Shortcuts: WD:PC, WD:CHAT, WD:?

Wikidata:Project chat

From Wikidata
Jump to navigation Jump to search

Wikidata project chat
A place to discuss any and all aspects of Wikidata: the project itself, policy and proposals, individual data items, technical issues, etc.

Please use {{Q}} or {{P}} the first time you mention an item or property, respectively.
Other places to find help

Other places to connect
Start a new discussion
Edit
On this page, old discussions are archived after 7 days. An overview of all archives can be found at this page's archive index. The current archive is located at 2020/11.

Representing United States Senate classes[edit]

I'd like to have statements such as Bernie Sanders (Q359442) position held (P39) United States senator (Q4416090) indicate to which United States Senate class the given senator belongs to. This could be done by making a qualifier for this statement indicating which class but there is no obvious qualifier that applies. Further, a qualifier on the statement puts it on a par with elements describing the statement itself such as start time (P580) or end time (P582) which describe details of the actual holding of the seat whereas the Senate class seems to describe the seat itself which is only occupied by one senator at a time. Thus, while the electoral district (P768) for both senators in a given state covers the same geography, it feels more natural to split each electoral district (P768) in two, one for each class since it's not possible to have qualifiers of qualifiers.

Thus while representing senate classes as qualifiers is undeniably simpler to manage it seems less true to what we are modeling.

Thus my proposal is to re-label entries like Vermont (Q58425109) as the "Vermont delegation" which will have two subclasses: "Vermont Senate class 1" and "Vermont Senate class 2", both of which will subclass "Vermont delegation". Relabeling will make it easier to disambiguate the United States senate constituency (Q58425000) from administrative territory. Having each class subclass from the delegation will allow us to do queries with electoral district (P768)/subclass of (P279) when we don't care about the particular class.

This should make it easier to generate tables like https://en.wikipedia.org/wiki/List_of_United_States_senators_from_Vermont from Wikidata information. @Andrew Gray: @Oravrattas: @Tagishsimon:

  • I think there is a misunderstanding of basic concepts about the way Wikidata works. "electoral districts" are unrelated to the seniority of a senator. To determine a senator's seniority, one should generally be able to compare start dates. Please don't create any fictional electoral districts or other fictional data. --- Jura 11:47, 6 November 2020 (UTC)
I think you misunderstand what United States senate classes are. They have nothing to do with seniority. I won't be entertaining more comments from you in this project. Gettinwikiwidit (talk) 12:55, 6 November 2020 (UTC)
@Jura1: - "seniority" in terms of time served isn't anything to do with it. As you say, that can easily be determined by looking at start dates. Please try reading a bit more about these issues before accusing people of "misunderstanding basic concepts".
The issue is that there are three classes of senate seats; the class dictates the year they are elected. "Class 1" senators are equal in all other respects to "Class 2" or "Class 3", and it doesn't affect seniority; at the moment, the most senior senator holds a Class 3 seat. The two seats are defined as having different terms (see en:Classes of United States senators) which is why we end up with strange things like this week having two different elections for senators in Georgia, one for Class 2 and one for Class 3. This is quite an unusual approach for a multi-member system - most countries with systems like this elect everyone at once, and don't distinguish between the seats - but it's not "fictional data".
@Gettinwikiwidit: I think this proposal sounds sensible and I support it. Breaking down the "constituency" into two overlapping seats seems to make a lot more sense, since we can then do queries like "find the split in party affiliation among Class 2 senators"; the alternative is to hack something together based on dates of first election but that would break down for anyone who started in a special election. I would quibble with "Vermont delegation" - maybe "Vermont delegation to the US Senate"? - but otherwise it sounds fine.
(PS - the pings won't trigger if you don't sign! So re-pinging @Oravrattas, Tagishsimon: in case they have comments.) Andrew Gray (talk) 13:06, 6 November 2020 (UTC)
 Oppose fictional districts. This can easily be handled by one of the available qualifiers. --- Jura 13:08, 6 November 2020 (UTC)
Before adding further data, please create a full sample with a sandbox item. These highly visible items aren't suitable for editing experiments and cleaning them up takes me and others too much time. --- Jura 13:13, 6 November 2020 (UTC)
Saying something is easy without making a suggestion isn't at all useful. Gettinwikiwidit (talk) 13:14, 6 November 2020 (UTC)
I think you finally implemented most of my suggestions. Congrats to that. We just need to clean up the fictional dates. --- Jura 13:18, 6 November 2020 (UTC)
Please try to read more carefully. You have just suggested that this is more easily done with qualifiers and that was the reason for your opposing. Please clarify what qualifiers you mean. These conversations don't have to be so painful. Gettinwikiwidit (talk) 13:36, 6 November 2020 (UTC)

State + class is a seat and is analogous to a district, but it isn't a district as such. The state is the district. We probably need a distinct way to model seats that are not districts. Same problem arises (if we care) about modeling the present structure of the Seattle City Council, which has seven district seats plus two distinct at-large seats. - Jmabel (talk) 16:58, 6 November 2020 (UTC)

I agree that, as there are two distinct seats per state, we should have a distinct item for each. I'm less convinced by the subclassing suggested here. I think we only need three new items, named something like, "US Senate class 1 seat", "US Senate class 2 seat", and "US Senate class 3 seat", and then each of the seats can subclass the relevant one, and be connected to the relevant geographic item for the state via coextensive with (P3403). Other than the Cook PVI column, this would then allow us to generate an equivalent table to that at https://en.wikipedia.org/wiki/Classes_of_United_States_senators#List_of_current_senators_by_class --Oravrattas (talk) 18:49, 6 November 2020 (UTC)

  • @Oravrattas: That sounds like a very good way to do it. - Jmabel (talk) 20:42, 6 November 2020 (UTC)
  • So, hmm, something like:
Seems safest to keep P131 as well, since both are true and a lot of queries will expect it. But I think this model makes sense and gives us a good framework to work with. Andrew Gray (talk) 20:49, 6 November 2020 (UTC)
So we completely abandon entries like Vermont (Q58425109)? Gettinwikiwidit (talk) 02:26, 7 November 2020 (UTC)
@JMabel: @Oravrattas: @Andrew Gray: Also it's not clear how to connect a senator with one of these entries if we don't use electoral district (P768). What exactly is being suggested? To make a new "seat" property? I'm not sure what value we get in the model by making this distinction. I don't think there's much risk of confusion and I'm not sure who would practically want to use the distinction in a query. Gettinwikiwidit (talk) 02:31, 7 November 2020 (UTC)
In any event I don't think it's worth holding up making this change on getting this property added. Someone can revisit later if they feel strongly about it. Gettinwikiwidit (talk) 02:33, 7 November 2020 (UTC)
How does this look? United States Senate state classes It's easy enough to take this table and generate QuickStatements or load into OpenRefine. If we have a consensus, I'll create the new entries. (FWIW, I have no problem with other people using the info in this repository to upload to Wikidata. It's all ultimately taken from either Government sources or Wikipedia.) Gettinwikiwidit (talk) 03:07, 7 November 2020 (UTC)
I think we'd want to keep electoral district (P768) for linking all of these claims; the fine nuances of differences between districts and seats, geographic and at-large areas, etc, are best modelled on the item rather than by using a different linking property. The existing property already has seat, etc as aliases, and I think it's always been used to cover the broad span of such things.
In terms of what to do with the old items, I guess if we're not creating an intermediate "Senate representation in Virginia" item then there's no real point in keeping them - I guess we can switch to the new items and then list for deletion, or blank and redirect/merge them into the state. Andrew Gray (talk) 10:48, 7 November 2020 (UTC)
I agree with Andrew Gray on electoral district (P768). Most standard queries for getting members of a legislature will expect that, and I see no need to force a different approach here. --Oravrattas (talk) 12:44, 7 November 2020 (UTC)
  • Before doing any changes, please provide a full sample of the planned changes so the wider community can provide an input. @Gettinwikiwidit: please bear in mind the reminder by @MisterSynergy: about seeking community input. --- Jura 05:27, 8 November 2020 (UTC)
@Jura1: Seeking community input is precisely what this thread is all about. I think you misunderstand how a community works. There is a proposal currently being discussed. If there is something you don't understand or have questions about feel free to ask. You made a proposal which lacked specificity and ignored my comments directed precisely to such a proposal. When asked to be specific you dropped the topic altogether at which point I wrote it off as being unserious. More directly, a community is about about a discussion back and forth. I'm more than happy to entertain questions which are focused on the utility to the community at large. Further, we all understand that this is volunteer work. None of us are in a position to give assignments to anyone else here. Regards, Gettinwikiwidit (talk) 06:53, 8 November 2020 (UTC)
It's probably also worth mentioning that before I started this work that the U.S. Senate info was spotty and inconsistent. It had been left that way for two years despite your involvement. I'll be so bold as to suggest that my supplying of the entire history of the U.S. Senate and continuing efforts to make it consistent do in fact a service to the community. You have focused on the fact that some entries have dates in the future (of which there are many examples in Wikidata. They are not "fictitious".) Refusing to acknowledge accomplishments does a disservice to the community potentially discouraging future work which would also benefit the community. Gettinwikiwidit (talk) 07:00, 8 November 2020 (UTC)
Per Jura, please publish a coherent proposal before launching into precipitate action. I have nodding familiarity with political office P106 but I cannot make head nor tail of the proposal here. Bernie Sanders currently has a set of United States senator (Q4416090) statements, qualified by electoral district (P768). Is the suggestion now that he will have instead P106=Q4416090 PQ768="Vermont Class 1 senate seat"? In other news, if the issue is a need to qualify a P106 Senate statement to represent seat class, would there be anything objectionable about using, instead, object has role (P3831) as the qualifier - object has role: class 1 seat? --Tagishsimon (talk) 12:25, 8 November 2020 (UTC)
@Tagishsimon: For what it's worth, this seemed to be a coherent proposal to me. It pretty much works as you described, except with P39 claims, not P106. The change is for those statements to be qualified electoral district (P768):"Vermont Class 1 senate seat" rather than electoral district (P768):"Vermont senate seat".
In terms of using object has role (P3831) on the statement, this would work, but I think it is conceptually a bit shaky. It suggests that his membership of the Senate (the object) is "Class 1", "Class 2", etc; but that qualifier properly attaches to the seat, not to his incumbency of it, so it makes more sense to describe it on the item for the seat. This is consistent with how we handle modelling the situation where there are different "kinds" of constituency existing alongside one another - eg with the NZ Māori seats, we classify Te Tai Tonga (Q7690987) as being a Māori seat, rather than qualifying Rino Tirikatene (Q3527980) as object has role (P3831):holder of a Māori seat. Andrew Gray (talk) 14:48, 8 November 2020 (UTC)
Colour me unconvinced. Whilst I accept that the current proposal is workable &c, and that the different approaches discussed here might be seen - reporting aside - a 2*3s & a 6 affair ... Te Tai Tonga seems to be the south island Maori seat, and its item describes it as such. We have not had to go out and invent a new item coextensive with Te Tai Tonga as we're purposing to do here. Your dismissal of object has role is valid, but only with the frame you selected. I could as easily frame it as 'A senatorial position for the Vermont electoral district which takes the role of furnishing a class II senator" or somesuch. The proposal we see here is to invent Class I, II & III districts which, strictly speaking, do not exist. That seems a bigger crime than merely taking a slight tyre-wrench to the meaning of object has role. To the extent it matters, coining new districts hitherto unknown in the history of US Senatorial history becomes problematic for the naive report writer who considers, not unreasonably, that the electoral district for a senator for Vermont is Vermont, rather than our confection. Find me the senators who have a seat that is coextensive with Vermont will just have to be one of those very unfortunate and IMO unnecessary ontological complications that those that come after us have to deal with ... when there were other, how to say, non-breaking, means of conveying class. I don't think you're going to get unanimity on this, and I think I've now said all I have to say; Gettinwikiwidit - athough I'm not a fan of the proposed approach, nevertheless I'm happy that you're giving this corner of wikidata the attention it needs. --Tagishsimon (talk) 23:45, 9 November 2020 (UTC)
@Tagishsimon: Not a new invention at all. It's straight out of the U.S. Constitution. See en:Classes of United States senators. - Jmabel (talk) 06:39, 10 November 2020 (UTC)
Well, no. The district is Vermont. The class is I, II or III. We are choosing to conflate class into district ... they are distinct things. The solution prevents us from a simple query for Senators from Vermont. Instead the user must know to search for senators from a district that is coterminous with Vermont. That's a very poor and unnecessary solution. --Tagishsimon (talk) 08:17, 10 November 2020 (UTC)
@Tagishsimon: Are you saying that Vermont (as constituency) should have a single item as a multi-member constituency, with number of representatives in the Senate equal to 2? I think it's slightly more accurate to say that Vermont has two single-member seats than one multi-member seat, and whilst there are definitely pros-and-cons of each approach, including wrt querying, I do think it's already generally accepted practice to have to take a separate step to get from constituency to geography: consider producing a list of all the United States House of Representatives (Q11701) members for Vermont constituencies. --Oravrattas (talk) 08:57, 10 November 2020 (UTC)
I'm saying that there is a single electoral district; and that it is more important to be able to support plain queries such as "find me the senators from Vermont" than it is to conflate class - the specific nature of the seat in the district - into 'electoral district'. --Tagishsimon (talk) 09:32, 10 November 2020 (UTC)
@Tagishsimon:This sounds a lot like the discussion of the difference between a "seat" and an "electoral district" mentioned above. I don't disagree that the two separate concepts are being conflated here, but the decision was to punt for now because of all the places the two concepts are being conflated. Being a model, it seems to me to be less a question of whether it's real or not as whether the modeling the distinction is useful or not. Do you have any examples of useful queries which would be made more difficult by this conflation? Gettinwikiwidit (talk) 07:31, 13 November 2020 (UTC)
@Tagishsimon: We currently differentiate Vermont (Q58425109) from Vermont (Q16551) so even saying "find me the senators from Vermont" paints over the need to work with the model. With the proposed model, it's still possible to fetch the senators from Vermont using wdt:P39/wdt:P131 wd:Q16551 Gettinwikiwidit (talk) 07:43, 13 November 2020 (UTC)
It's probably worth pointing out that this query works with both models. Gettinwikiwidit (talk) 07:46, 13 November 2020 (UTC)
@Tagishsimon: Thanks for the more specific suggestions about how this could be done with qualifiers but as mentioned in my initial proposal at the start of this thread and as Andrew says here this is more about a property of the "seat" itself and not about the incumbency of the seat. Also discussed above is the distinction between "seat" and electoral district (P768). If you have anything to add on that topic, feel free. The proposal is to change all the electoral district (P768) qualifiers in all US senator entries to reflect the class of the seat being held. Rather than have it be a United States senate constituency (Q58425000) representing only the state in question such as Vermont (Q58425109) it would be one representing the state and the class of the seat. All seats of the same class across all states would subclass a single entry representing the class itself. (I'm personally less convinced of the usefulness of this, but I'm happy to play along.) The new seat entries thus would look exactly like the Vermont (Q58425109) entry but also have a coextensive with (P3403) property with the same value as the located in the administrative territorial entity (P131) property. Gettinwikiwidit (talk) 21:25, 8 November 2020 (UTC)
More generally though, I think the priority should be getting the info into the Wikidata store over arguing over properties which can easily be modified later on. I understand the drive to get it right the first time to minimize work later on, but I would argue that it's pretty hard to know what works well until you've seen it in action for a while. As @Andrew Gray: alludes to above this proposal hews closely to other models which seem to be working well. Gettinwikiwidit (talk) 21:25, 8 November 2020 (UTC)
I'm happy to mock this up for one senator. Is there a preferred place to create scratch items? Searching for "scratch" in Help and Wikidata didn't help much. Gettinwikiwidit (talk) 21:59, 8 November 2020 (UTC)
@Gettinwikiwidit: P3403 should be to the "real state", I think - Vermont (Q16551) is the "master thing" that determines what the borders are, and all the things that are defined as being coterminous with Vermont get defined by pointing to it.
On your other question. we don't really have a scratch items setup (there's the sandbox, but that's mostly for testing individual edits). I'd normally make some sample items to demonstrate with, but for now I've made a stab at sketching it out below based on this discussion:
  • [new] Vermont Class I senate seat
  • [new] US class I senate seat
This is how the NZ setup handles it, though I can see an argument for dropping the instance of (P31):United States senate constituency (Q58425000) on the individual seats and just having them be instances of Class I seats (it's a bit cleaner and simpler). But as you say, we can easily fiddle with the upper structure once the new items are in place in the P39 claims. Andrew Gray (talk) 22:24, 9 November 2020 (UTC)
@Oravrattas: @Andrew Gray: I made one example: Vermont Class 1 senate seat (Q101435082). I left of coextensive with because it has a symmetric constraint that I don't want to deal with at the moment. I'm also not clear on its purpose. I've also not linked it to the current US senator, but this should give the idea. Gettinwikiwidit (talk) 09:19, 10 November 2020 (UTC)
@Gettinwikiwidit: I think this seems good. One tweak I'd suggest is that United States senate seat (Class 1) (Q101434824) should be subclass of (P279) United States senate constituency (Q58425000) rather than instance of (P31) - it's a class of Senate seats, rather than a single instance of a Senate seat - but otherwise seems good to me. Andrew Gray (talk) 13:38, 10 November 2020 (UTC)
@Andrew Gray: Done. Also given the symmetry constraint, I wonder whether coextensive with (P3403) shouldn't exist between the senate seats entries for a given state. I don't know that I like having the state entry pointing to the seat. Gettinwikiwidit (talk) 01:44, 11 November 2020 (UTC)
I have QuickStatements ready to go to create items for all the other seats. Changing all the position held (P39) statements is more work I think because I believe I'll need to generate all new statements and then delete the old ones. I'll have to pull all the current data to copy the position held (P39) statements, edit the electoral district (P768) qualifier. I can do this by generating an appropriate SPARQL query to pull all the relevant fields from these statements. Is there a better way? @MisterSynergy: @Andrew Gray: @Oravrattas: Gettinwikiwidit (talk) 06:54, 11 November 2020 (UTC)
@Gettinwikiwidit: In general, I find wikibase-cli to be by far the best tool for working with P39 statements. I'm happy to help out with doing the import/migration of any of these if that's useful. --Oravrattas (talk) 07:42, 11 November 2020 (UTC)
@Oravrattas: Thanks for the offer. I may take you up on it, but I'm also trying to learn here. I'll have a look at wikibase-cli. FWIW, this (somewhat messy) SPARQL query says that in this case it shouldn't be terribly difficult. 
SELECT DISTINCT ?qualLabel WHERE {
  ?sen wdt:P39 wd:Q4416090;
    p:P39 ?statement.
  ?statement ps:P39 wd:Q4416090;
    pq:P2937 ?term;
    ?pred ?val.
  ?qual wikibase:qualifier ?pred.
  ?statement prov:wasDerivedFrom ?ref.
  ?ref ?p1 ?v1;
       pr:P248 ?v2.
  ?prov wikibase:reference ?p1.
  SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
}
Try it! There aren't that many reference types either. I also learned that I can reference the Biographical Directory of the United States Congress (Q1150348) directly as well as supply the US Congress Bio ID (P1157). Gettinwikiwidit (talk) 07:56, 11 November 2020 (UTC)
I'm suggesting wikibase-cli because it makes it easy to edit existing statements. That's generally a much better idea than the copy/delete approach. See, for example, update-qualifier. --Oravrattas (talk) 08:47, 11 November 2020 (UTC)
I just looked it up right after you first posted. I'll probably use that. I had suggested similar functionality to the QuickStatements folks. Thanks again. Gettinwikiwidit (talk) 09:18, 11 November 2020 (UTC)
This all looks great. Agree with the suggestion that wd-cli is probably the way to go for handling multiple statements - it's incredibly good at this sort of thing. Andrew Gray (talk) 17:21, 11 November 2020 (UTC)
  • Okay, I think I have this all lined up to make the change. I've created all the new senate seat objects. 
    SELECT ?seat ?seatLabel WHERE {
  ?seat (wdt:P31/(wdt:P279+)) wd:Q101500234.
  SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
    Try it! }

I've also lined up the data to make the changes to all the located in the administrative territorial entity (P131) qualifiers. Last chance to comment before I make this change. Regards, @Tagishsimon: @Andrew Gray: @Oravrattas: @Jmabel: Gettinwikiwidit (talk) 07:23, 13 November 2020 (UTC)

  • Please see the note on your talk page. I asked an admin to stop this given that you have been requested to seek input of the wider commmunity before make such changes to our data. --- Jura 09:05, 14 November 2020 (UTC)
@Jura1:, this is literally a discussion on Project Chat - it is what constitutes "input of the wider community". It is the third such thread since @MisterSynergy: asked for wider discussion (see October 2-14, November 3-6). The model has developed over the course of these discussions - for example, the 'XXth Congress member' items have been removed. As you yourself said above, "I think you finally implemented most of my suggestions".
In this thread, we have discussed the proposed model for another week and provided examples. You have not posted anything since the example was provided, and now you turn up demanding an admin intervene. This is simply disruptive.
I would also remind you that you are under an editing restriction, agreed by admins in 2019, which says you are "forbidden from undoing another autoconfirmed editor's edits to items about politicians, politics, and government without prior discussion" This revert clearly breaks those restrictions. Andrew Gray (talk) 11:37, 14 November 2020 (UTC)
I don't think this thread satisfies to what was asked from Gettinwikiwidit, but MisterSynergy is best place to state that. I suggest the restriction for Gettinwikiwidit and Oravrattas is extended to Andrew Gray as well. --- Jura 11:55, 14 November 2020 (UTC)
@Jura1: I don't believe I'm under any restriction. Please point me at the relevant discussion. Regards, Gettinwikiwidit (talk) 13:20, 14 November 2020 (UTC)


The change that has been made to the electoral district (P768) of US Senate P39s continues to be really really stupid, predicated on nothing more than one person's desire to encode the class of the seat, and a feeble grasp of semantics.

An electral district is a geographic area. electoral district (P768), to remind, is in EN labelled "electoral district this person is representing, or of the office that is being contested". The electoral district of the Oklahoma class 2 seat is, unambiguously, Oklahoma. Electoral district != seat. Was never designed to = seat. Is documented as not being seat. Yet here we are with the electoral district (P768) now taking values such as Oklahoma Class 2 senate seat (Q101498943), which is not a district, but a seat.

If you just want to make things up as you go along, and decide that you'll stuff inappropriate data into property statements that were never designed for that data, we might as well all pack up and go home.

This is simply the worst large-scale change I've seen in wikidata. It really is appalling. --Tagishsimon (talk) 21:35, 14 November 2020 (UTC)

More. The class applies to the seat, not the district. Just as the district for the AG of Texas is Texas, so the district for the class 1 senatorial seat in Texas is Texas. The solution to the class problem is to be found in a seat - P39s main property statement - rather than in its district. The solution would be three items to be used as the main values in P39s - "US senator for a class 1 seat", "US senator for a class 2 seat", "US senator for a class 3 seat". --Tagishsimon (talk) 21:55, 14 November 2020 (UTC)
Let's see if I can make this debacle any clearer. All of these different seats have the same electoral district.
Positions versus Districts in Texas - IRL
Position district
Governor Texas (or Texas electoral district)
Lieutenant Governor Texas (or Texas electoral district)
Attorney General Texas (or Texas electoral district)
Comptroller Texas (or Texas electoral district)
Land Commissioner Texas (or Texas electoral district)
Agriculture Commissioner Texas (or Texas electoral district)
Railroad Commissioner(s) Texas (or Texas electoral district)
Senator for a Class 1 seat Texas (or Texas electoral district)
Senator for a Class 2 seat Texas (or Texas electoral district)
And now look at the mess we've made:
Positions versus Districts in Texas - current debacle
Position district
Governor Texas (or Texas electoral district)
Lieutenant Governor Texas (or Texas electoral district)
Attorney General Texas (or Texas electoral district)
Comptroller Texas (or Texas electoral district)
Land Commissioner Texas (or Texas electoral district)
Agriculture Commissioner Texas (or Texas electoral district)
Railroad Commissioner(s) Texas (or Texas electoral district)
Senator Texas class 1 senator seat
Senator Texas class 2 senator seat
It's very plainly nonsense on stilts. To perform some reductio ad absurdum, we might as well go the whole hog:
Positions versus Districts in Texas - future debacle
Position district
Governor Texas Governor seat
Lieutenant Governor Texas Lieutenant Governor seat
Attorney General Texas Attorney General
Comptroller Texas Comptroller seat
Land Commissioner Texas Land Commissioner seat
Agriculture Commissioner Texas Agriculture Commissioner seat
Railroad Commissioner(s) Texas Railroad Commissioner seat
Senator Texas class 1 senator seat
Senator Texas class 2 senator seat
I just do not know how else to say, senatorial class is an attribute of the position, not of the district which elects a person to that position. --Tagishsimon (talk) 23:11, 14 November 2020 (UTC)
@Tagishsimon: You seem to be saying different things and conflating them in odd ways. If we could take the histrionics down a notch or two maybe we can have a rational discussion here. First and foremost I think discussions on the data should be based on comprehensibility and usefulness to the community at large and intend to center my arguments on these precepts. Second, the only reason this change is so large is because I supplied the vast majority of the US Senate entries just a short while ago. There were inconsistencies before I arrived but didn't seem to raise anyone's ire. Completing this data was a boon for the community. Hopefully we can all agree on that.

Thirdly (I'll stop numbering points from here on in), I think we can all agree that a US Senate seat is not a fictional thing, no one is inventing anything here. The only question is how to model it. If anyone needs convincing, just look at the Georgia special elections this year. Both seats are up and candidates run for *one seat or another*. The class III seat has two more years left on it and the class II seat begins a fresh six years. They are distinct.

Is there a difference between a district and a seat? Yes. I believe this was stipulated by everyone in the thread above. Is the district identical to the geographic area? I would argue no. An electoral district has a property of the area it represents but is not equal to it. Entities in Wikidata which predate my contributions seem to conform to that notion as seen by the difference between Vermont (Q58425109) and Vermont (Q16551). (Incidentally, some of the inconsistencies which predate me were that the two were used interchangeably.) Even here in order to avoid philisophical rabbit holes (stating what an abstract concept *is* gets into murky territory) it's probably best to say this was *modeled* by an electoral district (Q192611) entity which had electoral district (P768) property pointing at Vermont (Q16551).

In many cases there is a single representative for the entire district. In others there are more than one but they are elected at the same time. The US Senate is relatively rare in that two people holding the same "office" covering the same "district" occupy distinct seats. We're left asking how to model this. Your last suggestion is that the "office" should incorporate the "seat". The rest of the discussion in this thread has thought that the "district" can model the seat. I think both of these are compromises. I don't believe anyone thinks of the office as being "Senator from Vermont Class III" and the difference between a "district" and a "seat" has already been stipulated.

Again, these are two different approaches to modeling the reality. As for the relative merits of either, I think there's a healthy degree of subjectivity which goes into that. As people talk about the "office" more often than they do the "district", I'm inclined to prefer the model which leaves the position as United States senator (Q4416090). In any event I haven't heard any strong arguments about how one model is more useful than another.

Also, I think a more precise way to critique the chosen method is that the property electoral district (P768) doesn't apply, not that the seat is non-existent. One solution might be to make a new property "seat" which has an "electoral district" thereby distinguishing the "position", from the "seat" and the "electoral district". In the vast vast majority there will be a 1-1 correspondence between "seat" and "electoral district". That probably explains why it wasn't done before. I personally am for such clarification which might help to address Jmabel's concerns mentioned above.

To get back to the community, I think it's better to have this data in the data store rather than not. Having the data in the store means it's available for future modifications. I make no claims that this data should remain immutable for time immemorial. And while I agree it could be more precisely modeled, I'd argue that the form it's in now is largely comprehensible and useful. For instance, we can now sort each seat by start time to calculate which senator followed and was followed by which senator and augment the data with even more useful information. You mentioned one query above, which I mentioned was no more difficult in this model than it was in the preexisting model. I'm happy to further discusses the comparative merits of the models based on how difficult it is to extract info out of them. Regards, Gettinwikiwidit (talk) 03:10, 15 November 2020 (UTC)

So your reductio ad absurdum version probably is the most consistent, though at that point I would call it a "seat" and not a "district" and the "seat" would have a property of "district". But again, all models contain compromises though I appreciate that not all compromises are created equal. Gettinwikiwidit (talk) 03:23, 15 November 2020 (UTC)


@Tagishsimon: My first thought is that focusing on electoral district (P768) as a "district" not a "seat", and emphasising the geographic aspect of it, is I think a bit of a red herring. We already use it for non-geographic electorates - eg Irish Seanad panel seats like Administrative Panel (Q4683477), UK university seats like Combined Scottish Universities (Q5150880), Hong Kong functional constituencies like Medical (Q3248957). It's better to think of it, IMO, as an item for all forms of "electoral division", however that division is handled locally; it happens to use "district", but "seat", "electorate", and "constituency" are also aliases.
I can see the arguments for using the administrative geography (the state) as the value for electoral district (P768) for senators, but I don't think that splitting out a distinct item for the electoral entity necessarily leads to the silly effect of us having to create a new district for every statewide elected post. In the examples you give, we would normally say that someone is eg Texas Attorney General (Q7707525), and then set it up so that that item is connected to Texas, rather than saying they were elected to the post of Attorney General for the electoral district of Texas. (If we don't have an item, the usual approach is to say they hold generic post:of (P642):place, and use the item for the place; note P642 not P768). So I don't think the Texas example here does suggest this approach is absurd.
To my mind, the thing that makes sense for Wikidata is to use distinct electoral entities, even if it does mean adding a bit of abstraction. It means we have a consistent model ("a person holds a post with P768 electoral district, and the electoral district is P131 of an admin geography") across countries, and within parliaments (eg we don't have to have a different approach for House seats in Virginia and Vermont). We already did this abstraction by having items set up so that Texas senators were previously using Texas (Q58425106), not Texas (Q1439). (I appreciate you're contending we shouldn't have been doing that in the first place, but just to emphasise it's not a new feature.) And if we're doing that abstraction, splitting the Senate seat into two overlapping class seats isn't any more complicated than having one. So I think this is all reasonable and consistent with existing modelling approaches.
@Gettinwikiwidit:, I would be cautious about creating a new "seat" property to try and resolve this ambiguity. It would just cause endless confusion further down the line as we have to redefine all existing uses of "district" against a new property, and figure out which one to use in the majority of cases where the two are treated as being the same, or how to handle cases where some people would use "seat" and other "district" within the same context (eg should Vermont's at-large congressional district (Q4311018) use "seat" but New York's 1st congressional district (Q12357474) use "district"?). Far easier IME to say "yes, we accept there's a tiny bit of ambiguity between 'seat' and 'district' in some cases, but it makes sense from a modelling perspective to use a single property"; this is similar to the way that we collapse all existing subtleties of party affiliation into the parliamentary group (P4100) qualifier. Andrew Gray (talk) 10:01, 15 November 2020 (UTC)
In the short term, I'm going to reinstate the vandalized entry because it's more useful to have a consistent set of data while we continue to discuss. Gettinwikiwidit (talk) 21:22, 16 November 2020 (UTC)

Options for modelling[edit]

I've taken the liberty of dropping in a section heading to break this up a bit, and tried to set out where we are and the options for modelling this. I hope everyone is OK with this.

At the start of the year, we had (incomplete?) Senate data which did not reflect class, and had a single "electoral unit" per state in electoral district (P768) qualifiers. As of this weekend, with the edits discussed above, we have (complete) Senate data which has two different "electoral units" per state, each embedding a particular Senate class, in electoral district (P768) qualifiers.

There are two overlapping issues here, and I think it helps to consider them separately. We've discussed all of them at some point and they all have pros and cons.

Should we model senate class, and if so how?

Option 1 would mean we couldn't reliably work out class in a query (we could infer it sometimes, but not always); options 2-4 would mean we could easily query for it. Option 4 makes most Senate queries a little more complex but not massively so.

What type of items should we use in electoral district (P768) for senators?

Option A is simplest, options B/C make some Senate queries a little more complex. There are examples of both options A and B in other parliaments, though B is generally more common. Option D has implications for all political data, not just Senators.

Personally, I have a mild preference for some approaches over others, but to be honest I think I would be happy with almost any of them assuming we were consistent. Most combinations of approaches make sense and can be justified from a modelling perspective.

The only one I have major reservations about is option D, implementing a new property model to distinguish seats and (geographic) districts - I think the current broad use of P768 is very well established, and trying to substantially change this up to accommodate an unusual edge case could be a lot more trouble than it is worth. Andrew Gray (talk) 12:56, 15 November 2020 (UTC)

Thanks, @Andrew Gray:. I would like to suggest that when advocating one approach or another that we supply sample queries to demonstrate the benefits/demerits of the model. I believe the query to list senators sorted by class and start time (in order to fill in follows/follows by qualifiers) is unwieldy with option 2, since you need to fetch another qualifier out of the statement. As mentioned above anything you could do with the previous model you can do with the latest model using p:P768/wdt:P131. (or changing p:P768 to p:P768/wdt:P131 if using geography over state)
Moreover, I think there is value in separating the district from the geography mostly because it allows us to distinguish the political entity from the geographic one so that over time as these objects get more and more descriptive the whole thing doesn't become a garbled mess.
Lastly, I'll reiterate that I believe the primary concern should the the data's comprehensibility and usefulness and thus clearly lean towards having more information modeled rather than less information modeled as well making easiest the most common queries rather than less common queries. Comparing people's passion over one model over another isn't likely to get us anywhere. Regards, Gettinwikiwidit (talk) 21:01, 15 November 2020 (UTC)
  • I think it would be good to see references for each option especially for electoral districts.
Also, if information is repeated several times (as options 2, 3, 4) do, it would good to see a usecase for that. --- Jura 12:04, 22 November 2020 (UTC)

Merge[edit]

Main article: https://en.wikipedia.org/wiki/180-degree_rule Shall be linked with it: https://de.wikipedia.org/wiki/Achsensprung_(Film)  – The preceding unsigned comment was added by 188.96.224.152 (talk • contribs) at 13:56, 10 November 2020‎ (UTC).

IM channel datatype[edit]

A few months ago, IM channel (P8009) was created with string datatype instead of external id datatype, because the last datatype didn't support many URI schemes. In this discussion Lea Lacroix said that it would be possible fix that problem. So ¿it's ok changing P8009 datatype to external-id?  – The preceding unsigned comment was added by Tinker Bell (talk • contribs) at 03:54, 14 November 2020‎ (UTC).

Did you mean "URL"-datatype? external-id has no limitation. Please mention it on the property's talk page as well. --- Jura 11:59, 22 November 2020 (UTC)

Some language names is not translated to other languages[edit]

Hello, i view that some language names, such as:

  • Central Bikol
  • Korean (North Korea)
  • Belarussian (Tarasklevica ortography)
  • Chinese (Min Nam)
  • Egyiptian Arabic
  • South Azerbaijaní

among others, is not translated to other languages. How can translate this? Thanks. --Rodney Araujo (talk) 20:59, 14 November 2020 (UTC)

  • If it really says "Egyiptian" rather than "Egyptian" that should also be fixed. - Jmabel (talk) 21:20, 14 November 2020 (UTC)
  • Ditto for "ortography" rather than "orthography". - Jmabel (talk) 21:21, 14 November 2020 (UTC)
Language names are translated via CLDR, https://st.unicode.org/cldr-apps/v#/es// is the edit interface (for Spanish), request an account at http://cldr.unicode.org/index/survey-tool/accounts. Editing there needs to wait as there is an edit freeze due to CLDR rolling out a new release.--Snaevar (talk) 22:32, 15 November 2020 (UTC)
Hey, thanks for pinging me. @Rodney Araujo: can you confirm that the examples you mentioned are taken from the termbox (the list at the top of an Item page, including labels, descriptions and aliases)? In which language was your interface when you noticed it?
If it's languages in the termbox: indeed, we're taking the content from CLDR, except for the languages that are not listed in CLDR, in that case we add them directly in Wikidata. Lea Lacroix (WMDE) (talk) 16:50, 23 November 2020 (UTC)

Denoting the domain of awards and prizes[edit]

I am wondering which property should be used to denote the domain of awards and prizes in Wikidata. In Governor General's Performing Arts Awards (Q3405815), the performing arts domain for the award is denoted with field of work (P101). However, the description for field of work (P101) is: "specialization of a person or organization". This description does therefore not seem to apply to awards. At the same time, the type constraint (Q21503250) does include award (Q618779) (as well as museum (Q33506)) as accepted values. Is field of work (P101), then, the right property to denote the domain of an award? --Fjjulien (talk) 14:51, 16 November 2020 (UTC)

author  TomT0m / talk page Mbch331 (talk) Jobu0101 (talk) Tubezlob (🙋) Flycatchr Koxinga (talk) Fuzheado (talk) Mfchris84 (talk) Manu1400 (talk) Daniel Mietchen (talk) Nomen ad hoc (talk) 07:53, 16 November 2019 (UTC)

Pictogram voting comment.svg Notified participants of WikiProject Award --- Jura 11:49, 22 November 2020 (UTC)

Making US general elections consistent[edit]

I propose to model all the United States general election (Q26252880) instances on the 1790 United States elections (Q18356754) example so that the labeling is consistent. In addition I propose to connect them with follows (P155) and followed by (P156) qualifiers. I would also like to collect elections by which Senate class is up for election as collected here. I propose connecting them to the respective United States senate seat (Class 1) (Q101434824) using significant event (P793) but am open to suggestions. 

SELECT ?election ?electionLabel ?pred WHERE {
  ?election wdt:P31 wd:Q26252880.
  
  SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
}
order by ?nounLabel

Try it! Gettinwikiwidit (talk) 07:33, 17 November 2020 (UTC)

  • Please make a sample item and link to the revision you consider a sample.
https://www.wikidata.org/w/index.php?title=Q18356754&oldid=884540192 has currently just three samples. Description of that version isn't useful. See Help:Description for that. --- Jura 11:46, 22 November 2020 (UTC)

Is Playmates.com the only database of its kind out there?[edit]

As there does not seem to be a WikiProject Pornography (Q14942913) then I leave the following message here. According to this query: 

SELECT distinct ?human ?humanLabel ?playmate_id
WHERE
{ 
  {
    # Human
    ?human wdt:P31 wd:Q5.
    # Playmate
    ?human wdt:P106 wd:Q728711.
    # with optional id in playmates.com
    OPTIONAL { ?human wdt:P5346 ?playmate_id }
  }
  UNION
  { 
    # Additional humans with id in playmates.com
    ?human wdt:P5346 ?playmate_id 
  }
  SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
}

Try it!

Of the 567 "playmates"¹ on Wikidata, only 230 have an Playmates Playmate ID (P5346), in addition to the fact that according to its web map it only has 301 registered playmates.

Do you know of a more complete playmates database that can be added to Wikidata as a property?--190.31.204.3 03:01, 18 November 2020 (UTC)

  1. It is worth clarifying that the query also contains 81 people who have an playmates.com ID but whose occupation (P106) is not Playboy Playmate (Q728711), so you might want to check that.
/:An then you may have a look here for more related properties. Thierry Caro (talk) 12:31, 19 November 2020 (UTC)
Looks like playmates.com is regional-based. For example, Cathy Lugner (Q20798793) was published in Playboy Slovakia, and there is no official website for Playboy Slovakia. However Cathy Lugner (Q20798793) is described at playboy.de for whatever reason. There are other regional websites for Playboy, like playboyrussia.com. Due to unclear url stability and seemingly low amount of identifiers per each website (<500), I suggest to use described at URL (P973) instead of creating a separate external identifier for each regional variant. --Lockal (talk) 12:54, 23 November 2020 (UTC)

About people with two or more eye colors[edit]

Through the following query: 

SELECT ?human ?eyeColorLabel
WHERE
{
  ?human wdt:P31 wd:Q5.
  ?human wdt:P1340 ?eyeColor.
  SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
}

Try it!

And filtering the results externally to obtain people with two or more eye colors I got the following:

I do not know how to indicate in the query that only return items that have a property with multiple values, but if someone knows how to do it would be good to explain it to me for future reference.

I don't think that all the people in the list have heterochromia iridum (Q461486) so it may be necessary to check in other sources to see if this is due to a bad import.--190.231.242.204 14:41, 19 November 2020 (UTC)

@190.231.242.204: Here's a query to get people with multiple eye colours: 
SELECT ?human ?colours WHERE {
  {
    SELECT ?human (COUNT(?eyeColor) as ?colours)
    WHERE
    {
      ?human wdt:P31 wd:Q5.
      ?human wdt:P1340 ?eyeColor.
      SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
    }
    GROUP BY ?human ?humanLabel
  }
  # can further filter ?human here if needed
  FILTER(?colours > 1).
}
Try it! --SixTwoEight (talk) 14:52, 19 November 2020 (UTC)
can also be done without a subquery, using HAVING: 
SELECT ?human (COUNT(?eyeColor) AS ?eyeColorCount) WHERE {
  ?human wdt:P31 wd:Q5;
         wdt:P1340 ?eyeColor.
}
GROUP BY ?human
HAVING(?eyeColorCount > 1)
Try it! --Lucas Werkmeister (WMDE) (talk) 10:49, 23 November 2020 (UTC)

P5058 needs reboot[edit]

Due to recent changes in the way Instytut Teatralny im. Zbigniewa Raszewskiego (Q11713227) make their data available on-line to the public, our property e-teatr.pl person ID (P5058) requires some technical work and I would like to ask for help (maybe some bot edits would be possible)? In fact all the P5058 values are now redirects to a different website run by the same institution, namely encyklopediateatru.pl. So the first task would to be to create a new format as a regular expression (P1793) for P5058. Secondly we need to check (hopefully using bot) if the numbers given as P5058 values are still valid. From my random tests I can tell some are valid, but some are not - the bot should take new values from the new website. Finally, it may be appropriate to change the labels and descriptions for P5058, to reflect the new website the property is now linking to. I would be more than happy to assist as a native Polish speaker, but I definitely lack the technical skills. Powerek38 (talk) 19:58, 19 November 2020 (UTC)

  • Please propose a new property if the scheme changes. --- Jura 11:39, 22 November 2020 (UTC)
    • @Jura1: It appears the ID format hasn't changed, just the domain that displays the items. --SixTwoEight (talk) 13:00, 22 November 2020 (UTC)
      • @SixTwoEight: Thank you, it seems that your recent changes to the property page fixed the issues with malfunctioning redirects to the new domain. Powerek38 (talk) 20:56, 22 November 2020 (UTC)

Member of[edit]

For a mason do we say member_of=masonic lodge (Q1454597) or say member_of=freemasonry (Q41726) or say member_of=freemason (Q23305046)? If the exact lodge has an entry I use that, as here: Special:WhatLinksHere/Q64966925. --RAN (talk) 18:23, 20 November 2020 (UTC)

  • If the information of the exact lodge is available I would create an item for it and use it. member_of=freemasonry (Q41726) looks better to me then member_of=masonic lodge (Q1454597). ChristianKl ❪✉❫ 18:29, 20 November 2020 (UTC)
    • Shouldn't the value of member of (P463) be an actual organization? Not convinced that the changed from "instance of" to "instance or subclass of" for the type constraint was a good idea [1]. --- Jura 11:38, 22 November 2020 (UTC)
      • The exact lodge is the only one that fits exactly, do you think we need to create something new that fits "member_of=" better, or can we modify one to be a better match? We need something to let people know someone was a mason, all the other fraternal organizations have an exact match such as "member_of=Phi_Bet_Kappa" or "member_of=Schiners" or "member_of=International_Order_of_Odd_Fellows" or "member_of=Benevolent and Protective Order of Elks". Currently we are split among the various choices I mentioned above. --RAN (talk) 00:45, 24 November 2020 (UTC)
        • While Phi Beta Kappa or the IOOF are each a single organization, the Masons are not. Most lodges are affiliated and recognize one another, but there are definitely exceptions (e.g. Continental Freemasonry). - 01:55, 24 November 2020 (UTC)~
      • I was of the understanding that the lodges do form an organization together, if that's not how Masons work then I agree that individual lodges should only be used. ChristianKl ❪✉❫ 14:07, 24 November 2020 (UTC)
        • As I say, most do, some don't. I don't know my way around this in detail; someone else probably does and could weigh in with a better explanation. A good example is that the Order of the Eastern Star accepts women as equal members (and always has), which the F&AM still don't allow. - Jmabel (talk) 21:03, 24 November 2020 (UTC)
        • Modelling individual lodges (even if we can reliably work that out) sounds like it could get very complicated - there are (were?) thousands of them, and presumably a lot will only ever have one person linked in Wikidata. 21:49, 24 November 2020 (UTC)

How long does it take a change to a formatter URL qualifier to take place?[edit]

The formatter URL (P1630) for US Congress Bio ID (P1157) was no longer working so I supplied a new one with preferred rank per the documentation. If I reload an entity using this property (Frank Leslie Smith (Q3068384) for instance), I don't yet see the updated URL. I don't see anything in the documentation about how this is technically implemented. Do I have to wait for some cache to be flushed? Are the pages generated on some sort of schedule? Thanks, Gettinwikiwidit (talk) 08:47, 21 November 2020 (UTC)

Some caches are really to blame. The developers should probably be made aware that neither purging an item nor even making an edit will force the new formatter URL to be used (so it is possible there is a cache entry for this which was not invalidated correctly). --Matěj Suchánek (talk) 13:27, 21 November 2020 (UTC)
@Matěj Suchánek: Thanks very much. It seems to be working now. I still don't understand the mechanism. It would be good if the documentation had a few words to set expectations on when changes can expect to show up. If someone who knows would do that I be greatly appreciative. Gettinwikiwidit (talk) 21:32, 21 November 2020 (UTC)
Confirmed, now I can see the URLs correctly formatted. So judging from timestamps the delay could be somewhere between 5 and 13 hours but I really don't know. --Matěj Suchánek (talk) 11:48, 22 November 2020 (UTC)
Tracked in Phabricator
Task T112081
Longstanding problem: see phab:T112081. As I understand it, the URL will never be reformatted, until the page is next edited. @Andrew Gray: Do you know if that is still the case? Jheald (talk) 18:20, 23 November 2020 (UTC)
I think that's still the case. Although it's possible this is kind of a blessing in disguise - if someone changes eg the VIAF formatter url to go to some spam site, it's probably good that doesn't instantaneously trigger changes to a million linked items! I'll run the purging script for P1157 tonight to clean up this one, though. Andrew Gray (talk) 19:09, 23 November 2020 (UTC)
@Jheald, Gettinwikiwidit: Purging script is now churning away, and should take about 24 hours to touch every item and refresh it. I've had a look at a couple of items and they seem to be correctly displaying the new URL. Andrew Gray (talk) 21:21, 23 November 2020 (UTC)
As I understand it, the URL will never be reformatted, until the page is next edited. As I commented above, I did an edit and even that didn't help. --Matěj Suchánek (talk) 09:58, 24 November 2020 (UTC)

Soucis de fusion[edit]

Note: originally posted at WD:AN. Nomen ad hoc (talk) 11:45, 21 November 2020 (UTC).

Bonjour, pourriez-vous annuler la fusion de ces pages et les rediriger vers les + anciennes ([2] vers [3], [4] vers [5]). Il y a aussi cette page dont le contenu est erroné, le lien en français n'est pas le bon ([6] vers [7]) ; je l'aurais bien fait mais depuis que j'ai installé la fonction merge.js, je ne peux plus le faire manuellement. Merci d'avance, Méphisto38 (talk) 5 octobre 2020 à 16:00 (UTC)

Which one of these external identifiers can be considered reliable reference?[edit]

Hey, I'm adding references that I got from Wikidata:Automated finding references input dump. So far 40K has been added (and will add around 30K more). here's an example. I have been adding references from external identifiers that are obviously reliable (e.g. National library of Spain). But I'm not 100% sure of these ones. If you know them, can you tell me if they are reliable enough to have their refs added to unreferenced statements?

I feel Internet Archive ID (P724) and MusicBrainz label ID (P966) are clearly unreliable but I'm not sure about the rest. Amir (talk) 15:32, 21 November 2020 (UTC)

  • The question is what statement (property) you want to reference with it.
Unfortunately, I don't recall the Automated finding references input-team providing us with the list to review. Maybe this is available in the meantime.
You can obviously add all of the above identifiers to items. --- Jura 15:40, 21 November 2020 (UTC)
Sources are not simply reliable or unreliable. Some sources are self-published and can be used about information on themselves (such as ORCID). Note VIAF and sources based on them are not always reliable as they sometimes pull data from Wikidata.--GZWDer (talk) 16:49, 21 November 2020 (UTC)
  • I would add all of them as references. Data users should decide by themselves which references they want to use and which ones not. —MisterSynergy (talk) 19:38, 21 November 2020 (UTC)
Eh, we have minds, we should use them. Be not a robot. -Animalparty (talk) 02:28, 22 November 2020 (UTC)
  • I feel Catholic Hierarchy person ID (P1047) is very reliable when it comes to Catholic bishops and dioceses. It is in fact a website run by 1 person, but he uses only official Vatican publications, often published only in Italian or Latin, as his sources. Powerek38 (talk) 21:31, 21 November 2020 (UTC)
  • I agree with MisterSynergy. Some of these are not reliable enough to use them as a source for new fact, but still, it is much better to have claim referenced by unreliable source rather than unreferenced. --Jklamo (talk) 22:47, 21 November 2020 (UTC)
  • @Ladsgroup: Igromania ID (P6827) is reliable source. It's Russian-language web site of a very large database of video game news and reviews. Used as a reliable source at least on ru.wiki and en.wiki respectively. Regards Kirilloparma (talk) 01:46, 22 November 2020 (UTC)
  • FamilySearch person ID (P2889) is 100% user-generated and thus generally unreliable, but it can be credible if the info for a given person is backed by documents. Use with caution. -Animalparty (talk) 02:28, 22 November 2020 (UTC)

Thanks User:Kirilloparma and User:Powerek38. I started the bot with Igromania ID (P6827) and Catholic Hierarchy person ID (P1047) which adds 10K more refs. Amir (talk) 03:28, 22 November 2020 (UTC)

  • SANDRE ID (P1717) is produced and administered by a National Agency and thus should be considered as a reliable source. Ayack (talk) 16:25, 22 November 2020 (UTC)
    Thanks. Added the bot. Amir (talk) 21:25, 22 November 2020 (UTC)
  • @Ladsgroup: I'm quite familiar with GRID ID (P2427) - it is curated by Digital Science (Q30338362) with data pulled from a variety of national and international databases. It is not always correct, but I would judge generally reliable. ArthurPSmith (talk) 14:15, 23 November 2020 (UTC)
    Thanks. Added Amir (talk) 18:34, 23 November 2020 (UTC)

Query always timing out[edit]

Hi folks - I've been using this query for a few years, and suddenly in the last few days it's timing out every time I try it:

Non-fiction writers (&subclasses) with labels in Spanish-but-not-English

Are other people having trouble getting queries to run? Or any suggestions for optimizing this one?

Thanks - Kenirwin (talk) 16:21, 21 November 2020 (UTC)

Try something like https://w.wiki/nWc --- Jura 17:03, 21 November 2020 (UTC)
Thanks @Jura1: -- can you tell me anything about what you did here or if there's any documentation on getting the queries to run better? It looks like you took out the part about being an instance of a person, but I'm not sure what else is happening here. Thanks! - Kenirwin (talk) 20:56, 23 November 2020 (UTC)

Modelling of holocaust victim[edit]

I was looking around how to model a Holocaust victim (Q5883980) and I noticed that Anne Frank (Q4583) has instance of (P31) set to Holocaust victim (Q5883980). That seems to be completely out of line with our normal modelling of humans. I propose we change this to significant event (P793) set to The Holocaust (Q2763) with qualifier subject has role (P2868) set to Holocaust victim (Q5883980) (example edit). This affects about 5200 items, see https://w.wiki/nXB . Please comment at Talk:Q2763#Modelling of holocaust victim. Multichill (talk) 19:38, 21 November 2020 (UTC)

Modelling unseated politicians[edit]

Should we be using position held (P39) for politicians who were never seated? If so, is there an appropriate qualifier to indicate this? A few of the people on this list have position held (P39) United States representative (Q13218630): John Willis Menard (Q6264430) and Matthew Vincent O'Malley (Q6791362). FWIW, they came to my attention because there is no corresponding US Congress Bio ID (P1157) for them.

significant event (P793) is being used as a qualifier for P39 to indicate UK MPs who were elected but did not take their seats. https://w.wiki/nSZ --Tagishsimon (talk) 03:30, 22 November 2020 (UTC)
@Tagishsimon: Awesome. Thanks very much. Gettinwikiwidit (talk) 04:57, 22 November 2020 (UTC)
I would add that there's a difference here between "never turned up" and "unseated". In the latter case, we'd model them as being an MP until the day their election was overturned and use a suitable end cause (P1534), with the new one then taking effect from the date of the ruling. We also have the phenomenon of contemporary MPs who intentionally never take their seats for political reasons, but are still considered MPs by everyone involved; the plan is to represent those with electoral district (P768) as well (edit: now added post-1950). I don't think the US has anything similar. Andrew Gray (talk) 14:17, 22 November 2020 (UTC)
If they never held the position, the rank should be deprecated. --- Jura 11:29, 22 November 2020 (UTC)
@Jura1: Also a good idea. Thanks. Gettinwikiwidit (talk) 12:26, 22 November 2020 (UTC)
It's not really an idea, it's just what comes from Help:Ranking.
BTW are you asking you about election winners that "who_never_took_their_seats" (text) or who were "unseated" (heading). The later could be that somehow their mandate ended prematurely after they took office.
For the later, the end date should explain it. --- Jura 12:34, 22 November 2020 (UTC)
@Jura1: By "unseated", I meant "not seated" not "removed from seat". Again, I ran across these entries. I did not create them. Gettinwikiwidit (talk) 05:23, 23 November 2020 (UTC)

Question[edit]

Does WD:QS no longer support add sources? I got stuck in the task of adding the source during execution. (｀・ω・´) (talk) 04:04, 22 November 2020 (UTC)

you mean references? it should BrokenSegue (talk) 04:05, 22 November 2020 (UTC)
@BrokenSegue: Yep I mean reference, Is it no longer supported? (｀・ω・´) (talk) 04:10, 22 November 2020 (UTC)
worked the last time I used it. what issue are you having? BrokenSegue (talk) 04:48, 22 November 2020 (UTC)
@BrokenSegue: It will keep such this page, and then there has been no progress. (｀・ω・´) (talk) 05:51, 22 November 2020 (UTC)
Well, I found the problem, I entered an extra "?" and it didn't work properly. (｀・ω・´) (talk) 05:55, 22 November 2020 (UTC)

Using sex or gender (P21) in Lexemes, pronunciation audio: I get "Potential issues"(constraints?)[edit]

While editing the Lexeme for English: "frog" I added to a pronunciation audio file that the gender/sex is female. I get the impression that Property P21 isn't supposed to be used in Lexemes or it needs an update in constraints? LotsofTheories (talk) 08:03, 22 November 2020 (UTC)

Use voice type (P412).--GZWDer (talk) 08:50, 22 November 2020 (UTC)
There's a citation needed constraint here. I'm a bass (Q27911). Should I add to my user page I'm a bass? In Sweden I once visited a choir and they told me I'm "2nd bass"(guessing the translation) or "andra bas"(Swedish). Does this help in creating a reference for my voice type? LotsofTheories (talk) 11:23, 22 November 2020 (UTC)
Given that there seems no justification for the citation needed contraint I removed it. ChristianKl ❪✉❫ 17:37, 22 November 2020 (UTC)
@GZWDer the pronunciation audio (P443) doesn't have voice type (P412) in allowed qualifiers constraint (Q21510851). Should voice type be added as an allowed qualifier to pronunciation audio? LotsofTheories (talk) 06:03, 27 November 2020 (UTC)
I realized now that on the talk page for pronunciation audio (P443) there's an example about Lviv (Q36036) which happens to have voice type in it. I guess I might try to add it. If I can't edit it for some reason please add it if you can. LotsofTheories (talk) 06:40, 27 November 2020 (UTC)

How do I open all Wikidata entries from a Wikipedia category?[edit]

How do I open all Wikidata entries from all articles in a Wikipedia category? How would I add the same description to all of them quickly? Thanks, --Tastenlöwe (talk) 20:23, 22 November 2020 (UTC)

Try PetScan (you will need to change "Pages with items" and/or "Use wiki: Wikidata"). --Matěj Suchánek (talk) 09:18, 23 November 2020 (UTC)

P276 for online events[edit]

To state the obvious, many events have been moved to online mode due to COVID-19. I tried to add location (P276) to one such event (namely Q102227964), but values such as Internet (Q75) or online (Q73368) activated constraints warnings. So how do we say in Wikidata that something happened online? Powerek38 (talk) 21:09, 22 November 2020 (UTC)

an online event has no location. maybe indicate the location property has no value. then maybe use broadcast by (P3301)? BrokenSegue (talk) 00:58, 23 November 2020 (UTC)
The location of an online event is online, as Powerek38 rightly suggests. We are here to try to model the real world. In the real world, the location of an online event is online. If you check out Property talk:P276, it gives the location of 'Passover' as being 'worldwide', and of 'bronchitis' as the 'bronchus'. P276's description points to P131 for administrative loatins and P706 for geographical locations. Taking all of these together, I cannot see any good reason why P276 should not support the plainly accepted truth that the location of an online event is online. The cure is to remove the erroneous constraint. --Tagishsimon (talk) 01:07, 23 November 2020 (UTC)

Identifiers[edit]

Did Identifiers on Wikidata get removed? I see that new creations don't have a section for them. Thanks--PremierePrush (talk) 22:15, 22 November 2020 (UTC)PremierePrush

The section will appear after an identifier is added. Ghouston (talk) 23:05, 22 November 2020 (UTC)

Item with wrong value[edit]

.NET Framework (Q5289) have wrong value: the version 5.0 is for ".NET" and not ".NET Framework". It should be removed. --2001:B07:6442:8903:C186:597:CF6A:C596 10:30, 23 November 2020 (UTC)

playtest (Q7203549) and playtester (Q102246013)[edit]

How to make "connection" between them? Eurohunter (talk) 11:44, 23 November 2020 (UTC)

practiced by (P3095), field of this occupation (P425) Jheald (talk) 18:09, 23 November 2020 (UTC)
(strictly speaking, those are maybe the relations between playtesting and playtester). Jheald (talk) 18:11, 23 November 2020 (UTC)
@Jheald: Thanks. Eurohunter (talk) 20:45, 23 November 2020 (UTC)

Wikidata weekly summary #443[edit]

Wikidata-logo-en.svg
Here's your quick overview of what has been happening around Wikidata over the last week.
  • Tool of the week
    • EntityShape.js is a script that allows you to check an Item against an EntitySchema. On the Item page, it will display various information about the Item, like the properties that should be present on the Item.
  • Development
    • Query Builder: We're making it possible to switch between querying for specific values (e.g. all Items with a specific post code) or any value (all Items with a post code statement regardless of which one it is)
    • Correcting the behaviour of what automatically happens on Wikidata when a page is moved on Wikipedia and co to an excluded namespace with the "suppress redirect" option (phab:T261275)
    • Making language fallback indicators not show up for language variant fallbacks in a few more places (phab:T267502)
    • Adding html language attributes for statements linking to Forms and Senses (phab:T267023)
    • Working on language and lexical category fields turning red on unfocus even though no changes are made after a change in ooui (phab:T266936)
    • Finished building the 1.35 versions of the Wikibase base and bundle docker images and published to dockerhub (phab:T264538)

You can see all open tickets related to Wikidata here. If you want to help, you can also have a look at the tasks needing a volunteer.

Read the full report · Unsubscribe · Mohammed Sadat (WMDE) 17:20, 23 November 2020 (UTC)

SciTLDR: one line summaries of research papers[edit]

Has anyone played with this? Semantic Scholar is rolling it out now (Nature article). Would make a nice addition, and could help adding topics? --SCIdude (talk) 18:36, 23 November 2020 (UTC)

Are you suggesting adding the one-sentence summaries to Wikidata items on papers? Assuming the text is public domain (AI and other non-human creations are generally non-copyrightable), how would this play out in practice? -Animalparty (talk) 20:14, 24 November 2020 (UTC)
So, there's an open question here about storing computed results in wikidata. A while ago I proposed computing pagerank and storing that as a new property. Now this SciTLDR has an advantage over that that for a given version of the summarization model the output is static (since papers are immutable). I'm generally in favor of doing this sort of stuff but I think we need to be careful to properly version and attribute it. BrokenSegue (talk) 19:24, 25 November 2020 (UTC)
The idea was that a summary statement would somehow make the process of finding "main subject" claims easier, or on the other hand, improve any search over all articles. Having abstracts might be similar---but abstracts have an introduction, these summaries apparently don't. Note also that you need to train the software yourself, data is not provided. --SCIdude (talk) 16:11, 26 November 2020 (UTC)

Vaudeville performer[edit]

Is there any way I can automate adding occupation=vaudeville_performer where description contains "vaudeville performer"? I was doing it by hand, but there are too many. --RAN (talk) 19:06, 23 November 2020 (UTC)

@Richard Arthur Norton (1958- ): This report will list them. I presume you can quickstatements? 2 minute job. --Tagishsimon (talk) 10:03, 24 November 2020 (UTC)

Q94521341[edit]

At Azariah Dunham (Q94521341) the conflict is giving a warning, can someone fiddle with the subclasses for American Revolutionary War (Q40949) so that the warning disappears? --RAN (talk) 20:20, 23 November 2020 (UTC)

It looks like it should work? American Revolutionary War (Q40949) is an instance of war of national liberation (Q1006311), which is a subclass of war of independence (Q21994376) > war (Q198) > armed conflict (Q350604) > conflict (Q180684) - which is a permitted class. Michael Collins (Q173196) has conflict:Irish War of Independence (Q208297), which is also instance of (P31):war of national liberation (Q1006311), and that's not showing any errors. Wonder if it's a caching problem somewhere? Andrew Gray (talk) 21:32, 23 November 2020 (UTC)
It looks like the fiddling fixed it, and the cache kept it un-fixed longer than expected. --RAN (talk) 23:40, 23 November 2020 (UTC)

Template subpage to be used as language link?[edit]

Maybe someone could answer to user:Wolverène's question: Talk:Q15692365?--Estopedist1 (talk) 16:39, 24 November 2020 (UTC)

It's Special:AbuseFilter/36; it should block addition of non-notable subpages, which according to Wikidata:Notability are "/doc, /XML, /meta, /sandbox, /testcases or /TemplateData" for templates, but it is also blocking links to others such as /end and /header for which the requirement in the policy is that the item has more than one sitelink. Peter James (talk) 20:27, 24 November 2020 (UTC)

Wikidata:Notability/sandbox[edit]

I'm posting it here as this page hasn't been edited for over two (2) months, but is there any interest in proposing it? Should all the individual changes be proposed individually or could these be proposed as a package? Perhaps there are some people here that want to propose their own changes here, as I think that the Notability policy needs updating.

Wikimedia Commons uses a Wikidata-based system for organising media files now and having a much more broader Notability policy could benefit this project. Furthermore, Wikidata has items for individual celestial bodies that may have only been observed by a handful of astronomers, while there are major local buildings and monuments with entries on Wikimedia Commons but not Wikidata. I'm not proposing the change here in the Project Chat, I'd rather see more eyes on this page so more improvements could be made and the discussion could continue. Especially since discussions about the Notability policy have been held for years that seem to end abruptly with no changes happening often due to disinterest and all parties moving on. -- Donald Trung/徵國單  (討論 🀄) (方孔錢 💴) 22:09, 24 November 2020 (UTC)

What problem, exactly, are you trying to cure? --Tagishsimon (talk) 22:59, 24 November 2020 (UTC)
That’s the big question. I might be wrong, but this might be a solution looking for a problem. @Donald Trung: Could you summarize which concrete discussions of the last few months could have been resolved by applying your proposal? --Emu (talk) 14:39, 25 November 2020 (UTC)
I think the only difference between the cases is that somebody has bothered to create items for the astronomical objects (probably by importing from another database), but nobody has bothered to create items for the "major local buildings and monuments". I don't think it's a notability issue. Ghouston (talk) 00:52, 25 November 2020 (UTC)

Should I merge Q10858442 in Cell (Q2349697) ?[edit]

Both pages contain the Wikipedia disambiguation pages of the cell.

And936 (talk) 06:39, 25 November 2020 (UTC)

Probably not, according to Help:Modelling/Wikipedia_and_Wikimedia_concepts#Disambiguation_pages: "Disambiguation pages are grouped by form, not by meaning. All the Wikipedia and Wikivoyage pages with sitelinks to a Wikidata page should have exact same spelling apart from the exceptions listed in the guidelines of the respective projects." Ghouston (talk) 06:46, 25 November 2020 (UTC)
No, disambiguation pages are for exact string matches. It might be possible to have a way with redirect pages that links those together, but that requires change in redirect policies on some Wiki's. ChristianKl ❪✉❫ 13:53, 25 November 2020 (UTC)
@ChristianKl: So you say I should merge these 2 pages. Am I right? And936 (talk) 16:31, 25 November 2020 (UTC)
No, you are exactly wrong. You should not merge them. As ChristianKl and Ghouston have explained. --Tagishsimon (talk) 16:36, 25 November 2020 (UTC)

Bulk import on data from Sports-Reference.com College Basketball site - data issues[edit]

Someone appears to have done a bulk import of data from this site. Problem is they did not account for data already on the site, leading to significant duplication.

At a minimum, there is now significant duplication under P118 (league), and P54 (member of sports team). For example : https://www.wikidata.org/wiki/Q58008659

Other issues include duplication of players in cases where the existing record did not have P3696 populated, an inclusion of 2020 as end time (P582) for all players, including those still active.

Can someone suggest the syntax of a SPARQL search to find all cases where Q94861615 is entered twice for a page under P118 to at least start to understand the scope of the issues/duplication? CanadianCodhead (talk) 17:48, 25 November 2020 (UTC)

--GZWDer (talk) 18:15, 25 November 2020 (UTC)

Thanks, 9000+ duplicated, ugh...CanadianCodhead (talk) 18:37, 25 November 2020 (UTC)

9000 duplicted for that league. There are other leagues. Fortunately, per league, it's a job of moments to get something like quickstatements to delete the duplicate statement - Help:QuickStatements#Removing statements. --Tagishsimon (talk) 19:02, 25 November 2020 (UTC)

Technically duplication is possible for other leagues, you can play in the Bundesliga, leave to go play in Ligue 1 and then return to the Bundesliga. So long as the start and end times are populated it is perfectly valid. You can not however go play somewhere else and then return to college basketball. Your entire stay is contiguous. CanadianCodhead (talk) 19:33, 25 November 2020 (UTC)

I've spent a few hours in the last few days trying to clean up hundreds of duplicate newspapers the same user had imported. They've been apppropriately contrite but have done nothing to help in the cleanup, as far as I can tell. --Matthias Winkelmann (talk) 20:43, 25 November 2020 (UTC)
@Matthias Winkelmann: when complaining about any user it usually makes sense to ping them to allow them to respond to the criticism. ChristianKl ❪✉❫ 10:03, 26 November 2020 (UTC)

What's the proper syntax/format to use QuickStatements to remove just statement 1 from the items? So for example to remove from Q16205693 statement Q16205693-266E0290-2DCA-4AC9-B1E1-D404C7A4C3C1 ? I cant figure out how to do that. It seems straightforward in the help file, but it is generating an error when I try it. I use quickstatements to load data frequently, so am familiar with the tool, but cant seem to get the remove syntax correctCanadianCodhead (talk) 16:54, 26 November 2020 (UTC)

As below. Seemed to work. The first hyphen gets changed to a $. diff --Tagishsimon (talk) 17:12, 26 November 2020 (UTC)
-STATEMENT Q16205693$266E0290-2DCA-4AC9-B1E1-D404C7A4C3C1

Thanks - it was the $ I was missing. Should not try to read documentation on 75 minutes of sleep...CanadianCodhead (talk) 17:32, 26 November 2020 (UTC)

Tourism[edit]

Hello. I have just initiated Wikidata:WikiProject Tourism/Participants. There is no actual project yet but at least we can now ping people interested by this topic. If you are one of these people, don't hesitate to put your name in the list. Thierry Caro (talk) 18:41, 25 November 2020 (UTC)

Merging duplicate items[edit]

Hiǃ I wanted to request the merge of items Q15684647 and Q5955430, which are duplicated but apparently reject an automatic merge. Thanks in advance! --NoonIcarus (talk) 09:05, 26 November 2020 (UTC)

They're not the same people. Different DoBs, amongst oher things. That is why someone has added Q15684647#P1889 to prevent erroneous merges such as your suggestion. You *really* need to look harder at items before suggesting merges. --Tagishsimon (talk) 09:12, 26 November 2020 (UTC)

Need Help With Authority Control[edit]

Hello out there in Wikidata land! I'm very new to WD and know almost nothing about Authority Control. While I've tried to learn as I go, I must admit to failure. I added the Authority Control template to the "Rocky Kramer" Wikipedia article and then did my best to update each of the data inputs on the corresponding WD page. That resulted in nothing propagating on the article, so I asked at the WP Help Desk, where I was told I hadn't inserted identifiers. I tried looking up identifiers and only became further confused. Someone at the WP Help Desk gave me some WD usernames to ping and one directed me to this chat. Therefore, I humbly ask if there's anyone who can help me get the Authority Control working for this article? Thanks much. --Warriorboy85 (talk) 09:44, 26 November 2020 (UTC)

Identifiers are the ID properties found, for example, at Q42#P214 and below. Rocky Kramer (Q100744874) lacks any of these. Find some values and appropriate properties for your musician and, if they are included in the IDs which the wikipedia template pulls through, then all will be happy. --Tagishsimon (talk) 10:08, 26 November 2020 (UTC)
That's where I get lost. First, where do I find the identifiers and once I have them, where do I insert them to make the Authority Control populate? I'm assuming I enter the Identifiers on the WD page and they will automatically propagate to the article? I apologize for being so ignorant when it comes to Authority Control, but it just seems complicated to me. Thank you for your help! --Warriorboy85 (talk) 10:51, 26 November 2020 (UTC)
P.S., I mean where do I find them for a musician? --Warriorboy85 (talk) 10:54, 26 November 2020 (UTC)
@Warriorboy85: For a musician, probably places like discogs https://www.discogs.com/artist/7596392-Rocky-Kramer or musicbrainz https://musicbrainz.org/artist/f55827bb-4224-4db8-bf72-0a19a3f106ef/works ... for these two the value 7596392 would be added with Discogs artist ID (P1953) and f55827bb-4224-4db8-bf72-0a19a3f106ef would be the value for MusicBrainz artist ID (P434). If you look at the property items, you'll see use examples which hint at which bit of the URL you need. To get ideas for other IDs, look at items for other musicians; does Rocky have an entry at the same places they do? Does that help? --Tagishsimon (talk) 10:59, 26 November 2020 (UTC)
@Tagishsimon: Sorry for the delay getting back to you. I went to sleep. Yes, that's very helpful. I have now looked at both Discogs and Musicbainz and I believe I'm starting to catch on. However, now that I have those two identifiers, where do I add them in Rocky's WD page? I see lots of places I can add numbers as references, but I don't see anything listed for either Musicbainz or Discogs. Should I add them as new categories? --Warriorboy85 (talk) 16:40, 26 November 2020 (UTC)
@Warriorboy85: Right at the bottom of the Rocky Wd page, a link saying "add statement". Click that. THe cursor will focus on the property cell os a statement box. Enter part of the 'Discogs artist ID' string into that, and chose the property from the list it provides. Then tab thru to the central cell and add the value. Hit publish. --Tagishsimon (talk) 16:45, 26 November 2020 (UTC)
@Tagishsimon: Okay, now that really helped me understand a lot! I can't thank you enough. I've added those numbers and now have a basic idea of what I'm looking for. I'll look at other musicians to find the types of identifiers that might be out there. I really appreciate your help. This WP and WD community is exceptionally helpful and I appreciate and value it greatly. Happy Thanksgiving! --Warriorboy85 (talk) 17:13, 26 November 2020 (UTC)
@Tagishsimon: Well, I got Rocky's ISNI and ORCID numbers posted and they now show along with his MBA number. Although I added his Discogs number and it took the number, it isn't showing. He also has a BMI Repertoire at [Kramer BMI] I have attempted to input his CAE/IPI # and even tried listing individual songs by their ISWC numbers, but the exclamation point keeps coming up saying there is missing information. Any idea what I am doing wrong? Thanks again. --Warriorboy85 (talk) 20:37, 26 November 2020 (UTC)

List of Q-items by number of statements[edit]

Hi, for Wikidata_talk:Lexicographical_data#Semi-automated_import_of_missing_lexemes_based_on_the_Q-items_with_most_statements I would like a list of Q-items (optionally that has a label in Esperanto) sorted by the number of statements. Is that possible without having to download the dump? If someone has access to a database where this information is available I would be very happy to receive a list of the first 50.000 items or so.--So9q (talk) 10:59, 26 November 2020 (UTC)

Here's what I came up with, because how can you not invest time in a request that includes the phrase "optionally has a label in Esperanto"): Query
This is rather not what you asked for, unfortunately. I couldn't find an existing list. Absent that, the straightforward query (as above without the line that includes "P31") is certain to time out, because it involves reading all of the data.
I am not entirely convinced that list would be helpful for your purposes, though. Intuitively, I believe a list by count of statements will include (1) countries, (2) movies (lots of actors etc.), and (3) other artefacts. Every paper coming out of the Large Hadron Collider has every single project participant as a co-author, for example. If that is reflected in our data, the list would be dominated by [Y particle flux in plasma conjunction is tri-symetric under Gedfeller conditionality]. There may also just be items with proper names at the top of such a list, because these are the same in any language and therefore have lots of (auto-generated) labels.
In the query above, I limited the list to instances of subclasses of human activity, which allows it to complete within the 60s we get. The chosen class(es) additionally restrict it to items that would seem to be meaningful for your purposes. You would vary that restriction to look at other promising segments. --Matthias Winkelmann (talk) 12:48, 26 November 2020 (UTC)
Oh, thanks for pointing out my flawed strategy :) You are completely right that most statements is not suitable for my needs it seems. I really like your query as it seems to count the labels instead. I will try to use that and maybe a list with statistically most often used words in Esperanto which seems to be exactly what I want.

"merge" of two erroneous items to a new item?[edit]

The items Premil Petrović (Q53110639) and Premil Petrović (Q95770358) seem to refer to the same musical conductor, actually Premil Petrović according to his premilpetrovic.com website. (The c-cedilla usually takes the c-acute form when at the end of slavic language words).

As a neophyte, can I request someone else merge these/ sort this out please? Scarabocchio (talk) 12:54, 26 November 2020 (UTC)

done --Matthias Winkelmann (talk) 13:34, 26 November 2020 (UTC)

Thank you! That looks like a merge into one of the existing entries plus an edit on the label? (I should have tried that). Scarabocchio (talk) 14:16, 26 November 2020 (UTC)
Yes... The merge function should be in the top-right corner, possibly under "more". It will automatically merge into the item with the lower ID, and throw an error if there are specific indicators that the items are not, in fact, duplicates (such as "different from" statements). After that, I straightened out the labels per your explanation and also added the website you mentioned. I tend to also check for duplication of data and remove excessive "imported from / XY Wikipedia" references, but there's no specific process prescribed here.
As always: start with just a few (thousand). And, if nobody comes screaming, assume you're doing ok. Like childcare, basically. --Matthias Winkelmann (talk) 20:01, 26 November 2020 (UTC)
We do have https://www.wikidata.org/wiki/Help:Merge that describes merging in more detail. One thing that's worth noting is that the merge function is only available if the gadget is activated. ChristianKl ❪✉❫ 23:19, 26 November 2020 (UTC)
Thanks, both. I think that I will be spending more and more time here, so need to master the basic operations. Scarabocchio (talk) 20:31, 27 November 2020 (UTC)
  • Great. Merging items is indeed one of them. So, in general, I think it's preferable to help users find Help:Merge than merge them directly for them.
    Thanks to that, we recently revised it and collapsed some sections users sometimes got lost with .. if there are other parts that need clarification, don't hesitate to mention them. --- Jura 17:37, 28 November 2020 (UTC)
I enabled the gadget following the links above, and have just merged my first pair of items (to Sanna Gibbs (Q64682995)) .. very smooth, very clear, no problems! Scarabocchio (talk) 18:09, 28 November 2020 (UTC)

New external identifier[edit]

I would like to have a new external identifier added. How do I go about this (or preferably have someone who knows how to do this do it)? The identifier would be: ArtFacts ID. DoSazunielle (talk) 14:54, 26 November 2020 (UTC)

See WD:PP.--GZWDer (talk) 15:58, 26 November 2020 (UTC)

Comment about mass import[edit]

Hi, last week B.Zsolt did a mass edit on some railway stations, importing some data. I saw that he edited the item about the railway station in my small city adding toilets. I think that he imported mass data from OSM without checking whether statements are valid (toilets are always closed). Moreover, I don't think he can import data from OSM as license are incompatible. As soon as I checked that edit, I wrote in the user page of the user but he ignored my request, so I write here. --★ → Airon 90 08:57, 27 November 2020 (UTC)

I do not think we can mass-import OSM data--Ymblanter (talk) 20:57, 27 November 2020 (UTC)

US senator statements clean-up needed[edit]

There are dozens of subclasses listed with has part (P527) at United States senator (Q4416090). Would someone kindly delete these malformed statements?

Also, the still existing malformed and unused items like Q98082299 should be listed for deletion. --- Jura 19:38, 27 November 2020 (UTC)

Not sure why you reverted my move to the relevant section, but you seem to be starting a topic about the same thing. You're seem to be leaving out why you are not doing this yourself. Maybe because the deletion request got half done by user:Wiki13 and half not done by user:MisterSynergy? Just starting a new topic out of the blue sure looks fishy. Multichill (talk) 10:05, 28 November 2020 (UTC)
This is about the item United States senator (Q4416090).
The other section is about electoral districts called "class A Montana" and similar. (There is also one about general elections in the US).
I have been asked not to clean up such stuff myself, but post it to the forum. So here it is. --- Jura 10:16, 28 November 2020 (UTC)
I have deleted the remaining ones as another data model has meanwhile been chosen and implemented. —MisterSynergy (talk) 12:42, 28 November 2020 (UTC)
Thanks. As long as they are subclasses, the P527 statements are wrong in any datamodel.
Can you also delete the two remaining statements: e.g. United States senator (Q4416090) has part (P527) Majority Leader of the United States Senate (Q28530268)? --- Jura 12:48, 28 November 2020 (UTC)

Consistency of dates between point in time and temporal processes[edit]

Hello, it might have already been discussed. But I'd have a question : is there a way to tell that a instance of (P31) (eg historical event (Q13418847) is about a point in time and therefore instances of that P31 should be having a simple point in time (P585). In the contrary, if a instance of (P31) (eg any temporary exhibition (Q29023906) ) is about a temporal process (it has commenced and it has ended or shall be ended == 2 dates) and therefore should get both start time (P580) + end time (P582) ? Would property "timeframe of that element" = point in time (Q186408) OR time interval (Q186081) "do the job?Bouzinac 💬●✒️●💛 11:14, 28 November 2020 (UTC)

Two cats for one thing[edit]

Hi all. Item Q9032983 and item Q7696450 are about the same issue: Category:Universities in Turkey/Universities of Turkey/Turkish universities or whatever name you like. Look at the WP links and you will see that some language WPs use one of these items and the others the other one, although the contents in the respective languages are the same. I could not merge them. Probably some WP has links to both? I do not know. For some reason I feel myself not so good today; therefore help of colleagues would be most welcome. Thanks in advance. --E4024 (talk) 17:18, 28 November 2020 (UTC)

They're not the same things, at least according to their EN labels - universities, versus universities and colleges. Greek has sitelinks for both, calling one a category of universities, and another a category of higher educational institutes. Were someone to wish to improve the situation, they would have to work out sitelink by sitelink what was found on each langauge wiki, and variously redistribute sitelinks and possibly create new items. Merging is unlikely to figure in the solution. --Tagishsimon (talk) 17:28, 28 November 2020 (UTC)
EN labels? I just added EN label to one item that I saw there orphan as it was. Now am I the one who created a "differentiation" between the two? :) I had just opened a cat for Universities of Turkey in CA:WP and went to make the WD link and saw one item without any EN label and linked it. I also thought I should write a label there. Oh my! --E4024 (talk) 17:57, 28 November 2020 (UTC)

Need help removing aliases[edit]

Is there any automated tool that can remove aliases in bulk? I ran an OpenRefine job that accidentally added the item ID of several items as an alias (e.g. it added "Q151521" as an alias to Texas A&M Aggies (Q151521)). QuickStatements apparently doesn't support using "-Aen" as a header to remove an alias, so I'm at a loss as to how I might remove these bad aliases. Any suggestions would be much appreciated. Thanks, IagoQnsi (talk) 20:06, 28 November 2020 (UTC)

I have a pywikibot script which can do this. All I need is a list that maps the aliases to remove to QIDs and the language code of the alias to be removed (any txt based format and formatting possible). If you can provide such a list, I could feed my script with it and remove the aliases. —MisterSynergy (talk) 21:41, 28 November 2020 (UTC)
@MisterSynergy: Oh thank you! Here's a CSV of the IDs and aliases: https://pastebin.com/98WifNWW. Not all of those aliases actually exist -- my batch upload only affected a few hundred of the 922 instances of university and college sports club (Q2367225) that exist on Wikidata. Thanks, IagoQnsi (talk) 23:09, 28 November 2020 (UTC)
✓ Done —MisterSynergy (talk) 23:34, 28 November 2020 (UTC)

Inverse of academic degree (P512)[edit]

Q item academic degree held by (Q66499315) currently claims it is the inverse property label item (Q65932995) of academic degree (P512). If that was the case, it should be a P item, not a Q item. Am I correct? Should this be changed? Mateussf (talk) 20:44, 28 November 2020 (UTC)

You are not correct. Properties may get an inverse property label item associated with them, whether or not there is an inverse property. This is one of the 'not' cases. All is good; we do not really have a use for the inverse property, but a user might well wish to reach for the label. --Tagishsimon (talk) 22:33, 28 November 2020 (UTC)
I don't fully understand the concept but I appreciate the answer and the attention, and I'm glad I didn't do anything wrong before asking. Thanks! Mateussf (talk) 01:46, 29 November 2020 (UTC)
I made an item about the gadget. Maybe that helps: relateditems (Q102435390). --- Jura 05:44, 29 November 2020 (UTC)

Agriculture[edit]

We haven’t focused much energy on building out our coverage of agriculture. There are lots of opportunities for merges, multilingual labels, external ids, etc. in the areas of agricultural tools and processes. Is anyone interested in a formal WikiProject:Agriculture? If not, I’ll putter along. - PKM (talk) 22:01, 28 November 2020 (UTC)

Cleaning up old model for US Senate data[edit]

As suggested here by @MisterSynergy:, I plan on removing position held (P39) United States senator (Q4416090) which do not supply a term. The only information contained in such statements not contained in those which do supply a parliamentary term (P2937) was spotty and occasionally inaccurate. The statements which do supply a parliamentary term (P2937) contain the complete historical record (excluding the results of the most recent election). Still to do is to supply elected in (P2715) but that is pending an organization of election in the United States (Q279283) which should distinguish between scheduled general elections and unscheduled elections in a rational way and perhaps have a clear model to indicate midterm elections. I'll probably do this work at the end of next week if there are no objections. Regards, Gettinwikiwidit (talk) 03:21, 29 November 2020 (UTC) @Tagishsimon: @Oravrattas: @Andrew Gray:

So you want to delete the information about the current senators?
Can you provide us with a sample edit? How do you make sure that other data is complete?
As this provides us with a way of querying the current composition of the senate (e.g. position holder with a start date, but no end date), I don't think it's a good idea to delete these statements before you fixed the other statements (deleting fictive dates, as previously mentioned).
To sum it up: No data should be deleted before we can check the same information is available otherwise. --- Jura 05:29, 29 November 2020 (UTC)
current members. Gettinwikiwidit (talk) 10:45, 29 November 2020 (UTC)

New badges for templates and modules[edit]

Some templates and modules are developed in one wiki (Wikipedia) and re-used in other wikis. I think it could be interesting to note this on sitelinks of items with badges.

Sample: Module:Biblio is a copy from fr:Module:Biblio. Badges on sitelinks at Q102438737 could indicate it. --- Jura 08:51, 29 November 2020 (UTC)

MediaWiki:Villagepump-url/sv[edit]

Could someone recreate MediaWiki:Villagepump-url/sv with the text Wikidata:Bybrunnen and MediaWiki:Villagepump/sv with the text Bybrunnen as they have been re-established? --Sabelöga (talk) 09:53, 29 November 2020 (UTC)

Retrieved from "https://www.wikidata.org/w/index.php?title=Wikidata:Project_chat&oldid=1315107116"