problem in using the existing items, especially if they are, by definition, the same geography (e.g. in a federal assembly). One downside to watch out for when there are separate items is people choosing the wrong one when adding data — e.g. there are currently 21 people with EUparl memberships with an electoral district of Czech Republic (Q213) (the country), rather than Czech Republic (Q5201764) (the constituency). To help detect (and potentially automatically fix) cases like this, it might be useful to always add a coextensive with (P3403) link between them. --Oravrattas (talk) 19:04, 7 September 2018 (UTC)

Dates for events expected in the futureEdit

When adding information on current legislative terms there is often an end date available which indicates when the current term is expected to the end. As this date is in the future there is obviously a degree of uncertainty about the date - while this may be the expected end date for the term, something may happen to trigger the term ending earlier than expected. There doesn't yet seem to be an agreed best practice across Wikidata for such future dates. Drawing on the example of Louis Vuitton Foundation (Q50376746) (thanks to Oravrattas) which uses a qualifier nature of statement (P5102) with a value of expected (Q50376823) for some future dates, I've added similar qualifiers to legislative terms in Brazil, for example Municipal chamber of Sao Gonçalo 2017-2020 (Q56460002)

Does this seem like a reasonable approach? Are there other approaches I should consider?

Thanks for any advice Owenpatel (talk) 10:28, 17 September 2018 (UTC)

Interesting. On the one hand this seems like a good approach; on the other, it is a trap for the unwary report writer. I wonder if in addition we should consider deprecating the rank of the statement. --Tagishsimon (talk) 10:42, 17 September 2018 (UTC)
The trap would be if someone is simply checking for the absence of a dissolved, abolished or demolished date (P576), rather than doing the FILTER(!BOUND(?end) || ?end > NOW()) dance, yes? If so, then that trap has already been sprung:
SELECT ?item ?itemLabel ?date WHERE 
  ?item wdt:P576 ?date FILTER (?date > NOW())
  SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
Try it!
Deleting all those seems doesn't seem particularly sensible, and though I can see some appeal in marking them as deprecated, that feels a little like abusing the fact that a `wdt:` query ignores those, rather than being what that rank is actually meant to be used for. The statement isn't erroneous — it's the best information we currently have. I would instead expect a deprecated statement like this to mean "This was originally meant to happen on date X, but actually didn't".
I'm inclined to agree - it seems like the wrong use of deprecated. To give an example of where you might need to use Deprecated in the usual sense with an expected date - in the UK if there was a snap election announced we would have a new expected date, so the original one would become deprecated (but would still have been an expected date), and the new expected date (still in the future until it happens of course) would be added. Owenpatel (talk) 14:50, 17 September 2018 (UTC)
nature of statement (P5102) is a fairly new property, so not widely used yet, but using it with expected (Q50376823) seems to be exactly what it was intended for, and at least this way we get to search for any dates in the past which were qualified like this, to investigate whether it turned out to be accurate or not (whereas now it would be very easy for these to go unnoticed for a long term).
This does reinforce the idea, though, that we should build up a good set of example queries, as there are quite a few odd corner cases and gotchas for the unwary all over the place. (Perhaps starting by adapting lots of the examples at Wikidata:WikiProject British Politicians/Sample Queries?)
--Oravrattas (talk) 13:25, 17 September 2018 (UTC)

Politician elected terms which do not match Legislative/parliamentary termsEdit

In many (most?) cases elected representatives are elected to a legislature for the length of the term of the legislature. So a position held (P39) statement for the representative can be given a parliamentary term (P2937) qualifier to link the position to the term. Obviously sometimes representatives leave or join the legislature mid-term, and this can be represented by adding start time (P580) and/or end time (P582) to the relevant position held (P39). For each term the representative is a member of a particular legislature they can have an additional position held (P39) which shows this membership.

However, there are other cases where a single elected term for a representative overlaps parliamentary terms. For example, this is true in the Brazilian Senate. The National Congress (bicameral) has terms lasting 4 years (list of federal deputies from Brazil of the 54th legislature (Q4640499), 55th legislature of the National Congress of Brazil (Q18479094)), but Senators are elected for periods of 8 years (covering two legislative terms) with a 1/3 and 2/3 seats being up for election every alternative legislative term. An example is Vanessa Grazziotin (Q7914656) - she was elected as senator in 2011 (at the start of list of federal deputies from Brazil of the 54th legislature (Q4640499)) and still holds the position (until the end of 55th legislature of the National Congress of Brazil (Q18479094)).

At the moment the position held (P39) statement doesn't accurately reflect the fact that she was a member of both legislative terms. In this situation for the position held (P39) I'm wondering if it is better to:

  1. Add one position held (P39) per legislative term resulting in two position held (P39) statements on each representative
  2. Add one position held (P39) per elected term resulting in one position held (P39) statement for the whole elected period with potentially multiple parliamentary term (P2937) qualifiers where the position goes over multiple legislative terms
  3. Something else...

Any advice/pointers are welcome Owenpatel (talk) 11:56, 9 October 2018 (UTC)

@Owenpatel: Interesting question. There are definitely a few cases where there isn't a clean mapping between personal mandates and legislative terms, but usually they're more out of sync than this one. In this case I would think that #1 would be more suitable: adding multiple different parliamentary term (P2937) qualifiers (or indeed multiple of any qualifier) to a single statement seems likely to cause issues for various SPARQL queries. The general rule of thumb is that once any of the qualifiers change, there should be a new statement reflecting the updated scenario. So when the legislative term (Q15238777) changes, then a new statement, with a new parliamentary term (P2937) qualifier, would be appropriate. Making sure that both position held (P39) statements have the correct (same) elected in (P2715) qualifier would likely be helpful for being able to query which seats in each term were re-elected then, vs carried over from the prior election. --Oravrattas (talk) 14:59, 9 October 2018 (UTC)

Members of committeesEdit

Just a notice that I asked a question on how we should model participation in committees on Wikidata_talk:WikiProject_Parliaments#Members_of_committees. Ainali (talk) 08:56, 18 November 2018 (UTC)

Suffrage / Women's suffrageEdit

How can we enter suffrage in state items, is a property needed? There are various dates for every country, like: universal suffrage, equal suffrage, women's suffrage (active and passive), voting rights of Australian aboriginals (1962), voting rights of black people in the USA ... --1rhb (talk) 12:24, 20 November 2018 (UTC)

I've wondered about this before; we probably need a similar approach for voting systems as well (first past the post, single transferable, etc). One problem is that these aren't always the same across all voters - for example, parliaments with both regional lists and constituency seats may have different voting systems for each one, and in some rare places different constituencies might have different suffrage rules (eg Wyoming had women's suffrage in federal elections for thirty years before it was universal across the US). This might be a problem to model in detail. Andrew Gray (talk) 13:07, 20 November 2018 (UTC)

Tanzanian GovernmentEdit

I've embarked on a slightly smaller scoped project (to begin with?) of updating all of the Tanzanian ministers/ministries and relevant MP pages for cabinet ministers on enwiki and here. Updating the pages on enwiki is fine, just a bit slow. Has anyone got any tools/queries that'll help me do the Wikidata bit faster? A lot of the ministries were merged, split and changed since the original enwiki pages were made, so I'm doing that leg work on enwiki. Particular things that are taking time on Wikidata that I think can be expedited are:

  • Finding all the existing (now probably defunct) ministry/minister items to add their replaced by/supersedes sets (any tools to do that work would be equally as useful!)
  • Creating and linking matching Minister/Ministry sets (particularly focus on actually creating the "position" (Minister) item so I can do the next bit)
  • Importing position data - after creating the minister page and then linking it up, there must be a quicker/automated way I can grab all of the info I've already put into enwiki infoboxes? Or can I do that here and use that data on enwiki? Either way WFM!

Any tips/advice would be great! --Lcawte (talk) 10:42, 19 January 2019 (UTC)

@Lcawte: that's excellent! (And I'll ping Tagishsimon here too, as I know he did some work on Tanzania before). I've added a report on the various minister positions Wikidata already knows about at Wikidata:WikiProject every politician/Tanzania/Q5015563/positions. That should pick up any item that is a subclass of minister, and has applies to jurisdiction (P1001) or Tanzania (Q924). (That generally finds more than part of (P361): Cabinet of Tanzania (Q5015563), though obviously it would be good to make sure they all have that too!)
That report will automatically also link to any pages that exist for showing the holders of those positions, should they exist at a known location (in this case Wikidata:WikiProject every politician/Tanzania/officeholder/[item id] — e.g. Wikidata:WikiProject every politician/Tanzania/officeholder/Q16147875. Those pages only need to contain a single line of configuration:
{{PositionHolderHistory|id=[item id]}}
and then that'll give a daily updating list of everyone who has held that position, including any obvious data inconsistencies.
There are a variety of tools for speeding up the data entry, though they depend slightly on whether you already have data in structured forms somewhere (or can manipulate it into such). QuickStatements is one of the most common, though you should be aware that it doesn't work well for adding the actual position held (P39) statements for people having held these positions, as if someone has held the same position more than once it will squish all the qualifiers together onto a single statement. There are a few different alternatives for that if you get that far though, that I'm sure people here will be very happy to help you with.
I'm not sure whether there are any good tools for reading this sort of position data out of infoboxes. I've toyed several times with investigating that further and/or writing something myself, but never quite get around to it. But maybe someone else here knows of something? My suspicion is that it would be easier to do it the other way around: add the information to Wikidata first, and then either switch to a Wikidata-powered infobox, or (if that's not yet acceptable for these specific inboxes within enwiki — I haven't kept up with all the nuances of that debate), construct a query that will output the relevant data on a one-off basis for inserting by hand into an infobox.
Hopefully that's a useful starting point at least. --Oravrattas (talk) 11:13, 19 January 2019 (UTC)
I'm also not sure whether you know any/much SPARQL, but a query like the following will also get you all the ministries Wikidata already knows about in Tanzania:
SELECT DISTINCT ?item ?itemLabel 
  ?item wdt:P31/wdt:P279* wd:Q192350 ; wdt:P17|wdt:P1001 wd:Q924 .
  SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
Try it!
--Oravrattas (talk) 11:24, 19 January 2019 (UTC)

Seating arrangements in parliamentsEdit

I think it would be useful to store data on where all members of parliaments are seated, in parliaments with stable seating arrangements. This would allow for interesting visualizations of votes, with indicators placed in each seat's location to show who voted what.

Modelling this sounds tricky. We could have an item for each seat, and a property pointing from the officeholder to the seat, but how would the location data of any given seat be stored? Geodata is one option, but that would require getting precise coordinates of each seat, which sounds difficult. Any ideas? --Yair rand (talk) 22:37, 29 January 2019 (UTC)

Geodata seems like overkill for this approach, and as you say getting actual values would be tricky - perhaps we could have a "seat number" listed in some way (catalogue code type qualifier?) and a blank SVG with the seat numbers linked to each cell? It's a bit complex, but anyone wanting to visualise this on an individual basis rather than just by filling in the right number of circles is probably willing to do the legwork. Andrew Gray (talk) 22:45, 29 January 2019 (UTC)
Hm, is there any way to store identifiers for areas in SVGs? (Would that be compatible with however the Graph extension stores things for use in charts like mw:Extension:Graph/Demo/HistoricalFertilityRates?)
Perhaps there should be a separate item corresponding to each parliamentary term's seating chart. "seating chart of the Nth Fooian parliament" could link to each member and seat, with one as main statement and the other as qualifier, while the item for the term could link to the seating chart. Some problems with this: Some parliaments have multiple locations (eg the European Parliament meets both in Strasbourg and Brussels, and has separate seating charts for each), and assemblies without distinct terms couldn't use charts... --Yair rand (talk) 00:04, 10 May 2019 (UTC)
I've started a proposal at Wikidata:Property proposal/Seating chart properties. --Yair rand (talk) 03:40, 14 May 2019 (UTC)


Any possibility of a members' list for Ghana, Oravrattas (talkcontribslogs) ... I forget how that element of Wikidata:WikiProject every politician/Ghana works (and why does the basic template not give us one?) thx --Tagishsimon (talk) 15:27, 13 February 2019 (UTC)

@Tagishsimon: The default template will give a couple of links to pre-generated SPARQL queries (for all members ever, and for a list of current members), as I don't know any way to actually autogenerate useful Listeria pages — but those queries only show if you haven't supplied a 'shortname' parameter yet. Once you've given that (as you have here), it replaces that whole section with a list of all the pages under "/Ghana/data/<shortname>/", on the basic principle that the SPARQL version is OK to get you started, but you almost always want to quickly replace those with custom pages. (Updates to the documentation to explain this better are very welcome!) I've added one at Wikidata:WikiProject every politician/Ghana/data/House/7th, but Listeria is returning a "503 - Service Not Available" at the moment, so we can't actually get to see the list yet! --Oravrattas (talk) 16:30, 13 February 2019 (UTC)
Oravrattas (talkcontribslogs) Much obliged, Tony; thank you, both for the report and the explanation. I might come back and look at the documentation. I'll be putting Ghana up to 3 stars now, since I've sorted out the missing genders, and we have DoBs for all but about 3/275 MPs. Masssly (talkcontribslogs) is organising editathons in Ghana to augment the data, so I think we'll get to 4 stars soon. --Tagishsimon (talk) 17:00, 13 February 2019 (UTC)
Excellent. I've merged a few of the newly created members into already existing Wikidata items, but this is great stuff! --Oravrattas (talk) 21:00, 14 February 2019 (UTC)
Oh, good man, thank you. Masssly and I have been doing the same, but evidently you're using a slightly different fu to us and finding things we missed. --Tagishsimon (talk) 22:41, 14 February 2019 (UTC)
@Tagishsimon, Masssly: How sure are you that the merge at Special:Diff/857636053 is actually the same person? The enwiki Agyenim Boateng might just be out of date, but there are a few things in that seem quite inconsistent with the details at --Oravrattas (talk) 11:09, 15 February 2019 (UTC)
Took a deeper look. They're indeed different people. —M@sssly 21:23, 15 February 2019 (UTC)

Members of European ParliamentEdit

A large number (maybe even all) members of the European Parliament from Italy link in their item to the poem Italy (Q19113443), not to the country or some regional electoral entity. Can somebody fix this? Steak (talk) 14:46, 23 February 2019 (UTC)

  DoneMisterSynergy (talk) 20:46, 23 February 2019 (UTC)

Parliament Diagram GeneratorEdit

I thought this project might be interested in a tool I've put together to generate semi-circular parliament diagrams based on information from Wikidata. It's available at It's based on, but with a UI to allow you to enter queries for whatever legislature you like. It also allows you to specify the language used for the legend (e.g. "en" for English) and the exact date to generate the diagram for (formatted as YYYY-MM-DD).

It will only work correctly if the item's position held (P39) has the following qualifiers: start time (P580), end time (P582), YouTube channel ID (P2397) parliamentary term (P2937), and parliamentary group (P4100). Additionally, the parliamentary group specified in parliamentary group (P4100) must have a sRGB color hex triplet (P465) statement in order for the group's dots to display in colour. The source code is available at Teester (talk) 09:32, 7 June 2019 (UTC)

YouTube channel ID (P2397)??? --Tagishsimon (talk) 10:16, 7 June 2019 (UTC)
Whoops. I meant parliamentary term (P2937). I've fixed it above. Teester (talk) 10:30, 7 June 2019 (UTC)
Worked great for 13th Saeima (Q57078934). Papuass (talk) 11:23, 7 June 2019 (UTC)
It is great, clever & very welcome; thanks. Three things: 1) wider data entry fields please (e.g. to fit the length of Member of the 57th Parliament of the United Kingdom (Q30524710)). 2) from & to dates have a day issue - I'm currently invited to find a date between 2017-06-08 and 2019-06-70 - e.g. with Position=Q30524710. Happens with the from date as well. 3) Typo on "The parameters proiduced no results". --Tagishsimon (talk) 12:24, 7 June 2019 (UTC)
I've made the data entry fields wider and fixed the date issue and typo. Let me know if there's anything else. Teester (talk) 14:16, 7 June 2019 (UTC)
Thanks. 4) Calendar UI for selecting the date (because, lazy me)? 5) Capitalise the label "language:" to match other labels. 6) Spacing between "language:" and its field seems to be larger than for the other three label/fields. Which is to say, I'm not finding much wrong with it ... except...
Consider which is a report on the current UK parliament. 649 members listed, but diagram-generator finds only 646; it's missing 2 Conservatives and 1 Labour member. All of the 649 seem to have the required parameters. Note also that 7) the stricture that they must have an End Date must be wrong/questionable for members of a current legislature? --Tagishsimon (talk) 14:39, 7 June 2019 (UTC)
Oops - parameters used to get the 646 were: Member of the 57th Parliament of the United Kingdom (Q30524710), 57th United Kingdom Parliament (Q29974940), 2019-06-07, en. --Tagishsimon (talk) 14:44, 7 June 2019 (UTC)
I've fixed the capitalisation and spacing. I've also adjusted the query slightly so it seems to catch everyone now in Member of the 57th Parliament of the United Kingdom (Q30524710)/57th United Kingdom Parliament (Q29974940) and still gives the same results for Teachta Dála (Q654291)/32nd Dáil (Q28976095), for example. Hopefully this fixes that problem. Next, to add a calendar! Teester (talk) 16:31, 7 June 2019 (UTC)
Thank you, Teester. I'll keep playing and see if I can break it in any other way; if so I'll let you know :) --Tagishsimon (talk) 17:30, 7 June 2019 (UTC)
Oh - you still might want to think about amending the end date advice to, for instance, * end dates (P582) (where appropriate). --Tagishsimon (talk) 17:32, 7 June 2019 (UTC)
I've made that change and also added a datepicker and some automatic input field filling. It now automatically fills in the last date of the chosen term (or today in it's still ongoing). Teester (talk) 13:12, 8 June 2019 (UTC)

@Teester: So, you're probably now sitting around, all glum, thinking "that project's finished, what do I do now???". Fear not. @Andrew Gray: has your back, and suggests [5] that you should amend the tool to provide a day-by-day animation of the changes to the make-up of a legislative assembly over a defined time-period. (Well, he didn't quite say this. He suggested I should grab 750 images & animate them, which seems a whole lot of work, no?) ;). Obvs, this is for no good reason other than it can be done. --Tagishsimon (talk) 20:26, 7 June 2019 (UTC)

I think this is a bit much for a feature request, to be honest :-) But I'm now wondering about running the code locally and feeding it "pre-programmed" answers (since I can probably generate a query for the number of seats each day across the lifetime of a Parliament).
@Teester:, this is an amazing tool. I was saying in a talk this afternoon how it might be possible to build something like this, but I never imagined it had been done a few hours earlier. Many thanks for it! Andrew Gray (talk) 20:35, 7 June 2019 (UTC)
@Teester: also one small bug report - for party changes, I think we usually give these as term A ends on day X, term B starts on day X (otherwise the queries sometimes get confused and think we had a one-day vacancy). The query seems to pick up both of these as being on the same day and return a duplicate - so if you run it for the UK on Tuesday 4th June 2019, it gives 655 seats (all the leaving Change UK members are counted twice). I think you can get around by this by changing the code to use ?end > rather than ?end >= - my test suggest it gives the right results that way. Andrew Gray (talk) 20:43, 7 June 2019 (UTC)
I've made the change you suggested to the query and it seems to work well. Thanks. Teester (talk) 13:12, 8 June 2019 (UTC)
  • I started a property proposal several weeks ago for improving generated parliamentary diagrams by storing structured data regarding parliamentary seating plans. I think that ideally, such diagrams would match the building setup. --Yair rand (talk) 05:49, 13 June 2019 (UTC)

What happened to Every Politician?Edit

Found this four year old catalogue in Mix'n'Match. Anyone know what happened to the (presumably) proposed property? Trade (talk) 23:55, 25 July 2019 (UTC)

@Trade: If I recall correctly, the Mix'n'Match catalog was set up at a point when the identifiers used by mySociety's project were neither stable nor unique, so wouldn't have made for a good Wikidata property. That project later moved to UUIDs, but I don't believe anyone set up a new catalog for them. --Oravrattas (talk) 15:11, 23 June 2020 (UTC)

Adding legislative periods with OpenRefineEdit

Hello ChristianKl Oravrattas Tagishsimon Jacksonj04 Owenpatel Markcridge Louisecrow Nomen ad hoc Tubezlob Siwhitehouse Mhl20 Alexsdutton Danadl Teester Zache a_ka_es Hasive Nat965 masti Papuass Jklamo ProtoplasmaKid Jmmuguerza Graemebp Pete Forsyth Jelabra Rfitzel Davidpar Canley Bodhisattwa CYAN Masssly MJL tdombos salgo60 Daniel Mietchen Lefcentreright Pedropaulovc Shahadusadik ミラP Xaris333 BrokenSegue M2545 Gnoeee DrThneed

  Notified participants of WikiProject every politician,

I added legislative terms of members of german parliaments. In QuickStatements it is not possible to add the same statement with different qualifiers twice. This is something what is possible in OpenRefine. Is there somewhere a place to share schemas created for the upload to Wikidata and can I create a page what describes how to add this with OpenRefine. I think this would be helpful and I can create some schemas if you want, from the structure what is already in Wikidata. I think that every Politician is a good Project and there are a lot of things that can be added to Wikidata during that project. --Hogü-456 (talk) 20:05, 14 June 2020 (UTC)

Open Parliament TVEdit

Looks like a new interesting project related to WikiProject every politician

- Salgo60 (talk) 06:25, 15 June 2020 (UTC)

Leave of absenceEdit

Edit: Posted the question in the Project chat instead, as it might be more general than just politicians.

Popperipopp (talk) 10:23, 16 June 2020 (UTC)

Politicians that leave their partyEdit

We have several cases in the Swedish where the member of parliament leave their party but continue working. Looking at the talk page of parliamentary group (P4100) it looks like no value would be proper, but there is also the item independent politician (Q327591). Any thoughts on what to use? Ainali (talk) 16:34, 21 June 2020 (UTC)

I don't have much of an argument but to me, no value looks like the most intuitive option. I'm not sure why the latter would be needed in the first place? Are there similar "null items" for other properties? Popperipopp (talk) 17:33, 21 June 2020 (UTC)
I originally was planning this to cope with the House of Lords, where there are two distinct kinds of independent - crossbenchers, who are the main grouping of independents, and "proper" independents, who either have been thrown out of their parties or choose not to become crossbenchers. Since then, we've had "named groups of independents" in the Commons (there were two different ones last year, it was a very complicated few months), which weren't quite a party but were distinct from the other no-party-independents.
We wouldn't be able to model this sort of detail using novalue, so it seemed to make sense to standardise on a generic item rather than use novalue for some independents but items for others. It also makes it a bit easier to group/count the independents in queries, which is handy. This may be more detail than you want to worry about, though :-) Andrew Gray (talk) 21:08, 21 June 2020 (UTC)
Thanks for you perspective. That reason didn't really occur to me, because in the Swedish parliament you can only be one kind of independent. Ainali (talk) 21:18, 21 June 2020 (UTC)
@Ainali: I don't know enough about how Sweden handles this, but one other more general consideration is that in countries with a strong parliamentary group model, it's often not strictly possible to be part of no group at all: if you leave one group, and don't join another one straight away, you're officially put in a group of one, and will be listed as such in voting records etc (see for example the history of parliamentary groups for Harry Harkimo in Finland). In general the trend in Wikidata has been to simply use independent politician (Q327591) for those cases, rather than creating a new parliamentary group item for the person, though the latter would technically be more correct. --Oravrattas (talk) 15:07, 23 June 2020 (UTC)
In Sweden they are not part of any group, and in the official records the field for party is just left blank. See this for a current example. Ainali (talk) 20:46, 23 June 2020 (UTC)

Use of P710 in item for legislative termEdit

Hello all. Here is the start of a legislative term modelled in Wikidata: Q96361127 (2019–2020 Massachusetts legislature in US). It utilizes participant (P710). Thoughts? Thanks. -- M2545 (talk) 14:03, 17 July 2020 (UTC)

M2545 Interesting approach! Generally speaking, the approach we've been using is the other way around - model the people as part of the parliament, rather than model the parliament as made up of a list of people. So in this case you could use position held (P39) statements on the individuals, with parliamentary term (P2937) qualifiers on them to link them up to terms.
One issue that comes to mind with this approach is that it might get challenging to maintain, with potentially hundreds of entries on each term item - eg for the MA legislature, you'd need 200 entries on the term for members, a handful more for officeholders like speakers, and then more for any mid-term changes through special elections etc. Andrew Gray (talk) 16:20, 17 July 2020 (UTC)
Andrew Gray-- Thanks for the comments. Wikidata entities already exist for many legislative terms (although at the moment they still need to be expanded with basic info like country, start time, etc.). So adding term names and dates to individual legislator's item would be just as much fuss as adding names to term articles. I also think it's more human-readable to have a long list of legislators. -- M2545 (talk) 22:13, 22 July 2020 (UTC)

Completed the list of senators in the 116th Congress of the United StatesEdit

Hi there,

I've only just heard about this discussion and I'm not up to speed yet, but I thought I'd leave a note letting you know that I grabbed the list of current senators from The XML version has a bioguide_id for each senator for which you cross-references to the congressional biography guide like so: This page also references an XML file for easy parsing. Following this trail I filled in the full 116th Congress into wikidata following the pattern of existing entries. The HTML version of the current senators has a link to a list of appointed senators which I used to clean up the start times. For all non-appointed senators I made the start time the same as the start of the term.

I'm also compiling a list of all 1985 historical senators using this method only starting with the search from the congressional biography guide mentioned above. Uploading that info means also adding some new political parties into Wikidata in order to follow the current pattern. I'm new to SPARQL, OpenRefine, etc, but I'm slowing getting up to speed.

Regards, Gettinwikiwidit (talk) 10:39, 22 July 2020 (UTC)

Challenge with P2715Edit

This comment relates to member of the Swedish parliament, but is probably relevant to other countries as well.

If a person is elected MP, then immediately becomes a minister (Q83307) in the government (or takes a leave of absence on the first day), we used to not to have corresponding position held (P39) statements for that person. But using that method, there was no way to record that the person was in fact elected in (P2715) in a given election. In order to validate that the number of MPs by district, party and election match up with official numbers, and to validate that the total number of MPs elected in a given election is correct (see query), I chose to add position held (P39) -> member of the Swedish Riksdag (Q10655178) statements for ministers etc as well.

Now all the numbers match up which is nice. But it also means that the total number of MPs for a given date will likely exceed the correct number (see query with a date chosen at random), as both the elected MP and her replacement will be listed. I'm not sure how to resolve this situation. Popperipopp (talk) 10:26, 25 July 2020 (UTC)

@Popperipopp: perhaps I'm misunderstanding, but won't this be a problem for any date where an MP leaves office and is replaced by someone else the same day? If so, I'm not really seeing why that's a problem: at the precision of a single day there really were, legitimately, more MPs than seats. That's not quite the same as saying that was true at any specific instant during that day, but we don't (currently!) record to that level of precision. --Oravrattas (talk) 07:13, 26 July 2020 (UTC)
Having exactly 349 MPs at any given date is not the main goal here. The situation I'm describing is rather that we'll always have substantially more MPs than seats if we record both ordinary MPs and their replacements as serving simultaneously. Popperipopp (talk) 11:28, 26 July 2020 (UTC)
@Popperipopp: Oh, you mean that you list the ministers as still being MPs even when they're in the Cabinet? The normal way to handle that in countries with a strict separation between the executive and legislature, is that person is first elected to parliament, and gets the P39 for that, but when they enter government, the parliamentary P39 ends, with qualifiers for end time (P582), replaced by (P1366), and a end cause (P1534) of entering government (Q51188211). Where a single party is in government, that could happen very quickly after the election, and someone might only be in the Parliament for a day or two, but in lots of countries it can take a little while for negotiations between parties to play out and the cabinet to form, so there could be weeks or even months in-between the two. So on days where someone leaves Parliament and gets replaced the list of people who were in office on that day will be higher than the number of seats, but on days where there were no changes, it should then be the correct amount. --Oravrattas (talk) 18:49, 26 July 2020 (UTC)
In Sweden it's not even a day or two. Since the government doesn't automatically resign so does not the ministers, and they are in service from day 1. Example: so even when Isabella Lövin (Q241208) was elected in 2018, since she already was a minister she hasn't been MP a single day. Ainali (talk) 20:20, 27 July 2020 (UTC)
@Ainali: what happens in the case where the election triggers a change in government? When do the previous cabinet positions stop and the new ones start? And similarly for the Parliamentary memberships of those people? --Oravrattas (talk) 07:20, 28 July 2020 (UTC)
@Oravrattas: That was actually the case in the last election, and you can sot of see that on the link for Lövin. The old government was in place until 2019-01-21 when the new one was formed. Now she happened to be in that one as well, but that is the reason for the two lines there. Ainali (talk) 15:29, 28 July 2020 (UTC)
Let's see if I can follow what happened here (and please let me know if any of this isn't quite right!):
So the question here is what to do with people who win the election, and so could take a seat in Parliament, but because they are a Cabinet member throughout the entire parliamentary period (or, if a current period, the period so far), they never actually take their seat (or have not yet done so)? (And so, in the official data, only ever have a "Kammaren — Riksdagsledamot — Ledig" entry, and never a "Tjänstgörande" one.) Is this all correct? Close but missing some nuance? Completely off? Sorry if this is belabouring things, but I'm trying to understand what's going on here in as much detail as possible. There are essentially two different factors to consider with anything like this: (1) how to enter the data in as generic a way as possible so that it's usable by people who aren't experts in an individual country's parliamentary and governmental structures and procedures; (2) how to capture all the nuances so that someone who is an expert can also get the precise information they need. So while there are probably several different ways of achieving the second goal, I'm keen that we also try to achieve the first. --Oravrattas (talk) 16:49, 28 July 2020 (UTC)
I think your description is completely correct. I also agree that 1. takes precedence over 2. Ainali (talk) 18:26, 28 July 2020 (UTC)
Phew, that's good! OK, now that I hopefully understand the problem better, I'll give this some thought. But my initial reaction is that there are a couple of possible avenues. The first is to give their P39 an explicit set of start and end dates of novalue and some other qualifier (perhaps object has role (P3831) or subject has role (P2868)) to an item that explains that they were elected but did not take their seat. The other is to have a separate position held (P39) target item, equivalent to the "Ledig" entries from the API, and have them hold that role for the period they're not actually sitting in Parliament. One advantage of the former would be that it would also let us record people who never take their seat for other reasons (e.g. if someone were to die between the election and the start of term). A key advantage of the second would be a much closer mapping to the data from Parliament itself, and I think (but need to ponder it a little more) it would mean that most standard queries for members would give the expected results, but queries that want to check something subtly different (e.g. whether the correct number of people were elected) could do so, albeit needing a slightly more elaborate query to check both position items (or a common superclass). --Oravrattas (talk) 19:41, 28 July 2020 (UTC)
Great summary! Going the second route would, obviously, simplify updates in the future as one can simply map the Riksdagen data almost 1-to-1 with WD. Have you come across this for any other country Oravrattas? I would guess it's not all that unusual, but perhaps I'm wrong. Popperipopp (talk) 17:44, 29 July 2020 (UTC)
There are certainly a lot of countries with a similar rule that if you take a Cabinet position, you can't also sit in the legislature, and so the basic principle of wanting to record that they would be members if not for that is not going to be unusual, even if there are variations in how that works in practice, and in how their replacements are selected.
It doesn't seem to be quite so common for the published parliamentary data to include people who haven't actually held the seats, however, although the timing of the various events does seem to make a difference in some countries. So for example, in countries where parliament is dissolved before the election (and everyone loses their seat), the status of members after the election, but before the re-opening of the parliament, does seem to vary.
In Estonia, for example, my understanding is that even if you are in a continuing ministerial role, the prohibition on holding incompatible positions doesn't really kick in until you would need to take the oath of office and actually sit in Parliament, at which point you need to actively resign one or other position. This became an issue after the 2015 elections when the Mayor of Tallinn, who had also won a seat in the parliamentary elections (in Estonia you can't hold both of those simultaneously either), was in a coma and couldn't officially choose between them at the end of March when everyone else did. As such he is recorded in the official data as having been a Member of Parliament until he was finally able to reject the seat in August, even though he was never "actively" so.
The closest parallel to the Swedish model and data provision I've seen is Norway, where this has been translated into Wikidata using separate items for member of the Parliament of Norway (Q9045502) and deputy member of the Parliament of Norway (Q16159375), and also a Q67205869 which seems to be used as a subject has role (P2868) target on the P39 when a substitute/deputy is replacing someone else: see for example on Camilla Strandskog (Q11963038). There are a few things in the modelling there that I don't think are quite right (particularly with subclass of (P279)), and this approach has someone essentially holding both positions (or really all three!) whilst they replacing someone, rather than toggling between them, but I haven't looked into whether that's an implementation decision here, or a 1:1 mapping to how the Storting reports the data. --Oravrattas (talk) 11:21, 30 July 2020 (UTC)

How to model which government proposed a billEdit

We've been discussing how to model bill (Q686822) by the Swedish government. Each proposition is signed by one or more cabinet members who presumably should be signatory (P1891) of the item. But how can we include what government they represented? E.g. Löfven I Cabinet (Q18176361), Cabinet Löfven II (Q60736270) etc. Using represents (P1268) as a qualifier seems to be a constraint violation. Popperipopp (talk) 12:04, 6 August 2020 (UTC)

@Popperipopp: cabinet (P5054) is currently documented as being restricted to being a qualifier on position held (P39), but it could potentially be widened to other scopes where a similar relationship needs to be expressed. --Oravrattas (talk) 12:42, 6 August 2020 (UTC)
Return to the project page "WikiProject every politician".