This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.
From an OTRS message (Ticket#2018080110008255) - the category c:Category:Cala Mitjana, Menorca has the wrong map. The poster stated "The Cala Mitjana in Menorca is located at 39.9345N, 3.9728E, but the map that is displayed on the page shows the Cala Mitjana in Mallorca, at 39.7522N, 3.4139E." I zoomed out the map and it is indeed showing the wrong island. I'm not sure how to fix this one. Can someone help here? Thanks in advance. Ronhjones (talk) 18:02, 2 August 2018 (UTC)
Should now be fixed. There were two sets of coordinates, one good, one bad. I've deprecated the bad one. - Jmabel (talk) 22:28, 2 August 2018 (UTC)
Special:Nearby message "Wikidata cannot determine your location. Please try again with a better signal."
When trying to use Special:Nearby under Firefox, I have recently been getting this error message. Initially, I thought that this might indeed be a connectivity issue (I use the feature most commonly while traveling on trains) but then I noticed that nearby items display just fine under Chrome on the same machine. I had a look at the Firefox cookies for Wikidata but could not find any that would be the obvious culprit here. I also looked at Phabricator, where T125820 seems to be closely related but separate. Anyone got a hint what the problem is here, and how to fix it? Thanks, --Daniel Mietchen (talk) 01:06, 31 July 2018 (UTC)
Thanks for thinking along with me. I have checked the Firefox privacy settings, and I don't think it's an individual ISP since I have seen this behaviour in a number of locations in several countries. --Daniel Mietchen (talk) 17:43, 31 July 2018 (UTC)
I find it works if I'm logged into a Google account, and doesn't work if I log out. Presumably because it's using a Google API. Ghouston (talk) 02:16, 1 August 2018 (UTC)
Although by "works", I'm getting a location a few hundred kilometres away that I visited last week, which was probably the last time I logged into the Google account on a phone. Ghouston (talk) 02:18, 1 August 2018 (UTC)
Needless to say, due to disagreement about whether atheism is a religion or not, some items are coded with religion = atheism (violating a constraint), and some are coded with religion = <no value>. Whether atheism is the same as absence of religion can also be debated. There's some discussion at Property_talk:P140 which I haven't read. Ghouston (talk) 00:29, 31 July 2018 (UTC)
People's opinions are not generally recorded. If somebody is known to be a member of a particular political party then there's a statement for it, but if they just wrote on a blog somewhere that they supported a particular candidate, there's no place for it. For religion, if somebody has publicly stated that they don't practice any religion, then religion = <no value>, but perhaps there's no need to record that they say they are an atheist unless in the context of joining a particular atheist organization, or being active in a particular movement. Ghouston (talk) 00:45, 31 July 2018 (UTC)
Whether or not to treat atheism as a religion is very complicated. Summarizing some relevant points:
Not all religions are belief-based, or even have associated beliefs. There are millions of atheists who are religious and millions of theists who are irreligious. See Nontheistic religion, Spiritual but not religious on Wikipedia. Atheism is neither synonymous with nor a subcategory of irreligion.
Atheism, strictly defined, is exclusively a belief, like monotheism, agnosticism, polytheism, etc. Formally, it has no further communal, ritual, broader philosophical, moral, organizational, or social aspects. As a result, many would further hesitate to categorize it as a religion.
Some definitions of "religion" are extended so far as to include all beliefs regarding spiritual matters. This would thus include atheism.
In certain societies, "atheist" as a designation has been placed firmly within the scope of the category of religions. The question "What religion, if any, are you?" could be answered by "I'm an atheist", without it seeming like a complete non-sequitur at all. This is not simply because in those societies all religious people are theistic and all irreligious are atheistic; in some of the same societies, "none/nothing in particular", "atheist", and "agnostic" are all common answers that are considered very different. Atheism is often taken to be a separate religious category by both its adherents and others.
All proposed solutions involving label changes have substantial consequences: "religious affiliation", "religious stance", "religious position", "religious attitude", "attitude toward religion" all exclude certain things which could be categorized as "religion".
"Atheism" is defined very differently by different people, debatably including, for example, agnosticism. See w:Atheism#Definitions and types.
To confuse things even more, there's only a moderate level of overlap between any of the conventional definitions of atheist and how people actually use the term as identification. In the United States, only one third of people who don't believe in G-d identify as atheists, and 8% of self-identified atheists believe in G-d. (See Pew research, point 4.)
Sources often do not state that "the person's religion is X", they just say "the person is X", so we can't just push the decision to the sources on a case-by-case basis.
I'm leaning weakly towards counting it as a religion. We've done similar things in the past for purposes of convenience of categorization. --Yair rand (talk) 00:00, 1 August 2018 (UTC)
That's a good summary of the problems. There's also the point that people's religious opinions can change over time, and they can say different things in different interviews. point in time (P585) qualifiers may often be needed. There's always the concept adopted for languages spoken, written or signed (P1412), just expand the label to encompass all the possibilities (I think Andy Mabbett already knows about this one). Ghouston (talk) 02:03, 1 August 2018 (UTC)
We could have a property for "supports" and one for "opposes", to allow the listing of all kinds of random things. Perhaps we'd know what football team somebody supports or who they voted for in 2006. It's not much different to listing religious views. Ghouston (talk) 07:05, 1 August 2018 (UTC)
Book title not italicized and given in English, not French?
Also one will need to define "standard" representation rules for your wiki. Because in Russian it's author name to be italicized, not book title. -- VlSergey(трёп)08:47, 31 July 2018 (UTC)
@Pigsonthewing:, goal for both templates is the same, but current implementation of "cite q" seems too simple for ruwiki. It doesn't format author names (according to rules about short-long notation), it doesn't check "parent" properties (for example, it doesn't retrieve ISBN from "parent" book or ISSN from "published in" journal, AFAIK), it doesn't print publication place name at all. Compare:
cite q: Vladimir Filimonov, "Антиковеды — авторы «Энциклопедического словаря Брокгауза и Ефрона» в коммуникативном пространстве Н. И. Кареева", Q20807896, WikidataQ20807885
(why author name is in English? It's Russian article; no journal name; no pages; no ISBN info)
From theory point of view one can just link cite q template and source template in ruwiki. As soon as source template clone is deleted from enwiki, of course. -- VlSergey(трёп)12:08, 1 August 2018 (UTC)
Thanks for the ping. I didn't post over on the Hawaiian WP, since it looked pretty much dead and the last post on that page was in 2014. I asked @T8_Expresso: on their WD talk page for help, since they were active on both projects recently, but got no response so far unfortunately. --Kam Solusar (talk) 00:48, 2 August 2018 (UTC)
Proposed property for Instagram tags
Wikidata:Property proposal/Instagram_tag was closed as "not done", because "there were no examples provided of official sources endorsing instagram tags". Such endorsement is not a requirement for property proposals (and evidence of it was not even asked for, in the discussion). Furthermore, there were more supports than objections, with the three objections being stated on the basis of the data-type, rather than objections to a property in principle. None of the supporters objected to the proposed datatype.
For those reasons, I have asked the closer, User:Micru, three times to reopen it, but they have refused, expressing his own partisan view on the datatype. The proposal should be reopened forthwith, so that it can be reviewed by someone else, with a neutral approach and without bogus criteria. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits14:22, 10 April 2018 (UTC)
I will agree with you that this was closed prematurely—not saying that it should be created at this point, though. I will reopen this within the next two hours, @Pigsonthewing:, once I get to a computer. Mahir256 (talk) 14:41, 10 April 2018 (UTC)
@Pigsonthewing: As you have said below, this discussion has not been resolved. I am thus respecting Micru’s request that a clear outcome be achieved here first before reopening the proposal. (To be clear, if no admin had asked me to hold off on it within the two-hour window, I would gladly have reopened it then.) Mahir256 (talk) 19:51, 19 April 2018 (UTC)
Question given the various comments of the participants, what outcome would you consider appropriate? --- Jura14:41, 10 April 2018 (UTC)
There were legitimate concerns raised about this property proposal, mainly that Instagram tags are not curated, not sourced, and not endorsed by anyone, and as a result I believe this should not be used as an external identifier. I suggested Andy to start a new proposal with data type "string", same as property hashtag (P2572).
The discussion was stalled since January, and while the number of support/opposition can give an indication about the general perception of a property, it is never enough to justify by itself the creation of a property. I feel attacked by Andy questioning my integrity as property creator just because the outcome was not in his favor. Micru (talk) 14:50, 10 April 2018 (UTC)
I am again restoring this unresolved issue from the archives; absent agreement here, the property proposal should be reinstated so that debate can be concluded there in the proper manner. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits11:31, 25 July 2018 (UTC)
Question given the various comments of the participants, what outcome would you consider appropriate? --- Jura11:37, 25 July 2018 (UTC)
(somehow I thought only admins were property creators back in April...) The issues with the tag properties we have proposed are IMO not at all specific to Instagram—in other words, at this point reopening this proposal does not in itself help resolve these issues—and given that the current active tag proposals pertain to Gfycat and Flickr, these issues ought to be resolved on either of those proposals as long as no property creator has closed them. (@Micru, ArthurPSmith, Pintoch: If you decide to close either of the current active proposals at any point, please start an RfC with respect to tags before closing them so that discussion about these issues can continue more generally.) I would like to adapt what I said on the second of the current active proposals more generally here, though: until we all agree to recast hashtag (P2572) as an external identifier, I generally support either property's creation as a string; on the worthiness of these properties' inclusion, I sometimes wonder, given many of the examples presented on those proposals and elsewhere, whether a single unified 'tag' property is in order, seeing as most of them pertain to exactly the same topic whether viewed on Twitter, Instagram, Gfycat, or Flickr. Mahir256 (talk) 14:38, 25 July 2018 (UTC)
There are 21 open RfCs, and I doubt that it would help opening a 22nd. Personally I am interested in knowing Andy's answer to Jura's question, as this seems to be a topic that bothers him for some reason, and I would like to understand better his reasoning.--Micru (talk) 18:56, 25 July 2018 (UTC)
"examples [...] of official sources endorsing instagram tags" is not a requirement for property creation. You should reopen the proposal which you improperly closed stating lack of them as you sole reason for doing so. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits20:55, 25 July 2018 (UTC)
@Pigsonthewing: As it has been explained in other places there are more reasons for not creating tag properties. Tags are not curated and include many elements that do not belong to the category, and they do not desambiguate concepts. For instance "train" has a mixture of trains, people training, and images totally unrelated. Even basic concepts like "river" contains mostly things that are not a river. Do you really want to keep persevering on this?--Micru (talk) 09:48, 26 July 2018 (UTC)
As I have explained before; you are entitled to have your opinion on the merits of the proposed property. You are not entitled to "super vote" and close the proposal based on those opinions; much less using an invented rationale. Reopen the proposal. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits13:40, 26 July 2018 (UTC)
As it is now there are no guidelines regulating the closure of property proposals, so it is left to property creators to decide how, when, and why to close a proposal. Of course this situation can be changed in future proposal closures, as you can propose how or under which circumstance a proposal can be closed in the corresponding RfC.--Micru (talk) 15:55, 26 July 2018 (UTC)
Are you arguing that you are entitled to "super vote" and close a proposal based on your personal opinions, or to impose arbitrary criteria like requiring "examples of official sources endorsing instagram tags", or both? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits17:30, 26 July 2018 (UTC)
I'm saying that "As it is now there are no guidelines regulating the closure of property proposals, so it is left to property creators to decide how, when, and why to close a proposal." If you feel that additional criteria are needed, please feel free to propose them.--Micru (talk) 20:37, 26 July 2018 (UTC)
The proposal needed to be closed anyways. If it's just to be closed with the same outcome, you are just wasting everyone's time. Try to formulate a constructive conclusion for the discussion based on people's input. --- Jura12:02, 27 July 2018 (UTC)
Hello.I suggest a new property for this site to be suggested by someone who is expert in French language and the field of games.Thanks --David (talk) 09:40, 31 July 2018 (UTC)
Comment For which elements on that site would you want a property? The only elements I see are the character (some ~100 of them), but I don’t think this warrants a property. Jean-Fred (talk) 09:59, 31 July 2018 (UTC)
Then what is the need for property, you can create the page for yourself. I also play Tekken but didn't find anything unusual on the page after which you've named this discussion. There is no sense of creating any property because I fear there are no more than hundred characters in the entire roster of Tekken franchise. Harsh Rathod Poke me!17:58, 31 July 2018 (UTC)
I've noticed a surprising number of people with country of citizenship (P27) countries that did not exist during their lifetimes. United Kingdom (Q145) came into existence in 1927, but there are nearly a thousand results for individuals with that country of citizenship (P27) with death years before 1927. (This query could be refined to take account of death dates with century-level precision).
@MartinPoulter: I know for a fact that @Bodhisattwa: had fixed most, if not all, of the India cases; whether or not this preserved appropriate qualifiers and references I'm not as certain of (a quick look at the relevant contribution history suggests that if anything was lost, it was most often a imported from Wikimedia project (P143) qualifier), but in any case bulk edits are definitely in order. Mahir256 (talk) 21:01, 31 July 2018 (UTC)
I've amended a few tens of thousands of UK P27s in recent times, and opened a discussion on talk P27, here. Depressingly though there is not consensus that the job should be done at all - people like Ghouston seem to want to confect a dummy UK to cover the various UK states, which for me makes no sense whatsoever and is an insult to everything that I think wikidata is about. iirc, where I left it, was the set of (wrong) UK P27s whch had references, since the job of moving the citizenship to the appropriate value without losing the reference information, takes that much longer. smh. --Tagishsimon (talk) 21:11, 31 July 2018 (UTC)
I find it highly dubious to claim that the UK pre-1927 and post-1927 are completely different states with different citizenships. There's more continuity than discontinuity. The sitelinked article en:United Kingdom doesn't even mention 1927. But I said that already at the other discussion. Ghouston (talk) 21:44, 31 July 2018 (UTC)
There are plenty of statements that I do find dubious. Modifying the query above shows that there are 4293 people who died before 1949, but somehow obtained Australian citizenship, which wasn't created until 1949. Cases of anachronism (Q189203)? Ghouston (talk) 22:43, 31 July 2018 (UTC)
When, Ghouston, your best evidence that 'nothing really happened here' is a badly written wikipedia article, and your major contribution is to say 'I can't be bothered to work out for which period my confected UK should cover' ... I think we're entitled to weigh your argument and find it wanting. It is incontrovertable that the legal name of the state changed in 1927, per w:en:Royal and Parliamentary Titles Act 1927. It is incontrovertable that UKians extant in 1926 were citizens (or subjects) of United Kingdom of Great Britain and Ireland (Q174193). They even had their own passports - File:UK passport 1924.JPG - confirming the country of which they were citizens/subjects. It is incontrovertable that UKians extant in 1928 were citizens (or subjects) of United Kingdom (Q145). They had passports, too - File:UK Passport Pre 88 change.jpg - confirming the country of which they were citizens/subjects. It is incontrovertable that UKians born after 1927 were not citizens of United Kingdom of Great Britain and Ireland (Q174193) because a state of that name did not exist in their lifetime. It is incontrovertable that UKians who died before 1927 were not citizens of United Kingdom (Q145) because a state of that name did not exist in their lifetime. Would it be all the same to you if we went with facts like these rather than your gut feeling? --Tagishsimon (talk) 23:13, 31 July 2018 (UTC)
I don't consider a change to the name of a state to be the same as the creation of a "new" state. There's more continuity than discontinuity. Ghouston (talk) 23:35, 31 July 2018 (UTC)
By the way, even if the enwiki article is "badly written" for recognising a UK that dates back before 1927, we still need a wikidata item to sitelink it to. Either United Kingdom (Q145) needs an earlier inception date, or the article should be sitelinked to a different item. Ghouston (talk) 23:43, 31 July 2018 (UTC)
I'm thinking the population of the 32 counties of Ireland probably thought it was more than a name change (even if the name change took 5 years). I'm thinking there is some point to Q145#P571 and Q174193#P576. You don't have a whole lot going for your argument beyond "I don't consider" and "I'm dubious". I'll leave it here: if you are not going to merge Q145 and Q174193, then I'm going to presume that we have items for a couple of time-limited states, and continue to work towards person items having P27s that are logically consistent with the existence of Q145 and Q174193. --Tagishsimon (talk) 14:43, 1 August 2018 (UTC)
@Tagishsimon: I think Ghouston has a point. The government of the United Kingdom of GB/NI in 1928 is the same entity as the government of the United Kingdom of GB/I in 1926, other than in name and in territory. 1927 is also a somewhat arbitrary cutoff point, since it is based on the Act of Parliament that renamed the country rather than the territorial change. I don't know whether it would make sense to use United Kingdom of Great Britain and Ireland (Q174193) as an item for "the UK between 1801 and 1928" and United Kingdom (Q145) as "the UK, the government created in 1801"; this would be useful for former countries where there is a legal successor state, such as Serbia and Montenegro (Q37024) (which would then be "Serbia between 1992 and 2006"). Either item could then be used for country of citizenship as appropriate, since both would be technically correct for those who – in the case of the UK – were born there between 1801 and 1928. In any case there should be some way of modelling this relationship. Jc86035 (talk) 15:12, 1 August 2018 (UTC)
Yeah. If you ignore Ireland. If you ignore the absence of constituencies returning MPs to the UK parliament as a result of the 32 counties. If you ignore enough stuff, you can for sure find some way to 'model this relationship', to solve a problem which does not exist. --Tagishsimon (talk) 15:18, 1 August 2018 (UTC)
And really. Why stop at 1801? Kingdom of Great Britain (Q161885) is up for grabs, because what, really, was the difference? Another Act of Parliament? A few more or less Irish constituencies - and who cares about them? That gets you to 1707. Or Kingdom of England (Q179876) because the difference was just Scotland, and we're all the way back to pre-Conquest times. Yay. "Problem" solved. --Tagishsimon (talk) 15:27, 1 August 2018 (UTC)
(I have edited my previous comment to add strikethrough tags.) Of course, this all depends on what we define the existence of a country as being. I'm assuming that a territorial change (e.g. the annexation of Crimea; Belgium–Netherlands "land swaps") or a name change does not cause all countries involved to cease existing. According to the current ontology the Irish Free State split from the UK of GB/I in 1922, and then the UK of GB/I continued existing until 1927, when the Act of Parliament to change its name was passed. I think this is logically inconsistent. Jc86035 (talk) 15:32, 1 August 2018 (UTC)
It doesn't seem to me that this change in the territory ruled necessarily makes for a different entity in terms of a government. We consider the U.S. (and its government) to have continuity at least back to the Constitution, and generally to the Decalaration of Independence, even though there were originally 13 states and are now 50 plus assorted overseas territories, and because in the areas that were part of the Confederacy in the Civil War, effective rule by that government is not continuous. - Jmabel (talk) 15:52, 1 August 2018 (UTC)
Well. Good luck with coming up with a workable set of rules which can be applied given the succession of 4 named states we have right now - KoE, GB, UK+I, UK+NI. Each succession is marked by the same three things - acts of parliament, name changes, territorial changes. Jc86035, I'll tell history, when I see it, that you're not happy with how it conducted itself. Meanwhile, as I said, I'll continue with the logic of current four items we have as the basis for P27s. Ping me when you've effected a material change. --Tagishsimon (talk) 16:09, 1 August 2018 (UTC)
@Tagishsimon: I'm not trying to disagree with history. How we represent this in the data structure is an entirely different question. If a structure which isn't perfectly rigid is used for defining the creation of a historical country (Q3024240) in the data, why should the date be 1927, when the obvious material changes occurred earlier? Jc86035 (talk) 16:37, 1 August 2018 (UTC)
┌────────────────────────────────────────────────────────────────────────────────────────────────────┘
Just throwing this out there - I think we have this problem because we have not come up with a way to handle "nationality", so editors use "country of citizenship" as a proxy for "nationality". I believe we need "nationality" in a fuzzy and general sense, across time, so that we can say "British author, Russian poet, Italian artist" as found in our sources. And I despair that we'll ever get consensus on this issue. - PKM (talk) 00:27, 2 August 2018 (UTC)
@Steak: There are plenty of completed catalogues that remain on mix'n'match for others to view; as far as I have seen @Magnus Manske: generally deletes a catalog only if someone requests it due to a problem with the imported catalogue entries. Mahir256 (talk) 14:53, 1 August 2018 (UTC)
@Steak: Keeping the matched entries in the system can actually be quite useful. For example, if one's pushing a list of names through M'n'M search, it's very helpful to see all the different Wikidata items in different MnM catalogues that have been matched to that name, to give an idea of the sort of different possibilities for who it might be. Jheald (talk) 12:47, 2 August 2018 (UTC)
If I were being strict, since "Charity Navigator ID" = "identifier for a charitable organisation in the United States, in the Charity Navigator database", I would create a separate item "Charity Navigator database" <instance of> database, and use <stated in> "Charity Navigator database" qualified with "Charity Navigator ID". - PKM (talk) 00:16, 2 August 2018 (UTC)
Is it desirable and advisable to store the shares of important reserve currencies as in this table in wikidata? Advantage would be to generate the charts from that table. If yes, how is it done? --Rabenkind (talk) 07:49, 2 August 2018 (UTC)
For tennis players, most language versions of Wikipedia have a list of tournament wins and some even of all lost finals, i.e. runner-up placements. For reference purposes, please see examples for Serena Williams:
Even though different in styles and overall presentation, there is a common set of information displayed in all versions - and this just calls for a use of Wikidata!
Generally, the following information are included which may be represented by the following structure (example: Angelique Kerber winning Wimbledon):
Of course other details my be applicable for other kind of sports, but I hope the main idea is clear.
I am by no means an expert in Wikidata, but I see a lot of potential for automated lists/templates as shown for Serena Williams above. This is especially important for less known players that play on 2nd tier tournament tours and have a lot of wins there, but are rarely updated in all wikis. Therefore, I am seeking your feedback on the following:
Thanks for your response, but this is not really what I am looking for. As mentioned in my intro, I would like to be able to create the lists of tournaments won (or being runner-up) by a certain player. Therefore, I am basically looking for the inverse element to what you described in order to be able to maintain the information more easily and more central at the player level rather than the tournament level (side note: of course we will also try to create such lists automatically as well, but this would be something like https://en.wikipedia.org/wiki/List_of_Wimbledon_ladies%27_singles_champions in contrast to the links given in my original text). Also, I am aware that we derive a couple of information from the tournamament, but this is already a detailed look.
@Unnited meta:Thanks again for the continuing feedback. However, after a little bit of thinking, I don't share your views of creating a new item per match played. This idea has been formulated before (see https://www.wikidata.org/wiki/User:Sannita/Matches), but has some flaws:
If we wanted to pull the information the way presented in the various Serena Williams examples by way of a wiki template, we would have to run very expensive queries on Wikidata every single time a tennis player page is looked up at a given wikipedia version, because we basically had to look for every match played by a specific player in the whole of Wikidata. I don't think that this is the way to go here, but I am not a data expert.
The match-based system you are proposing is very difficult to maintain from a single players perspective. As an example: if I wanted to include a template for tournament wins and runner-up placements in a given wikipedia version, I would check the wikidata page of that player looking for completeness of the data before I included the template. This would be difficult to look at because either (i) I would have to be a WD expert knowing how to run the queries that provide me with the respective information I want to have (and I can assure you that most people writing about tennis, including me, are not) or (ii) I would have to include the template first and then create the matches that are missing.
As long as the information on WD are not complete (and we haven't even started on collecting them), we can only activate above-mentioned templates on a player-by-player basis once completeness (at least to respective point in time) is confirmed. Adding the information required could be done at a player-by-player basis way easier with the approach I am proposing.
On a side note: we have many instances of inverse elements that store the same information from different POVs.
All of those issues are addressed with the setup I am proposing. Lastly, let's look at a potential "endgame" for tennis information on WD which would be being able to fill all of the tournament draws with WD information rather than having all wikipedias inserting them individually (as an example, look at https://en.wikipedia.org/wiki/2018_Wimbledon_Championships_%E2%80%93_Women%27s_Singles and all (!) of connected languages which all portray the same information!). I couldn't find the numbers of matches played every year on the WTA and ATP World Tours (and let's not forget about the 2nd tier series and the ITF series that are also covered by wikipedias), but only for the four grand slams, we have 127 ladies' single matches, 127 men's single matches, 63 ladies' doubles matches, 63 men's double matches and 63 mixed doubles matches, i.e. 443 matches per event per year, i.e. 1,772 matches per year. This does not include the wheelchair, junior and senior competitions. Please note that Wimbledon runs for more than 130 years... Another example for a rough estimate of matches played per year: in WTA there are 52 tournaments excluding Grand Slams, the two WTA Final Events FedCup/Hopman cup. If we assumed fields of 32 players per average event (which is actually more) and assumed 16 players in qualifications with 4 qualifiers (which should be a good average), we will have 31 main draw + 12 qualification matches in singles and 15 matches in doubles, i.e. 58 matches per event on average. This would make over 3,000 matches per year for the highest tier of women's tennis alone, excluding tier 2 (definitely relevant) and tier 3 (mostly relevant) and all of the men's events. I don't think that it is a feasible approach to have somewhat 10k new Q-items per year (!) only for tennis.
Still, I am happy for the discussion and would embrace any additional feedback provided by everyone in the community. --Mad melone (talk) 06:42, 30 July 2018 (UTC)
@Mad melone: I agree with Unnited meta that you would need to have one item for each match. Matches can be notable in their own right (e.g. Isner–Mahut match at the 2010 Wimbledon Championships (Q30801)) so it would probably make more sense to use the existing data structure, and perhaps have a bot generate data tables on Commons in some way so that the data doesn't have to be reprocessed every time a tournament or player page is viewed. Jc86035 (talk) 04:42, 31 July 2018 (UTC)
┌────────────────────────────────────────────────────────────────────────────────────────────────────┘
@Jc86035:Point taken. I think the following structure could work
results (P2501) -> Runner-up (currently in creation process)
Then we would have to think of the result. Options are a text field where you could enter something like "6:3, 6:3" (or any other agreed format) or describe it as follows:
match time of event (P1390) -> First set (Q-ID to be created for at least 5 sets and restrictions for P1390 have to be lifted or new Property has to be created)
@Mad melone: I'm not sure. match time of event (P1390) only accepts time units right now, and games aren't strictly "points". I've never edited much in this area, but any structure would have to leave open the possibility of adding data for individual points in some way (because eventually someone will want to and be able to import that), probably through Commons data tables. Allowing the match score to be input as "6-3, 6-3" (with agreed syntax for tie breaks, walkovers and so on) seems like the simplest thing to do right now, and if desired in the future that data could be interpreted into a more structured format. Jc86035 (talk) 09:56, 1 August 2018 (UTC)
According to the English-language Wikipedia, Golden Potlatch (Q5579718) existed 1911-1914 and 1935 (or possibly 1934, the article contradicts itself on this)–1941. I can find no reasonable way to express this, and Help:Dates certainly does not cover such a case. - Jmabel (talk) 04:46, 30 July 2018 (UTC)
In the more general case this sort of thing usually boils down to a question as to whether it was actually the same thing during both time periods, or whether each could be considered a distinct enough concept to have a separate item for each, with their own inception/abolition dates. In the first case, you could use significant event (P793) to show that it was temporarily suspended, and then again for being reinstated, or, as this was a periodic event anyway, perhaps have linked articles for each of the individual occurrences of the festival. It would probably also be worth looking at the model at Wikidata:WikiProject Cultural events, and/or asking there about this. --Oravrattas (talk) 08:06, 30 July 2018 (UTC)
One option is this; if you wish to keep it, please add sources. Alternatively, create two "child" items, add one set of dates to each, and make them, say, "part of" the existing parent. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits13:22, 30 July 2018 (UTC)
@Pigsonthewing: I reverted that, because I think it is more confusing than helpful, but thanks for trying.
A single item may be enough, and would work better with multiple periods of dormancy. There's already an item dormancy (Q162267), although I don't think it's appropriate because it applies to biological cycles. But perhaps something similar exists, or could be created. It would be like
Editing Wikidata, I’m sort of challenging myself on how much to curate data and I’m seeking input on this – and maybe even a reference to some policy? Let me give two examples for you to comment on.
As it takes some time to register a company in Denmark and as the registrations are publicly available, it’s quite common that is you want to create “Great Company” and go public on January 1st, then a few things happen in advance.
A lawyer will during the fall register a company called “Lawyer XYZ company 34” and instate himself as CEO and put a number of his employees on the board. Then by January 1st, he will change the organisations name to “Great Company” insert the true founder of the company as CEO and instate the real board of “Great Company”.
When I source data from the national company register and add verbatim to Wikidata, it seems that “Great Company” was founded a few months before January 1st, that some random people has been on its board and that this lawyer guy has been heading the company. As I try to provide the “Great Company” story, I then tend to leave out these data, as they exist as technicalities and not as indications of the “Great Company” true origin.
In another case, I want to convey the fact that “Great Company” has bought a small Danish Island, which was a story that made the headlines. However, in Denmark, an Island cannot be owned by anybody.
In practical terms, an island consists of one or more pieces of land, registered in the national Cadastre. So, seen from a data perspective, what I can source is a number of ownership changes for each of the pieces of land that constitutes that island, but only by understanding the specific structure of that island and combining all of the ownership changes can actually lead to the insight that “the island has changed ownership”.
So, I end up putting an ownership statement on the island, pointing to the owner and thus representing the fact that I want to publish, rather than creating an elaborate structure with items for each of the land pieces, etc.
So, please help me to understand if my actions represent an acceptable label of curation or if I’m completely off. --VicVal (talk) 12:18, 2 August 2018 (UTC)
I think what you're describing is the distinction between legal definitions and common human understanding - in general I would favor Wikidata hosting statements that are close to the common understanding of people, rather than whatever the precise legal situation is. However, there may be reason to have both - and Wikidata does allow for that, with appropriate references, qualifiers, etc. ArthurPSmith (talk) 13:26, 2 August 2018 (UTC)
That's actually a good point - that i can create the correct legal statements that are clearly sourced and then do statements which are inferred from these!? --VicVal (talk) 15:11, 2 August 2018 (UTC)
Only one column (text) is common to both tables, however is not an exact match. There is just one or two words that make a link.
I need an easier method where I can link the two tables on a monthly basis
Table 1 has product specific data from suppliers, with cost of each product
Table 2 has product specific data from customers, with sale price of each product and QTY sold
If you can help in any way, ie a software or website that converts the data
Help is much appreciated, – The preceding unsigned comment was added byAkhan1984 (talk • contribs) at 9. 8. 2018, 00:37 (UTC).
Property for relation Church parish and administrative parish
I think I have seen a property for connecting a Church parish with a corresponding "administrative" parish but cant find it... Anyone who can help? - Salgo60 (talk) 15:43, 17 July 2018 (UTC)
Thanks for trying - maybe I just saw it in proposed properties
It is different WD objects that should be connected using this property
The basic idea is in Sweden we had civil parish or a administrative parish socken (Q1523821) then in parallell we had Parishes of the Church of Sweden. As those administrative units are related this property should make this connection...
PmtFYI: We are cleaning the data in Wikidata for Swedish "församlingar" and in the next step we will see how we can integrate with a project from the Swedish National Archive called TORA that is a Topographical register concerning historical settlements in Sweden using Linked DATA see projekt and property P5324
When "cleaning Wikidata" I can see that if you are in a location and in the parish of church like you look in the books "kyrkböcker" then many resources created are connected to the administrative parish "socken" ==> somehow it would be nice to "walk" from the "församling" -> "socken" -> resource
As P131 is only for the governmental hierarchy, it seems there should be a more general property which then can be used for the ecclesiastical hierarchy - the various Christian churches, but also e.g. Thai Buddhism has such structure. And there are also other territorial hierarchies, e.g. military command. Similar like we have the catalogue property for miscellaneous catalogues, and some specific properties for special catalogues. Ahoerstemeier (talk) 13:36, 27 July 2018 (UTC)
I will suggest a property that connects a civil parish and the related church parish. Please help me to formulate this in good english '
Name: Related parish
Description: Property to connect the administrative parish with the church parish and vice versa.
Motivation: In Sweden we have "socken" civil parish and "församlingar" church parish. As some properties are just connected to the civil parish and some to the church parish it is useful to connect the related parishes so that you easy can find items that are related
I suggest two distinct properties, one in each direction.
Also, you might want a broader notion than just "parish," since (for example) a Catholic diocese could have similar considerations.
Also, you will want to think about whether you intend this only when the correspondence is 1-1, or whether you are interested in things where there is an overlap but not an exact correspondence. - Jmabel (talk) 16:07, 3 August 2018 (UTC)
Thanks for the feedback. Anyone with a better name? The normal case is 1-1 but as always there are exceptions over 400 year and the last years the church parish is more "unstable" - 20:42, 3 August 2018 (UTC)
@Pmt: that is my idea BUT this is for older days. The Swedish Districts is a new construction that we really dont know why we have ;-) (my opinion). The main interest right now is to navigate the Swedish parish we have had since 1600 and has been very stable to the Swedish Church parish "församlingar".
Asked about the Nobel nomination deletions again, Pierrette13 says (according to google translate) "As for the "almost Nobel", obviously I will not go back to reinsert, it's great anything in my opinion. Can you tell me the list of others 'almost Nobel' and I will make a group protest, thanks to you".
I'm at a loss to know how to deal with a user who seems, to me, to have a thing for removing data on a whim. It's discouraging to see wanton destruction of people's work. @Pierrette13:. --Tagishsimon (talk) 22:07, 17 July 2018 (UTC)
I'm relatively new here, so perhaps I don't understand Wikidata's criteria. To be on Wikidata, doesn't something have to be citable and notable, not merely citable? - 04:26, 18 July 2018 (UTC)
Anyone can be nominated by the "thousands of members of academies, university professors, scientists, previous Nobel Laureates and members of parliamentary assemblies", but there is an official short list maintained by the Nobel Committee that is made public 50 years after the award is given out. I see no reason to exclude the data. --RAN (talk) 19:45, 18 July 2018 (UTC)
A "short list" is probably notable. But the full list? There were 330 official nominations for the Nobel Peace Prize this year. I guess that's tractable, if we really want it, but a lot of it has to be like the sort of candidate for the U.S. presidency who gets 200 votes. - Jmabel (talk) 03:53, 19 July 2018 (UTC)
The Nobel Price seems to me a notable prize where the nominations are of general interest. It's valuable to be able to ask question such as how many nobel price nominations citizens of a particular country got in a certain year. Wikidata allows the same with alumni of specific universitites. Wikidata might be the only place that gives you an answer if you want to know how many Harvard alumni got a Nobel Price nomination in 1961. The idea that question like that aren't of interest to anyone and thus shouldn't be able to be answered seems mistaken to me.
Apart from that we have a rule that items have to be notable with our own definition of what we mean with notable with is a lot broader than the Wikipedia version. I'm not aware of any Wikidata policy that says that individual statements need notability just that they need trustworthy sources. ChristianKl ❪✉❫ 07:25, 24 July 2018 (UTC)
I would like to see that we set up federated searches with the Nobel Price database and could create queries in Wikidata and in the Nobel Price database - Salgo60 (talk) 20:55, 29 July 2018 (UTC)
How to set up a Feederated search Nobel Price SPARQL -> Wikidata
Ethnicity and religion should have at least one reference
"Statements for religion should have at least one reference" the same for ethnicity. Does that mean we delete the field and the current data if there is no reference currently? Someone is making a large number of deletions and I want to see if consensus was formed to delete the information, if currently not referenced. I know they are deleting them in the English Wikipedia but I have not seen a discussion here. The person deleting them has been warned twice about making mass deletions at Wikidata based on rules set at the English Wikipedia. Surly Eddie Murphy's daughter is African-American and a reference can be found. We really should not be deleting data at Wikidata because it is being deleted at the English Wikipedia. We went through this before when they mass deleted links to Findagrave at English Wikipedia and started deleting them at Wikidata as an "unreliable source". See Special:Contributions/Nikkimaria for examples of the mass deletions. --RAN (talk) 23:20, 29 July 2018 (UTC)
If you want to find a source and re-add, by all means do so. However, as the description for the property states, "consensus is that a VERY high standard of proof is needed for this field to be used. In general this means 1) the subject claims it him/herself, or 2) it is widely agreed on by scholars, or 3) is fictional and portrayed as such". This is particularly important for living people. See also relevant discussion here. Nikkimaria (talk) 00:47, 30 July 2018 (UTC)
I see a discussion between five people, I do not see a formal consensus for mass deletion. You were twice warned about getting a formal ruling before you start mass deletions. If a formal consensus to delete is acknowledged, a bot will make the deletions for ethnicity and for religion. --RAN (talk) 01:39, 30 July 2018 (UTC)
We already have consensus that it is not appropriate to have these properties without sources. Note also that both religion and ethnicity are considered "likely to be challenged" for the purposes of Wikidata:Living people. Nikkimaria (talk) 02:35, 30 July 2018 (UTC)
Why are you deleting one here and one there on an ad hoc basis? You did the same thing when you deleted Findagrave when it was used as a reference. If you established consensus for deletion, we delete them all at the same time. Perhaps we keep them and make them unviewable by the general public until they are sourced. There are several possible solutions, all requiring firm consensus, not a discussion between a few editors. Remember, you were warned twice before about doing this sort of thing, and each time you just moved to a new data field and started deleting again. Consensus at the English Wikipedia does not apply to Wikidata. Wikidata serves all language Wikipedias, and consensus must be established here. --RAN (talk) 05:48, 30 July 2018 (UTC)
I'm not sure that it would be technically feasible or desirable to have claims that are unviewable by the general public. That leaves 2 options: a) allow the unsourced statements b) delete them. There could also be an option to add references to them all, but it would presumably take a long time, and there's no guarantee that anybody would do it, which means in practice it would be similar to a). Ghouston (talk) 06:56, 30 July 2018 (UTC)
Again, we have already established consensus on Wikidata that these fields must be sourced. This has nothing to do with what enwiki does or doesn't do. If you'd like to propose a bot to remove the unsourced values, by all means go ahead, but at this point AFAIK no such bot exists. Nikkimaria (talk) 11:32, 30 July 2018 (UTC)
"Must be sourced" and "must be deleted if unsourced" are not synonyms, and you have been warned twice before about mass deletions without getting a clear consensus to do so. Yet, you just keep moving to new data fields and disregard the previous warnings. --RAN (talk) 04:41, 31 July 2018 (UTC)
I also read this as meaning that we don't want unsourced statements with that property. And I also tend to remove those statements if unsourced, especially at the items of living people, if I can't find a source in the first results of a web search. I would definitely support a bot removing those unsourced statements at items for non-fictional persons. If somebody cares about this information, this person can add it again, with a source. - Valentina.Anitnelav (talk) 07:29, 31 July 2018 (UTC)
Although, oddly, Valentina, any reading of SHOULD and MAY in Wikidata:Living people should take due regard of rfc2119, per Wikidata:Living people#Notes. In that document, we read 5. MAY This word, or the adjective "OPTIONAL", mean that an item is truly optional. and if we understand the may we are discussing to be may be removed, then we are left with the conclusion that removal is truly optional. And hence far from mandated. All emphasis mine. --Tagishsimon (talk) 07:46, 31 July 2018 (UTC)
Should we be removing them ad hoc one at a time, as people browse through Wikidata? Should the ones removed be restored? Should we add a disclaimer that the information is tentative until referenced? --RAN (talk) 02:27, 2 August 2018 (UTC)
deletion is not a quality improvement process, i would invite the people deleting claims to instead start a "missing citation" task flow, and start referencing claims. anything else is wasting everyone time with empty pontificating. Slowking4 (talk) 13:10, 4 August 2018 (UTC)
Farsi-only "religions"
There are several items such as Q5923778 and Q5925628 that are sitelinked only to fa-wiki, described as instance of (P31)religion (Q9174), and which appear to be sites with Iranian National Heritage registration numbers. I don't read Farsi at all, but this seemed unlikely to me; Google translates suggests strongly to me that these are not religions. Can someone who reads some Farsi have a look? - Jmabel (talk) 06:12, 30 July 2018 (UTC)
If you can find the others please do so and ping me. Apparently some keyword in the title has tricked the bot to add them which is understandable for a Persian speaker but should be removed −ebrahimtalk11:17, 4 August 2018 (UTC)
Confusion about instance-of vs subclass-of
Repeatedly I've come across the problem of having to determine what exactly the instances of a given class are. I think I understand Help:Basic membership properties but it hasn't answered my question.
For example, take head of state (Q48352): it's described as "official who holds the highest ranked position in a sovereign state". Should I think of it as a class whose instances are certain humans, or as a class whose instances are certain public offices? If I follow the subclass chain to person (Q215627), I find that its instances are in fact persons; if I follow its subclass chain to public office (Q294414), I find that its instances are public offices.
@AxelBoldt, Jmabel: This is an aspect of a very general issue: how do you model a single entity that has regular significant changes over time, whether a repeating event, a competition, a periodical publication, a body of work such as a legal code, or a political position or office like this. In my view Wikidata entities that are not specific events should be regarded as much as possible as timeless concepts. In Wikidata we don't regard a person or organization as a class of the various instances of that entity as they changed over time; I think the same should apply here. So in my view specifically in this case, president of Germany (Q25223) should be said to be an instance of head of state (Q48352). Subclass would apply only if there are multiple simultaneous holders of the position, for example for members of a legislature, where the instances would be individual seats (if they can be identified as distinct entities). ArthurPSmith (talk) 20:10, 1 August 2018 (UTC)
Totally disagree. There is no modeling difference between multiple instances over time and multiple instances at one time. - Jmabel (talk) 23:30, 1 August 2018 (UTC)
@Jmabel: It's not an issue of modeling differences, but conceptual, regarding our ontology. Would you consider "Einstein in 1905", "Einstein in 1922", "Einstein in 1955" instances of the class "Einstein", with "Einstein" a subclass rather than an instance of "human"? That is conceptually plausible, but it's not what we do. ArthurPSmith (talk) 13:55, 2 August 2018 (UTC)
I certainly would not do that about years in a person's life.
So, do I understand from this and Jheald's remark below that here "President of Germany" is used more like what I might call "presidency of Germany"? As in, we don't say that Hindenberg is an instance of "President of Germany", but that he "held the office"? Because to me, speaking normally, if I refer to the "President of the United States" I am referring to a person; if I refer to the "presidency of the United States" I am referring to an office. - Jmabel (talk) 16:16, 2 August 2018 (UTC)
We don't make individual humans members of any other classes than human (Q5). "President" or "Mayor" or whatever should always be considered to denote an office, not a person or group of people. None of these positions should ever have been added into the subclass tree of person (Q215627), it only creates confusion if such relations are there.
If there is only one office, like President of Germany, then it should be instance of (P31)head of state (Q48352) and not subclass of (P279). If it is possible to distinguish the office in different ways -- eg if one distinguished different offices for time periods, only then would a subclass of (P279) relationship make sense. I'm aware that people have made a real ploughed field of this in the database, but this is the principle we should be aiming for. Systematic clean-up in this area is long overdue. Jheald (talk) 13:02, 2 August 2018 (UTC)
@Jheald: I was asking for general rules and principles, and the principle "humans are only ever instance of a single class, human (Q5)" is an important such principle, that helps a lot. Is it documented somewhere, is there a page that collects these ontology principles? Or should we maybe put it in the description of human (Q5)? We currently have numerous subclasses of human (Q5), which I assume is fine, as long as none of them ever has an instance, right? AxelBoldt (talk) 08:31, 4 August 2018 (UTC)
Hi, we got a proprety « official name » that can store the name of stuffs, say a place, in the official language of the place it belongs to. It’s obviously useful to store the names in the original language in an easier way than to find the good language(s) and choose the label in this language(s). Labels in other languages are free to be whatever fits for that language, which is useful for example in French in which place names are often translated (or translitterated) in something in french.
It’s also useful to store the successive name of the topic if it was rebranded, and it’s the use I’m interested in here. It may be that the name change in the original language, but that the translation also change in an unofficial language… But if community choose not to create two items for before/after the renaming, how do we deal with translation of unofficial names, is there a practice/guideline/property?
My ideal solution would be to create several items but I’m not sure this is always a good fit. Do you have example for this? I’m looking for usecases.
I may try to create a global lua module for the result to be cross wiki reusable if relevant.
This question origin is a request in the french wikipedia biography wikidata infobox to show a name for a birth place at the birth date of a person, and has it’s counterpart of the french wikidata chat, but I feel like this may need a broader audience so I ask here.
You can now query the constraint violations with the Query Service
Hello all,
We started integrating the constraint violations into the Query Service. That means you can build queries using the constraint violations, with the predicate wikibase:hasViolationForConstraint. This will hopefully help you to watch better the quality of Wikidata content.
Please note that this is a first step. Not all constraint violations are exposed yet, only the ones that can be checked fast enough. We're working on having more available in WDQS.
You can base your queries on these few examples:
#10 statements with constraint violations that are currently includedSELECT*WHERE{?xwikibase:hasViolationForConstraint?y.}LIMIT10
#Map/timeline/image grid of items that have a statement with a constraint violation#defaultView:MapSELECTDISTINCT?item?itemLabel?image?coordinate_location?point_in_time?date_of_birthWHERE{?swikibase:hasViolationForConstraint?y.?item?z1?s.SERVICEwikibase:label{bd:serviceParamwikibase:language"[AUTO_LANGUAGE],en".}OPTIONAL{?itemwdt:P18?image.}OPTIONAL{?itemwdt:P625?coordinate_location.}OPTIONAL{?itemwdt:P585?point_in_time.}OPTIONAL{?itemwdt:P569?date_of_birth.}}
#Bar chart of statements that have a constraint violation, grouped by instance of the regarding item:#defaultView:BarChart#TEMPLATE={ "template": { "en": "Bar chart of statements that have a constraint violation grouped by ?property the regarding item" }, "variables": { "?property": { "query":"SELECT ?id WHERE { VALUES ?id { wd:P31 wd:P17 wd:P571 wd:P361 wd:P19 } }" } } }SELECT?instance_ofLabel(COUNT(?instance_ofLabel)AS?count)WHERE{?swikibase:hasViolationForConstraint?y.?item?z1?s.BIND(wdt:P31AS?property)SERVICEwikibase:label{bd:serviceParamwikibase:language"[AUTO_LANGUAGE],en".}OPTIONAL{?item?property?instance_of.}}GROUP BY?instance_ofLabelORDER BYDESC(?count)LIMIT30
Thanks, IMO this is already now definitely the best new feature of the month! Am I correct if I find that there are currently only ~10k constraint violations in total exposed to the Query Service (count)? It would also be great if there was a list of supported constraint types (like this one perhaps), and I already noticed that the Query Service autocompletion tool does not know wikibase:hasViolationForConstraint yet. —MisterSynergy (talk) 11:10, 3 August 2018 (UTC)
Good to hear you like it :) Unfortunately we can't give a list of supported constraint types because it doesn't depend on the type of constraint but on the time it takes to execute the specific check on the specific item. If it's done before the query service update it gets in, otherwise not. We wanted to see if there is interest in this feature before making the necessary more complex changes that are required to get them all in. But it seems people really like it so we'll do that and then all of them will get in. The ticket for that is phabricator:T201147
Excellent. Looks like quite a lot of data is still missing. I got the first one to run on lighthouses, but somehow the constraint violation statement disappeared from query server (without being fixed). --- Jura21:53, 3 August 2018 (UTC)
I'm trying to make out from the Phabricator ticket exactly how this got implemented. Apparently it is not part of the regular RDF changes feed, as I'm not seeing it on my private server. And there is some discussion of how it interacts with caching in the wbcheckconstraints API, so maybe it is only updated when someone looks at the item. Bovlb (talk) 00:14, 4 August 2018 (UTC)
User Lakokat (speedy) deleted the items Q55862320 and Q55863756 without any discussion and any check of its usage at the page information. I ask anybody who has the right to undelete them.
Both items are used on the German Wikivoyage article of Montreux in Switzerland. Q55862320 is a historic market hall, the other one a youth hostel. Unfortunately the market hall is not yet described in Wikipedia. The items are necessary to share information across various language editions. --RolandUnger (talk) 09:31, 4 August 2018 (UTC)
I restored both items, as they were in use at dewikivoyage per [1][2]. It would nevertheless be great if you could complement more information about both entities, to make it easier for Wikidatans to understand what those items are about (particularly references to external sources and external IDs, if available). —MisterSynergy (talk) 10:32, 4 August 2018 (UTC)
I am sorry for deleting the items you have mentioned, I agree that they should not be deleted in the first place. They have been restored by User:MisterSynergy, thank you MisterSynergy. When I delete any item, I will check its properties, sitelinks and inbound links (for example, Special:WhatLinksHere/Q55862320) to check its usage and whether it meets the notability policy. It does not produce any inbound links or transclusions thus I am not awared of the item being used when the data are queried by wikidata or other wikimedia projects. I will definitely check also the item 'info' to better trace the usage when I delete any items from now on. Lakokat11:01, 4 August 2018 (UTC)
Yes but to be honest not super high. It is one of the projects I offer students when they ask for projects around Wikidata to do as part of their thesis or similar. If there is larger agreement that this is very important to have (compared to all the other things that you want) I can bump it in priority. --Lydia Pintscher (WMDE) (talk) 16:08, 4 August 2018 (UTC)
Have we started simple '-s ending' adding plurals as aliases in English?
I ask because after I removed one from drum (Q11404), User:Jura1 has reverted me, more than once. His latest edit summary claiming that some unspecified part of Help:Aliases (which makes no mention of such cases) and, bizarrely, Help:Description support his action. My suggestion that he start an RfC if he wants us to add such aliases seems to have been ignored.
"drums" seems valid per "The label on a Wikidata entry is the most common name that the entity would be known by to readers. All of the other common names that an entry might go by, including alternative names acronyms and abbreviations; and alternative translations and transliterations, should be recorded as aliases.". Help:Description was just mentioned as you apparently ignore it, looking some change on my watchlist. I think the only problem with plurals that came up in the last discussion, may be that they could refer to another item. (This isn't the case here). --- Jura22:21, 3 August 2018 (UTC)
I (a native English speaker) have never known anyone, ever, refer to "a drum" as "drums"; it is not a "common name that the entity [N.B. singular] would be known by to readers". Your quote - "...acronyms and abbreviations; and alternative translations and transliterations..." - does not include "plurals". "Drums" does, contrary to your assertion, refer to another item; it is a valid alias on drum kit (Q128309). Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits22:29, 3 August 2018 (UTC)
As a native English speaker, we do say "(s)he plays drums" rather than "(s)he plays drum" with reference to a percussionist whose normal instrument is a drum kit (whereas we don't say "(s)he plays guitars" or "(s)he plays piano". So this one is a little tricky. - Jmabel (talk) 08:49, 4 August 2018 (UTC)
I am trying to use quickstatements, and I encounter the problem that identical values of the same property can not be added. In my case, Elo ratings shall be added, and if on two or more elo lists, the elo of a player is the same (happens often especially for inactive players), then of course this has to be documented, but, as it says at Help:Quickstatements, "Existing statements with an exact match (property and value) will not be added again". How can I force the tool to add those statements notwithstanding the identical values? Steak (talk) 18:54, 2 August 2018 (UTC)
I don't have much experience with importing large amounts of data to wikidata. Can somebody recommend some other possibility? Steak (talk) 10:15, 3 August 2018 (UTC)
Not so difficult. However, I recently ran into trouble using PAWS with my regular account without a bot flag. It might be wiser to set up a SteakBot account, apply for a bot flag and do the import with that account, using PAWS. —MisterSynergy (talk) 13:13, 3 August 2018 (UTC)
The problem for me is that en:Normandy is connected to Q18677875 and is:Normandí is connected to Q15878. This creates a situation where my query for pages that are on the Icelandic Wikipedia but not the English Wikipedia shows that the English Wikipedia has no article for Normandy. But it does have an article, both articles contain all the definitions of Normandy, but can only be connected to one data item. There are 18,000~ pages on the Icelandic Wikipedia not linked to a corresponding English Wikipedia page. But how many of these are similar mismatches rather than missing pages? Frayae (talk) 11:29, 3 August 2018 (UTC)
@Liuxinyu970226: You keep pushing this, but it's not going to happen. Ever. For a taste of what happens when multiple sitelinks are allowed, consider Commons category (P373). It becomes a nightmare to identify what the actual principal correspondence is. Which is why people like Mike Peel have been working so hard to create Commons sitelinks, precisely because they are 1-to-1. Other secondary linking mechanisms might get added, but the basic principle of 1-to-1 for sitelinks is not going to change. It's too valuable, too much of the time. Jheald (talk) 22:55, 5 August 2018 (UTC)
Newbie question: what does one do when 2 items have the same English Wikipedia article.
While editing a category for the Ainu language at English Wiktionary, I noticed that the the Wikidata-derived Wikipedia link pointed to w:Hokkaido Ainu language, which is a redirect to w:Ainu language. That's because Wikipedia has only one article for the Ainu language, but Wikidata has several items- as it should- and Wiktionary's data module points to Q20968488. Now that I've thought more about it, I think Wiktionary should be pointing to Q27969, the parent category.
Before I realized that, though, I tried to change Q20968488's Wikipedia en property to point to the only Wikipedia article on the subject, but it failed due to Q27969 already having that in it's Wikipedia en property. Given that one Wikipedia article may be the only page at Wikipedia for more than one Wikidata item, it seems strange to have a one-to-one relationship between Wikidata items and Wikipedia articles. Reserving the Wikipedia article for the parent category won't always work, because Wikipedia's coverage decisions aren't always based on the data structures that Wikidata uses. Or am I misunderstanding something? Chuck Entz (talk) 03:04, 6 August 2018 (UTC)
I'm working on a large upload to Wikimedia Commons of images of Classical sites. Many of the sites and even objects in the upload have Vici.org identifiers. It so happens that this identifier has a property in WD as well (Property:P1481), but it is used only a few times. It obviously makes sense to add this property to the larger sites and monuments (for example Ecbatana, Q696193), but what about small monuments? For example, the upload has images of a monumental stone lion in Ecbatana. Should that be imported to WD? And if not, are there any guidelines on where and how we should draw the line? As another example, the upload contains material on the Mausoleum of Avicenna, which is also on Wikidata (Q5952145) and on vici.org. Any advice would be welcome! Best, --AWossink (talk) 11:28, 1 August 2018 (UTC)
This is more a matter of WD:N than anything else. Just create the items that are relevant, that is when they have more than one reference.--Micru (talk) 19:36, 1 August 2018 (UTC)
@AWossink: If in doubt, lean towards creating it (provided of course you've tried & failed to match it). The time is going to come when CommonsData will be live and we are going to want to start adding depicts (P180) and main subject (P921) statements there for images on Commons. Very useful to have the item here to link to, and more information about it. Beyond that, if something is worth creating a category for on Commons, then definitely make sure it has an item here, which will mean that c:Template:Wikidata infobox can be added to the category. Once a thing has an item here, with an identifier uniquely distinguishing it, that opens the doors for all sorts other sites and other cross-referencing data to potentially be added in future -- and all sorts of query results that it could start to appear in. Jheald (talk) 12:57, 2 August 2018 (UTC)
Now edit summaries are automatically generated, and editors can only add customized edit summaries to reverting, restoring, or merging. Is it possible to allow editors to add edit summaries to all kinds of edits? The auto-generated edit summary can provide just basic information on what the edit is, but cannot provide information in details or explain why the edit is made. Allowing editors to add edit summaries can provide additional and useful information. Of course the editor-added edit summaries would not replace, but just supplement, the auto-generated edit summaries. I would appreciate it if anyone could discuss this issue here. Thanks. --Neo-Jay (talk) 05:59, 5 August 2018 (UTC)
Pick one item to be used in statements (an instance of taxon, although the label can be "dog" for convenience), and minimize use of the duplicates created for Wikipedias. Ghouston (talk) 05:37, 24 June 2018 (UTC)
Apparently using a taxon-centric model with other names being properties or aliases is unacceptable although it not entirely clear what use-cases that model would fail in. The claim for using a name-centric model is that one item for every name keeps them independent of the organims being referred to because every taxonomic author can have their own view of what the circumscription (characters defining a group) of a taxon (a group of organisms) is. This would presumably also require items with labels that are meaning-free - maybe taxon entities that are unnamed and use a UniqueID. See also Property_talk:P1420. Pinging also @Peter_coxhead: @Kaldari:. Shyamal (talk) 11:36, 25 June 2018 (UTC)
So where are properties - like heart rate, or number of legs, or gait, or... - which are about the (class of) organism, not the name, supposed to go? And, anyway, aren't items about the names of things supposed to go in the Lexeme namespace? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits17:32, 25 June 2018 (UTC)
Andy Mabbett, and the common name, not the taxon!, has a "taxon common name" Q144#P1843. Also subclassOf "domesticated animal" AND "pet", with "pet" having subclassOf "domesticated animal". It is easy to solve, but users that want to mix items about subclasses of biota with items about names of these subclasses stand in the way. 178.5.32.20112:33, 24 June 2018 (UTC)
This is the heart of the problem. Yes, organisms have physical properties, which may be shared by a group of organisms (a taxon); names do not have physical properties. Yes, Wikidata muddles taxa and their names. However, as discussed at length at Property talk:P1420#taxon-centric, at present there's no known way of representing taxa in Wikidata and allowing different taxonomic views to be shown. Peter coxhead (talk) 06:14, 26 June 2018 (UTC)
Even given the problems with multiple overlapping taxa, is there really a need for separate items for common names? Most taxa don't have them. Now we have three items for "dog", when 2 seems sufficient. Ghouston (talk) 10:12, 27 June 2018 (UTC)
Perhaps the question should then be - Should wikidata aim to (attempt to) capture details of taxonomic histories and circumscriptions rather than serve as an index to Wikipedia entries that explain those subtleties? Shyamal (talk) 11:01, 27 June 2018 (UTC)
@Pigsonthewing: The problem you've identified here is only the tip of the iceberg. Currently Wikidata conflates taxons and taxon names into the same items. And thus when a species has multiple synonyms, things like heart rate, range maps, interwiki sitelinks, etc. get randomly spread among those various synonyms. See the lengthy discussion at Property_talk:P1420#Changing_this_to_a_string_property. Clearly the current system is not tenable. We probably need to do a project-wide RfC about this at some point. Kaldari (talk) 22:35, 27 June 2018 (UTC)
In practice, P3395 is not a property that applies to a taxon; it is a human-centric property, used for persons and for what are commonly perceived to have personalities, also (that is, pets). Thus there is no real conflict in these items. Just a case of raising trouble in this Project chat. - Brya (talk) 07:19, 7 July 2018 (UTC)
It beggars belief that anyone with even a school-level education in biology would claim that only "persons and what are commonly perceived to have personalities" have a heart rate (for that is what heart rate (P3395) is). But if you really think that all I am doing in this section is "raising trouble", you know where the admin noticeboard is; raise your complaint there. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits20:12, 7 July 2018 (UTC)
It beggars belief that anyone with even an average helping of common sense would go by surface appearances and would refuse to look at how a property is actually used. Wikidata has several properties that are used only for pets. - Brya (talk) 06:18, 8 July 2018 (UTC)
Knowing the cause of a problem is part of the solution. I am not sure the few participants here who are discussing it in good faith should be expected to provide instant solutions when they were not around when the problem was created in the first place without much discussion. It seems clear that creating synonyms and names as items is best set aside for now until there is a larger discussion that examines a practical ontology and one that does not get archived out from here. Shyamal (talk) 02:15, 30 June 2018 (UTC)
I know this paper very well, but it certainly does not support your conclusion. For me the essence is summarized as „[...] research along the taxonomy/ontology interface should focus either on strictly nomenclatural entities and relationships or on ontology-driven strategies for aligning multiple taxonomies, but not on building static networks for large portions of the tree of life.“ This is were WD could do a minor contribution. But this is not closely related to the issue raised above. --Succu (talk) 20:44, 30 June 2018 (UTC)
A useful step then would perhaps be to explain what kinds of queries you would be running from your ideal taxonomic database and what your results are expected to be. Shyamal (talk) 11:50, 1 July 2018 (UTC)
What's your „ideal taxonomic database“? I havn't one! See the paper you cited. What queries do you want to be answered? How many species exists is out of scope here. --Succu (talk) 21:37, 1 July 2018 (UTC)
You misread me - there was nothing mentioned anywhere above that had to do with wanting to know the number of species. I do not see an purpose in your statements other than dismissing discussion. Shyamal (talk) 04:01, 2 July 2018 (UTC)
A first goal should be to get the nomenclatural aspect running. Afterwards there is plenty room to do integrate all the different taxonomic viewpoints. --Succu (talk) 21:31, 3 July 2018 (UTC)
I thought a bit about the issue and the solution is very similar to the human (Q5)/Homo sapiens (Q15978631) split. It would be good to solve both issues.
I don't think entities in the Q-namespace should be instance of (P31) of common name (Q502895). I think it's better to model names in our new datatypes than in the Q-namespace. We will have lexeme's for "dog" and "human" and any other species that might have the same problem. Ontology-wise that clears up the problems very well.
It does have the problem that this means "human" and "Mensch" wouldn't get interwikilinks from Wikidata anymore given that they are different lexemes but I think the amount of items is small enough that we can solve the issue with a bot that creates manual interwikilinks. ChristianKl ❪✉❫ 15:34, 30 June 2018 (UTC)
See eg phab:T199119#4412377 for the kind of unhappiness and confusion this gives newcomers to Wikidata. This ticket was to investigate walking up the subclass of (P279) tree (and others), to find additional keywords to add to indexes along with depicts (P180) values to facilitate image searches.
@Succu: The latter two would be difficult to do automatically, because there is no of (P642) qualifier given. But would there be any objection to going ahead with this, in cases where there is a target given by P642 ? Jheald (talk) 21:00, 22 July 2018 (UTC)
I prefer knowledge over automated "healing". I dislike instance of (P31) (and related) statements with qualified by of (P642) as well. Could you please give two or three non trivial examples you have in mind? --Succu (talk) 21:11, 22 July 2018 (UTC)
@Pigsonthewing: Sorry to have been slow to get back to you. Rather max'd out with a number of other projects at the moment, plus real life :-) But I might make some trial edits next week, to see how they go down. Jheald (talk) 10:27, 4 August 2018 (UTC)
Yes, "instance of: common name" is a somewhat makeshift construction, but a "said to be the same as" (P460) construction won't work across the board as these are not necessarily symmetrical relationships. The same applies, but even more so, to "instance of: synonym : of" which is never reciprocal. In the latter case we actually need a separate property, "is a synonym of" or "is a synonym of taxon", the inverse of "taxon synonym" (P1420). - Brya (talk) 03:05, 6 August 2018 (UTC)
A good consideration. How did you identify this list? Is there a clever query we can run regularly to identify missing properties like this? --99of9 (talk) 00:47, 6 August 2018 (UTC)
Just because @GerardM: gets away with cryptic posts of this sort in the Wikidata Facebook group doesn't mean your own posts here should be cryptic, @Thierry Caro:. That being said, I'm sure Ethiopia, Pakistan, and Vietnam all most likely have census codes (and I'd be willing to work on Pakistani census codes if I were able to find any in a consistent place); I imagine @Beat Estermann: and others might find good properties related to dance, seeing as such folks have worked with theatre; I do wonder if there are Hinduism or Islam-related encyclopedias that we could create properties for (as we already have for the GAMEO and the Catholic Encyclopedia)—perhaps the same may be said for magic, circus, sociology, and psychology ontologies. Mahir256 (talk) 02:08, 6 August 2018 (UTC)
Thank you for calling a message like: these are important subjects and we know nothing about them "cryptic". That means that too little words are used for you to get the message. As you may know from Facebook, I have involved myself in bringing out more information about Africa, I use #AfricaGap as a hashtag, because there is so little data and the data that is there is of poor quality. Ethiopia is in Africa. These links is about its national politicians, the other is about its and geographic entities.
When you consider Africa, less than 1% of the humans in Wikidata is from Africa. The discrepancy is huge. Do you really think en.wp will call African sources notable? Not really. The information that I work with is fragmented and it is often wrong. However, it is 100% better than the no information we currently provide. The wrongness is often in it being outdated.
It is wonderful for the "pillars of our community" to easily dismiss the ideas of others. Obviously when it is pointed out that our data is lopsided and they want more where we have so much, it does not help when this is called out. When you consider how little money is spend on Wikidata, it is certainly a miracle. It works and it works well enough. But at the same time data on sources do not help us to indicate what sources are fake. We do not get a fair impression of our data at wikidata itself. Query is so necessary when you are not used to using tools like Scholia or Reasonator. (Being cryptic here but read my blog for clarification. \
@GerardM: The issue I and at least ten other people—judging from those who liked @Ijon:'s comment—had with your Facebook post, and which I see at least @Tagishsimon: had with Thierry's post above, was rather clearly not that either of you want to close the knowledge gap with respect to developing countries and other major subjects on Wikidata. It was never any indication of dismissal or a lack of caring for the point you were trying to make; you should have at least noticed in this instance, for example, that I have indicated some interest in some of Thierry's mentioned topics, some of which I am fully aware overlap with your own. I believe Asaf's comment was intended as an attempt to spell it out to you, but there is little use being brief if others do not see clarity in that brevity.
I cannot speak for what those others thought when seeing a Reasonator link accompanying your message, but there were multiple problems with the page (a general paucity of information? missing label translations? unorganized information layout?) that I could perceive upon seeing it, any of which could have been the problem indicated by your message when taken by itself. This inability on our part to determine what the problem may have been is in no way indicative of our thoughts on the problem you had in mind, and if the term 'cryptic' is too strong a word to describe your post in relation to others' misunderstanding of it, then I sincerely apologize as in that case I have in fact picked a bad choice of words. This, however, should not alter the need to be thorough and clear in all of your explanations the first time around.
In roughly this same vein, @Thierry Caro:, please understand that your comment definitely has the potential to confuse people when taken by itself (to suggest that this isn't the case is to deny Tagishsimon's personhood). I do think answering @99of9:'s questions helps add a great deal of necessary context, but this doesn't negate the ultimate benefit of it being there in the first place.
It is certainly a regrettable situation, Gerard, that not everyone may be immediately aware of your work on Africa-related topics, this work being praiseworthy in itself; even being well aware of this context myself through some of your blog posts and other on-wiki forums, I still had the confusion I expressed above. We are human after all, and it is only natural that some people will not immediately pick up the appropriate background for comprehending some messages on this page and elsewhere. As such it is always appropriate to repeat the appropriate information needed for context and to elaborate to the extent needed to eliminate ambiguity. Mahir256 (talk) 15:51, 6 August 2018 (UTC)
I agree with you, Mahir, about the original post above being unclear, and about the difference between not understanding the intent of a post and not caring about that intent.
However, since you mentioned me by name, I feel I should point out to you that it was an incivil way to express your confusion to make that assertion about GerardM "getting away with it". Consider avoiding this kind of snark in the future, to increase the odds of being heard and understood by your interlocutors.
Talk about "on message". This whole thread is high jacked by the form of conversation, lack of understanding, and "how poor an idea was expressed". To be blunt, we are to be about the sum of all knowledge and Wikidata proves the extend we are little more than a stamp collection. Wikidata has a really strong data set about biomedics but it is years behind a project I was involved in, Wikiprotein, because it does NOT allow you to infer knowledge about what those papers stand for and it does not tell you anything about the validity of those papers. We are really strong improving on the gender gap and, yes it is important, but it is to the exclusion of everything else.
With red links associated with Wikidata items we will bring much tighter linking with Wikipedias, it will help combat fake news. But who has an interest in a plan that makes the Wikipedias less insular? The Cebuano Wikipedia is the best source for information on many subjects about Africa but who reads Cebuano? (with deep linking to Wikidata we would all know this). Both Wikipedia and Wikidata suffer from an ingrained preference for attention of itself and personal interests. It is how it discriminates and it is why it has hardly anything to offer to Africa.
PS When you learn about the growth of the Swahili Wikipedia on MuddyB's blog.. it is because of the Cebuano content, an easy to ignore fact. "We" at Wikidata boycot Cebuano. Thanks, GerardM (talk) 05:41, 7 August 2018 (UTC)
Just chipping in here, since the reason we have properties for GAMEO and the Catholic encyclopedia is because people went out and created those freely accessible resources and Wikipedians like myself made use of them several times in Wikipedia articles, creating a basis for wider usage through Wikidata. We should create properties for all of these, like The Jewish Encyclopedia (Q653922). We can really use these for all sorts of things (such as yet-to-be-described religious paintings) Are there any other religious encyclopedias? Same for native language encyclopedias. Jane023 (talk) 06:53, 7 August 2018 (UTC)
Keep them if someone has bothered up upload them. Ten years from now, one of the minor players might win an Oscar, and it would be cool to have their complete filmography here. - PKM (talk) 00:19, 2 August 2018 (UTC)
IMO it's one thing to list every single actor who played a bit role on that film's item, but I don't see much worth in creating items for all those unnamed characters (cab driver, female bartender, etc.), at least as long as they aren't referenced anywhere else.
And on another note: I'm a bit concerned about the large amount of data we've been copying from non-free databases like IMDb into Wikidata over time. --Kam Solusar (talk) 01:01, 2 August 2018 (UTC)
I agree that this seems pointless past a certain point. There should be items only for the protagonists and for more minor characters appearing in multiple works. Thierry Caro (talk) 11:27, 3 August 2018 (UTC)
They may be relevant if one is interested in the representation of features like occupation, ethnicity, gender in films. There are also other questions (e.g.: what kind of occupations did an actor represent) where this information could be relevant. - Valentina.Anitnelav (talk) 07:50, 4 August 2018 (UTC)
In theory if we could have it, would we want all data from IMBD? I think yes, right? Has there been previous discussion?
All of these roles are tied to disambiguated actors who have their own Wikidata items and their own IMDb ID (P345). This means they connect at multiple points to other Wikidata content. This seems good to keep. Blue Rasberry (talk)14:43, 8 August 2018 (UTC)
How to add a qualifier with "unknown value" via QuickStatements?
From the help page and searching online, I could not figure out how "unknown value" can be added programmatically with a qualifier for an item. My use case is: In the large collection of 20th Century Press Archives (Q36948990), some folders contain a certain number of works (P3740), of which - due to intellectual property rights - only some are online (covered by the newly created number of works accessible online (P5592)). However, for some dossiers (started after 1948) which are still stored as paper or microform, we only know about the existence of the folder, but we don't know how many articles it actually contains. I think a good way to document this would be to add the P3740 qualifier to PM20 folder ID (P4293) with "unknown value" as value, but could not figure out how to do that (with Quickstatements or otherwise - the situation occurs in a large number of cases, so doing it manually is not an option.) Jneubert (talk) 07:17, 6 August 2018 (UTC)
Are there similar items for no value / deprecated / preferred? Allowing rank to be specified as a temporary qualifier would be quite helpful. Jc86035 (talk) 09:06, 6 August 2018 (UTC)
@Jc86035: I don't think so, at least not as yet. A similar "placeholder for <novalue>" along exactly the same lines should be very straightforward. For statements that are to be deprecated, I usually add a reason for deprecated rank (P2241) qualifier, and then ask at Bot Requests for somebody to deprecate all the statements marked that way in a particular group. But a better / more automated mechanism might be an idea. Perhaps a new property "edit request" that could take various values to request bot intervention? Would be useful to hear what eg Pasleim thinks. Jheald (talk) 09:14, 6 August 2018 (UTC)
Maybe something like "edit request find" (one each for QIDs, properties and strings), "edit request replace" (ditto, but not needed for deprecated, etc.), "edit request scope" (property / value / qualifier property / qualifier value), "edit request property scope" (PID)? I had to manually change a few dozen values of placeholder 1 (Q55560743) and placeholder 2 (Q55560744) because QuickStatements would not let me add a new statement with the same property and value as an existing statement. Jc86035 (talk) 09:28, 6 August 2018 (UTC)
@Jheald: Just to make sure we talk about the same thing, I've manually made up an example (G. Meir) of what I want to achieve. In QS, it did not work for me to insert a non-numeric placeholder as the qualifier value - neither Q80596|P4293|"pe/012253"|P3740|Q53569537 nor a string like Q80596|P4293|"pe/012253"|P3740|"unknown_value". Both were rejected by QS as Error. So I suppose I didn't get your suggestion right. Jneubert (talk) 10:24, 6 August 2018 (UTC)
Well, some artificial numeric value (like -9999999999) could work, but that's very special, could overlap with real values in unforeseen cases, and probably wouldn't work with strings or other datatypes. Jneubert (talk) 10:54, 6 August 2018 (UTC)
Excuse me, if this is a often-before discussed issue - but is it known if the limitation of QS re. special values is just a limitation of the current QS syntax, or is the restriction imposed by the underlying API? Jneubert (talk) 10:57, 6 August 2018 (UTC)
It's a limitation of QuickStatements. If it were the API, bots wouldn't be able to fix it and we wouldn't be able to add the statements manually either. - Nikki (talk) 11:28, 7 August 2018 (UTC)
Thanksworthily, User:Maxlath has extended wd-cli to handle no_value and unknown_value claims in statements and in qualifiers!
The following commands worked for me in order to qualify the first PM20 folder ID (P4293) property for A. Miller with an "unknown" number of articles:
claim_guid=`wd data Q80596 | jd 'claims.P4293.0.id'`
wd aq $claim_guid P3740 '{"snaktype":"somevalue"}'
This should be scriptable for a large number of items (will apply for an according extension of my bot accout). Thanks again to everybody involved! --Jneubert (talk) 15:20, 8 August 2018 (UTC)
I think the common sense answer is yes, but this is problematic because songs don't usually have line breaks or punctuation. Different sources might split lines differently; for example, one source might split the lyrics by clause instead of by sentence. Furthermore, it's often not clear what the difference is between vocalizations in a song which aren't part of the lyrics and vocalizations which are. Jc86035 (talk) 16:03, 8 August 2018 (UTC)
I think one has to rely on the source, report what the source said. Have two P1922s if there is more than one source telling more than one story. Agree that 'yes' is the overall answer. (I guess it must be possible that different publications of the same poem will have the same issue.) --Tagishsimon (talk) 16:09, 8 August 2018 (UTC)
How would you call this?
Hello all,
The development team is currently working on the termbox, to improve it both on desktop and mobile versions. While doing some user research, we realize that we don't have a clear name for the content of this zone on Wikidata: the combination of languages, labels, descriptions, and aliases. This is an issue for us, because if we want to work on things, we need to name them properly first.
Before having an official voting or anything, I wanted to ask you: do you already have a name in mind? How do you describe this specific zone, when you're showing Wikidata to someone, for example? And if you never thought about this before, what name would make sense for you? :)
It's worse than you think. We don't have an official term for the interface that presents this box. That is, if, from the Wikidata main page, you click "Random item" on the left edge of the window, you are taken to an interface that displays information about a random item. We don't have an official name for this interface. This is important because this interface behaves differently from some other interface, such as the API, so it's necessary to have a word for the (whatever it is) when discussing how it differs from other interfaces. Jc3s5h (talk) 11:27, 7 August 2018 (UTC)
I usually describe it as "the box with the labels, descriptions, etc". The only I can think of that would work better than term box would be "names and descriptions box" (where "better" means people should be able to work out what that is without any explanation - if it has to be explained, we might as well stick to term box) . - Nikki (talk) 11:43, 7 August 2018 (UTC)
The "header" for each "language" consists of a "label" and a "description". It can include one or several "aliases". --- Jura11:45, 7 August 2018 (UTC)
If I understand correctly, there is one for yolk, one for egg yolk and one for chicken egg yolk ... but interwikis links are also big mess.
Mikani (talk) 15:37, 9 August 2018 (UTC)
Wikidata weekly summary #324
Here's your quick overview of what has been happening around Wikidata over the last week.
Past: State of the Map, the OpenStreetMap equivalent of Wikimania, was held in Milan from 28-30 July. There were two meetups on interactions between Wikimedia and OSM, and Wikidata was mentioned in several sessions, including:
Thanks for the update! I am curious, is there a Phabricator task (or any other way) I can follow to see how far Senses are from being actually deployed? Or could you just give us a guesstimate? Cheers, --Denny (talk) 19:24, 7 August 2018 (UTC)
The query minerals named after women mentioned above uses the syntax WITH ... AS %name and INCLUDE %name. This seems to be a useful extension to standard SPARQL. Is it Wikidata specific? Are there other such syntax extensions? Toni 001 (talk) 13:43, 10 August 2018 (UTC)
UTSA Roadrunners (Q7876032) (about their football team in particular)
The second seems to be about the football team and should be renamed accordingly. In Wikipedia, this would be accomplished via the "Move" link and then renaming as "UTSA Roadrunners football". Would appreciate if someone could review and resolve, such that the football content follows the expected naming convention.
Here are examples of the expected naming conventions:
It's ok to have more than one claim for the same property on an item, if you have sources that state that. However, for books we probably should have separate items for each edition - see Wikidata:WikiProject Books for further discussion on the books data model here (where I note that ISBN is mentioned as a property of the edition, not of the work itself). ArthurPSmith (talk) 18:03, 9 August 2018 (UTC)
ISBN-13 supersedes ISBN-10, and the latter can be calculated from the former. Arthur is correct about items for each edition, but where such items do not yet exist, I tend to use the ISBN from the first edition, (hardback if applicable), if I can find it. But any valid ISBN is better then none! Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits18:37, 9 August 2018 (UTC)
i use the isbn13 from a first edition, if i can find it. worldcat is a fine source.[3] oclc number works also. isbn is more a stock number, oclc is a library database number. reports are isbn10 breaks sometimes. Slowking4 (talk) 12:59, 13 August 2018 (UTC)
I could create an item "Qantas airport" and make every airport where Qantas has had a service an instance, but it doesn't seem very desirable. Ghouston (talk) 00:44, 10 August 2018 (UTC)
You'd kinda expect London Overground stations to have UK railway station code (P4755). London Overground is that part of the regular railway system operated by TfL; it manages some stations, uses stations managed by other ToCs, etc. Meanwhile, the property Amtrak station code (P4803) will identify the station as one serving Amtrak trains. That's just the way it is. Not many Amtrak trains passing through UK stations. --Tagishsimon (talk) 05:35, 11 August 2018 (UTC)
Bless. SEO ftw. I've pruned the three items added by our new American entrepreneur- Internet business owner and search engine optimization (SEO) expert friend & dropped a note on his talk page. --Tagishsimon (talk) 22:38, 9 August 2018 (UTC)
Different Language Wikipedia Article Conflicts for Game "Dandy"
There is a video game called "Dandy" that was released in 1983 for the Atari 8-bit line of computers, and then was substantially altered version and released for other platforms in 1986. The versions were different enough that the game database MobyGames lists the versions under separate articles. The English Wikipedia article for the video game Dandy (Q5215764) focuses primarily on the 1983 version but also lists the platforms it was later released on and only mentions the later versions in a single sentence under the section "Legacy". In contrast, the Italian Wikipedia article describes only the 1986 version and mentions that it is based on the 1983 version, which is linked to as [4], an article that doesn't exist.
My question is, since the English article is about both versions of the game, should the Wikidata item for the English article Dandy (Q5215764) contain a duplicate of all of the information from the Wikidata item for the Italian article Dandy (Q25409308), including database identifiers that are supposed to be unique? Is there a standard policy for when the content in Wikipedia articles for different languages doesn't line up and causes conflicts/redundancies like this? Rampagingcarrot (talk) 05:55, 10 August 2018 (UTC)
I think the enwiki article is about the 1983 version. The other versions are only mentioned briefly in the "legacy" section. It's like en:Barack Obama which is about Barack Obama despite having a "legacy" section that mentions other things. Ghouston (talk) 06:18, 10 August 2018 (UTC)
@Ghouston: I think you are right that the article is mostly about the 1983 version. An issue I see is that the platform and release date in the infobox on the right include the information for both versions. Should the Wikidata item Dandy (Q5215764) for this article exclude information about the later version in its platform and release date sections? And do you have a recommendation for what property to use to link this item with the item Dandy (Q25409308) for the second version, based on the Italian article? Perhaps modified version of (P5059)? Rampagingcarrot (talk) 19:28, 10 August 2018 (UTC)
In general, the Wikidata item should be about a coherent thing in the world, more than about a Wikipedia article (which is bound to evolve over time anyway). So I would think Dandy (Q5215764) should exclude information about the later version, but it can use followed by (P156) or replaced by (P1366) to refer to the distinct item for the later release. - Jmabel (talk) 20:35, 10 August 2018 (UTC)
Hello. In an old census, the populations of two villages are counted together as one. In there a way to add this information to both villages items? Or we should not add that information because is about both villages? Xaris333 (talk) 23:43, 10 August 2018 (UTC)
First thought is that you need a new item called 'village 1 plus village 2' with a description of 'villages combined for census purposes'; and join the discrete village items to the new item using part of (P361). I think you cannot add the value to the two discrete items since it is always about more then any one of them. --Tagishsimon (talk) 23:49, 10 August 2018 (UTC)
After that old census, all the other census have separate population for each village. So, its useless to do that. They were always 2 villages, never one. Xaris333 (talk) 00:34, 11 August 2018 (UTC)
Sadly, that's the structure the data calls for, if you want to use it at all. They were always two villages, but they were combined as a census area. That's life. --Tagishsimon (talk) 00:41, 11 August 2018 (UTC)
Looks like none of the villages or municipalities of Paphos District have area (P2046) statements, and so are all excluded from the results. See, for instance:
Does Wikidata:Notability apply to places? For instance, should instances of church building (Q16970) be added to Wikidata? If it does apply to places, how does "clearly identifiable conceptual or material entity" work? Does it's very existence as a physical place qualify (especially since it is a named physical place)? I understand that we wouldn't want to have every single addressed place in the world, but perhaps if they are named places that's ok? --U+1F360 (talk) 16:45, 12 August 2018 (UTC)
@U+1F360: We still struggle to explain what to include but every church in the common understanding of the concept should have a Wikidata item. Notability is established on the basis of having more structured data. Lots of buildings and organizations can have structured data.
There is a sense of persistence also, so a transient concept like pulling government registration records of businesses and nonprofit organizations which never do anything after registering (a common situation) is out of bounds. Typical churches will occupy land, connect to a larger organizational body, report finances, and expel data into the digital record. Model them all if you have references that you can cite and integrate! Emphasize the data sources over just the entities. Blue Rasberry (talk)17:31, 12 August 2018 (UTC)
(ec) Notability applies to every record. I think the latter part of criteria 2 is in part the key: "it can be described using serious and publicly available references". Churches surely fall into that category by virtue of publications like Crockford's Clerical Directory (Q5187367) and counterparts around the world. Of course, in truth, many/most addressed places in the (western?) world have 'serious and publicly available references', especially as government property databases become accessible as a result of open data initiatives. So value judgement or the application of so-called common sense is still required - Blue Rasberry's example is good - if we are to escape the importation of, say, the contents of the collected telephone directory (Q220393) of the world. Where the line is drawn will differ from one person to the next; but, equally, there is no scarcity of space on wikidata for new records and so I think we can lean towards a liberal interpretation of our guidelines. --Tagishsimon (talk) 17:38, 12 August 2018 (UTC)
With tracklist (P658), how is it possible to indicate that a track is on the second CD of an album or has a non-integer track number? I thought using a generic identifier property might work, but it might be better to use a track-specific property if none exists (particularly since a specific property could have the format standardized to {side or CD}–{track number}). Jc86035 (talk) 13:03, 13 August 2018 (UTC)
pt.wikipedia has a stupid captcha on it which I cannot get through. (also wtf.) Is there anyone here with the patience & language skills to get something done there? This was what I was wanting to post:
@Tagishsimon: no problem, glad to help! I've never had to fill a captcha there, so I don't know how it is hahaha. Anyway, I'd do the merging, but in this case I don't know what the proper procedure would be, so I prefer to wait for someone with more experience. Best regards! Guilhermebm (talk) 22:44, 15 August 2018 (UTC)
No. Looks scary. But the merge has been done by user:Caio!, for which many thanks. I would post my thanks on pt.wiki, but as well as the captcha, my short note saying thanks was recognised by pt.wiki as being likely vandalism. smh. --Tagishsimon (talk) 14:00, 16 August 2018 (UTC)
I propose this to become an official Wikidata policy. If you disagree with the proposed text, please say so before the thread is archived.--GZWDer (talk) 19:41, 10 August 2018 (UTC)
As it's just css and js, not interface in general, shouldn't it be named accordingly? Some wikis have interface editors that can edit mediawiki namespace. --- Jura12:19, 11 August 2018 (UTC)
It looks like the description doesn't match the actual rights granted. I'm not sure if the proposal should differentiate between users beyond grandfathering existing ones. --- Jura14:26, 11 August 2018 (UTC)
Despite your claim to be "exact", your list of people, whom you describe as having a "lack of notability" includes, on a cursory glance, a good number who are unambiguously notable - regardless of their generous endeavours as Wikimedians - by virtue of being the authors of scientific papers published in reputable, peer-reviewed journals, some of which are cited in Wikipedias, not to mention in other scientific papers. Many of the people in your list are also the subject of multiple Wikipedia articles; in one case in 157 languages. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits23:43, 11 August 2018 (UTC)
Help: Vandalism
A number of properties and items are currently being vandalised. Please check [5]. I have reverted some of them. Please take action. John Samuel20:30, 17 August 2018 (UTC)
How authoritative is that? I was a member of this (defunct) organization, and while the Seattle chapter was one of the three or four most active, I'm pretty sure the organization was headquartered in the Bay area. Doug Schuler may have been CPSR's last president, and he lives in Seattle and was for many years our chapter president, but I don't remember the organization ever having been considered to be based here.
Just to be clear: main upshot of my question isn't this one statement but the reliability of data from GRID. - Jmabel (talk) 00:17, 7 August 2018 (UTC)
Well, in general I'd characterize GRID as pretty reliable, and improving over time. They respond to correction requests - I've sent them hundreds (often for duplicates based on wikidata matching). It's certainly the best openly available database of research-related organizations, but far from perfect. In this case, if you look at the wikipedia entry for CPSR you'll note it lists the headquarters as Seattle, so that at least agrees with GRID. We do have an issue with use of located in the administrative territorial entity (P131) vs headquarters location (P159) for organizations which would probably have been the better property to use in this case (but many items in GRID like hospitals, colleges, etc. really do have specific locations). ArthurPSmith (talk) 13:57, 7 August 2018 (UTC)
@Jklamo: In my experience GRID is fairly reliable - it does have errors, but which source does not? I would not call it "authoritative", as this database is itself derived from others. (So it is less authoritative than, say, a national company register.) For the locations, I would expect that most errors that end up in Wikidata come from reconciliation issues (primarily because of the cebwiki / GeoNames mess). − Pintoch (talk) 09:09, 12 August 2018 (UTC)
Wikidata:WikiProject WLE
Hi. I have left a discussion about monumental trees in Wikidata:WikiProject WLM but I realized we don't have a place for coordination about WikiLovesEarth. Shouldn't we have it? The workflow is not as intense as with WLM maybe, but I a was surprised to find nothing at all.--Alexmar983 (talk) 20:16, 11 August 2018 (UTC)
Count me in - the past months I was working on cleaning up the protected areas in my home area, and would be happy to have some formalized rules. E.g. how to best handle the common mash-up of the landscape, the nature preserve (sometimes split into several legal parts due to administrative boundaries) and the natura2000 site all covered in one Wikipedia article, whereas from a data point of view all parts need separated items. Ahoerstemeier (talk) 19:14, 12 August 2018 (UTC)
Ahoerstemeier the big work on WLM ends in October-November. We can think about this at that point (if noone else did). We just put an introductory page with a list of common IDs for parks and nature, than we ping in the talk page some WLE organizers (author of pages on commons). You can do it yourself tihgt now, i can hel you with the ping if you create the page.--Alexmar983 (talk) 20:29, 12 August 2018 (UTC)
@Abductive: apologies (and welcome!), the link in my reply is a way of doing what U+1F360 suggested for catalogs that don’t have yet a dedicated property in Wikidata.—-DarTar (talk) 21:12, 12 August 2018 (UTC)
I am frequently creating items for groups featured as "companies" in the database MobyGames that publish or develop video games. However, according to the descriptions in the articles, these companies are often not separate businesses but are a named part of a different business that already has a Wikidata item. It seems appropriate to me to have a property linking the two items together, but I have been unsure what the best property or properties are for this purpose. I have been using the property pair has subsidiary (P355) and parent organization (P749), but it is my understanding that that subsidiaries are defined to be separate companies on their own right that are owned by their parent companies, and I don't think many of the groups I find fit this description. Often times Mobygames lists the group as a "publishing label" or "label", but the property record label (P264) seems to apply only to music. Other times the group is simply listed as a "team" or a "name" that the bigger company uses for publishing games. I thought maybe I could use business division (P199) as a all-purpose property, but it doesn't appear to have a matching property "division of" - both of the examples mentioned in the property are also listed as subsidiaries of their parent organizations, leaving no guidance for business divisions that AREN'T subsidiaries. Should I simply be using part of (P361) and has part(s) (P527)? Rampagingcarrot (talk) 19:03, 12 August 2018 (UTC)
@Ghouston: Thank you, that makes sense. I had worried that people stumbling across some of the more obscure groups may not be able to easily understand they are part of better known organizations - some of the groups have no description on their accompanying MobyGames article, and it would take me time to hunt down which organizations they belonged to. But it sounds like you are saying that making things bi-directional in these can cases can be problematic, and it might be best to simply do business division (P199) on the "parent" business. Rampagingcarrot (talk) 21:12, 12 August 2018 (UTC)
Yeah, business division (P199) alone is sufficient. It's a bit confusing because Wikidata is actually a database, not just a way of constructing nice-looking item pages in a user interface. The item pages only show the claims where that item appears on the left-hand-side, and ignore all the equally valid claims where the item appears on the right. Ghouston (talk) 21:53, 12 August 2018 (UTC)
You might also consider imprint (Q2608849), "trade name under which works are published; a publishing division of a publishing company", assuming we're agreed that video game publishers are publishing houses (not sure how video game editors treat that). I tend to draw a distinction between a business division with its own leadership and formal structure, and a brand that gets slapped on products for a certain market sector when there is no separate corporate division in charge of that brand - understanding that it's often hard to tell the difference from outside the company. - PKM (talk) 21:24, 12 August 2018 (UTC)
I don't know whether this is the proper place to raise this issue, but there seems to be a conflict whether polyploidy (Q213410) and polyploid (Q44445817) should be merged or not. Initially I thought it was a nobrainer, so I merged them. This was reversed by Andreasmperu. Since than, I have tried to enter into a meaningfull discussion here, but seem to hit a brick wall. As far as I can see, the topic discussed is exactly the same on all Wikipedia's, so no need for two items here. Andreas has not been able to give me any pointers on why it should be otherwise, but maybe here someone can, or otherwise can persuade Andreas to talk. BoH (talk) 21:23, 4 August 2018 (UTC)
English labels, English descriptions and statements explain the difference between the two items. --- Jura22:44, 4 August 2018 (UTC)
In that case the labels are wrong. Have you taken a look at the Enlish article. If you do, the issue should be resolved quickly. BoH (talk) 22:58, 4 August 2018 (UTC)
The question is if the sitelinks are on the correct items depending on how are they are titled in each language. Please avoid re-purposing items to match some Wikipedia language version. --- Jura23:00, 4 August 2018 (UTC)
I am not repurposing anything. There is an invalid statement on polyploid (Q44445817): polyploid is not a thing, it is a trait. As such, it is an adjective of the noun polyploidy. If you take a look at all the articles involved, you will see they belong together. BoH (talk) 23:07, 4 August 2018 (UTC)
Have to say that BoH makes a good case; and for me it is telling that there are no language wiki overlaps amongst the sitelinks, just as if a single concept had been split between a redundant pair of wikidata items. --Tagishsimon (talk) 23:26, 4 August 2018 (UTC)
I have long found the en.wikipedia redirect of "polyploidy" to "polyploid" to be unhelpful, but everything is such a mess I have never gotten around to fixing them. People regularly say an organism is "a polyploid" in English. They should not be merged, any more than tetraploidy (Q453166) should be merged to tetraploid (Q16610359). Abductive (talk) 07:05, 5 August 2018 (UTC)
So you say that de:Polyploidie is a different subject than en:Polyploid and should have no link to each other? Read the articles and see that that is not the case: they discribe the same thing. Just like your example of tetraploidy and tetraploid by the way.
Note that the current article en:Polyploid does not describe a polyploid, but starts the definition with a cell with a polyploid trait.
Also note that as Tagishsimon aready mentioned, there are no language overlaps. If these two should really be separated, how come that there are no separate articles about them?
Maybe it is helpful to change 'polyploid' to 'red', so:
Polyploid cells and organisms are those containing more than two paired (homologous) sets of chromosomes.
becomes:
Red cells and organisms are those containing red chromosomes.
If that was really a thing, in the field you would probably talk of 'reds' if you describe organisms with red chromosomes. It would not justify a separation however. BoH (talk) 12:32, 5 August 2018 (UTC)
Generally, there are three different questions:
(1) Do the items represent different concepts or different aspects of a concept?
(2) On which item should sitelink to language ABC go?
(3) Does Wikipedia in language ABC want to have a sitelink to language XYZ?
Whether all sitelinks would fit on a single item isn't relevant (2) and (3). That a page in language ABC mentions, describes or discusses all other concepts, doesn't answer (2). (3) can be answered outside Wikidata and can be achieved with local sitelinks or a LUA module. --- Jura14:15, 5 August 2018 (UTC)
Indeed we come across different questions. As a contributor to Wikipedia, I care less about into how much detail Wikidata wants to go. My main concern is the sitelinks. In the past, as everyone knows, sitelinks were taken care of by the Wikipedia's themselves. If that would still be the case, there would be no issue: it is very obvious that the topic discussed is only one topic.
It seems that in this case, we fall victim to some form of hypercorrectness, where 99.9% similarity is insufficient for some to merge.
That might seem to be the case, but it really is not. There is simply a mistake made by giving polyploid (Q44445817) a definition for the noun polyploid and than adding sitelinks to Wikipedia's that use polyploid as the adjective for polyploidy. BoH (talk) 14:34, 5 August 2018 (UTC)
Support merge. Useful to look at these in Reasonator, polyploidy (Q213410) and polyploid (Q44445817), to see incoming links to the items, as well as the statements on the items directly. Two things to think about here: (i) sitelinks (ii) the web of statements on Wikidata. In respect of (i): having all the sitelinks together would be a good thing. In respect of (ii), we need to look closely at the incoming links to the items, to see if they create a critical need to keep a distinct item. In this case it is clear that if we merge polyploid (Q44445817), we would also need to be merging tetraploid (Q16610359), Triploid (Q44449573), allopolyploid (Q9607621), and allotetraploid (Q25532160) to their corresponding items under polyploidy (Q213410). But it looks as if the merger would create no difficulties beyond this, so should be clear to go. A final thing: it is important that the labels in all languages get updated to reflect their new item -- eg Dutch in particular would have to be changed from the equivalent of xxx-Ploid to xxx-Ploidy. But, *provided that this is done*, I see no reason to object to the merger. There's no particular reason to insist that the label in xx-language of the item here 1:1 matches that of the article in xx-language wiki. What is important is that the sitelinked articles cover the same subject and ideas; and that the label of the item here in xx-language corresponds to ontology of the item here. It's a general point: because of the desire to maximise sitelinkage here, the ontology of an item here may not 1:1 match the ontology of the article in xx-wiki, so long as they cover the same subject. That's okay, at least in my opinion. We're trying to build a workable wikidata ontology here, with sitelinks to as much as is relevant, not slavishly trying to describe the ontology of each and every xx-wiki. Jheald (talk) 19:47, 5 August 2018 (UTC)
Indeed. If that were not the case, the Wikipedia's should abandon Wikidata as central point for sitelink and probably have to go back to the old interwiki's. I do understand the desire here on Wikidata to be as precise as possible, but one should not forget what are the means and what are the ends. BoH (talk) 20:04, 5 August 2018 (UTC)
That is the difficult thing, he is not taking part in any meaningful discussion, but still reverts. On a Wikipedia that would disqualify one from taking any reverting actions, but I don't know enough of the norms here. BoH (talk) 12:46, 7 August 2018 (UTC)
Even if one places all sitelinks on a single item, there isn't really a need to merge the items. What do the related WikiProject participants think? --- Jura13:00, 7 August 2018 (UTC)
I think this is the best option. It recognizes that on the topic discussed on all articles, the trait is the primary issue, while the cells and organisms with this trait are secondary. But it still allows room if someone ever would like to write a separate article about polyploid cells and organisms. It is an option I already performed, but this was reverted as well. BoH (talk) 16:07, 7 August 2018 (UTC)
My view is that we should not merge polyploidy (Q213410) and polyploid (Q44445817) but think instead about better ways to handle interwikis for such cases, which are collectively referred to as the Bonnie and Clyde problem, for which there is a dedicated WikiProject that is exploring potential solutions systematically. I suggest to move this discussion there and see whether this particular case is covered by the subcase typology that has been elaborated there. --Daniel Mietchen (talk) 19:34, 8 August 2018 (UTC)
I don't think this is a Bonnie and Clyde problem. All articles are about Bonnie and on the sideline mention Clyde as well. We can keep Clyde, but have all sitelinks on Bonnie, as Jura suggests above. BoH (talk) 21:42, 8 August 2018 (UTC)
There was no "full query request logs" published, please do not say that, it is misleading. We have discussed it in https://phabricator.wikimedia.org/T197777 and I though I have provided information about this. What was published is (valid) SPARQL queries, with SPARQL query text was sanitized to exclude any information that could potentially identify any person or contain any personal data, and broad classification of request agent (e.g. browser or bot).
It is not possible for anyone to "opt-in" exactly because there is no link between person (or user) and the query. If you said "please opt-in all my queries" (or "opt out") we couldn't because we do not know which queries are yours! Additional query processing was done in case there's some clever way to use information inside queries (like coordinates, or names, or text search strings, etc.) to link them to any person. This information was removed.
As I said on the ticket, if you have a concern that any PII might have still left (or even potentially left) there, please identify it. So far nothing of the sort has been brought to our attention. We are talking about SPARQL queries to public database. Note that https://quarry.wmflabs.org/query/runs/all lists all queries on Quarry, as far as I can see, while also identifying the user running them, and so far nobody has expressed any concern whatsoever about it. We are releasing very strictly sanitized data which has no link to any personal data at all. I'd like to know the source of the concerns.
As for setting up mirrors - Wikidata Query Service is an open source software, and you are certainly welcome to run it, including with Wikidata data mirror. I've made a number of improvements (relatively) recently to make it easier, and there's excellent work implementing docker install for WDQS. There are docs on how to run your own instance and on how to run standalone install, and I welcome you to reach out to me if you need any help with running WDQS, anytime. Smalyshev (WMF) (talk) 03:56, 9 August 2018 (UTC)
Normally, I'd have expected this to be announced in advance and given the users the option to opt-out or not opt-in to the publication. I think WMF should provide a service that isn't publishing full query logs. --- Jura13:49, 10 August 2018 (UTC)
I think there's misunderstanding here. Again, nobody is "publishing full query logs". And I am not sure how could any user opt out if we don't know which user the query belongs to? I think we're walking in circles here. Smalyshev (WMF) (talk) 21:26, 13 August 2018 (UTC)
Design advice for church parish start and end and name changes
One church parish can change name at a specific point in time. Question: What is the best way to model that?
I can see that Wikidata has plans to support contemporary constraints Task T141859 and contemporary constraint (Q25796498) if this is done we solve a never ending discussion we have in sv:Wikipedia when a person has an "event" and is connected to a parish and the comment is that the name is wrong it didnt exist at that time ==> we need to modell it in Wikidata correct to support contemporary constraints
I'll provide a report about the contemporary constraint when I finish its implementation. I fear I don't fully understand the problem you are trying to solve, but I'll try to answer some of your questions anyway. To determine if a violation exists or not, the most permissive values are considered. This rule is applied to all cases, including low precisions, lack of statements for start/end points and special snaks (somevalue and novalue). When a start or end time isn't defined or has somevalue or novalue snaks, it's considered -∞ or +∞, respectively. However, you shouldn't use somevalue if you know the date with a certain precision; for example, if you know a value is in the 15th century, you should use that value, 15th century, with the precision of centuries, and the same for millenia, 10,000 years, million years, etc. It's usually a bad practice to use somevalue for dates. --abián09:57, 11 August 2018 (UTC)
Thanks. The problem we have seen is
The Swedish National Archives has been writing bios for people in Dictionary of Swedish National Biography ID (P3217) and in the old days they used names of church parishes that were not contemporary with the name used when the person lived
We are now adding WD driven templates to sv:Wikipedia and many people complain when they see the wrong naming for a church parish that today has another name as it had 200 years ago
, but she actually attended it when it was known as the "Inns of Court School of Law", which has since been merged into another institution and renamed. enwiki considers it to be a single entity with inception 1852. I'm sure there are countless examples in Wikidata when entities change name. Ghouston (talk) 01:10, 12 August 2018 (UTC)
Thanks it sounds logical it will also impact how we have written our sv:Wikipedia articles (sometimes a name change of an old church parish has got a new article) - Salgo60 (talk) 10:22, 12 August 2018 (UTC)
The whole question of continuity over time is quite difficult, because you have a mix of continuity and discontinuity, and at particular events you can deal with it either way (with separate items or a combined item). We also have that decision being made independently on different projects, so can even have both types of Wikidata items for the same real entity. On en:Margaret Thatcher, the infobox lists her alma mater as "Inns of Court School of Law", which redirects to the City Law School article. In theory, you could look at Thatcher's dates of education and the official names in City Law School (Q5123253) and write a script that would give the appropriate name, but it would be complicated, and you'd have to set up all the data properly to make it work. Ghouston (talk)
Add a date before year 1600 ?
The problem is that we in the Swedish Wikipedia article has parish from Middle Age and I guess we lack good sources from this timeperiod....
Best would be in WD to state something like before 1600
The answer is "Male", with the definition being the description from Wikidata item Q6581097, "human who is male (use with Property:P21 sex or gender). For groups of males use with subclass of (P279)"
Difficult to see how that description is compliant with Help:Description. We have a sufficiency of examples of good practice, and sufficient users running reports looking & dealing with abarrent values, that we might now think about killing such descriptions with fire upon sight. --Tagishsimon (talk) 21:28, 11 August 2018 (UTC)
@Tagishsimon: you might not know about the phab ticket and previous discussions on the topic. We came to the conclusion that these can't be avoided for now. --- Jura21:41, 11 August 2018 (UTC)
Thanks Jura. So Phab T97566. I see in it reluctance to implement a method of surfacing P2559 data, and a presumption that instructions cannot be removed until that method is implemented. I see no evidence that anyone has considered/measured whether instructions in descriptions work. I note comments that instructions are often only in one of the many description languages - suggesting that they are de facto not considered that important. My view has not changed: instructions should not be in descriptions, or, at least, not in item descriptions. --Tagishsimon (talk) 22:03, 11 August 2018 (UTC)
@Tagishsimon: It shouldn't really be needed, but the substitute hasn't been implemented yet and re-users were advised to filter it differently. For P21, we could probably scrap it as the one-of property constraint offers a valid alternative. --- Jura22:17, 11 August 2018 (UTC)
Jura1 has reverted both changes, plus those made by colleagues in other languages after mine (which were in English). There is nothing in the cited Phabricator ticket (which is three-and-a-half years old, and has not been edited for over a year) which precludes my changes. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits21:59, 11 August 2018 (UTC)
┌────────────────────────────────────────────────────────────────────────────────────────────────────┘ Because - as I indicated - that does not preclude the changes I made, as discussed above. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits22:16, 11 August 2018 (UTC)
┌────────────────────────────────────────────────────────────────────────────────────────────────────┘ I was and am aware that the delay in making use of P2559 in Wikidata's interface in no way precludes the changes I made, for the reasons discussed above. And don't edit the section heading here, which is both accurate, and used in links from elsewhere. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits23:09, 11 August 2018 (UTC)
@Ghouston: it's most likely an entity resolution problem by Siri (either: 1. matching the word "angel" to some other Wikidata entity, or 2. performing some kind of other inference in its own knowledge graph, unrelated to Wikidata, concluding the answer is "male", and returning the corresponding description from Wikidata).--DarTar (talk) 17:42, 12 August 2018 (UTC)
Not really. We haven't identified the source of the error. Even the amended description isn't correct in the context (provided they be male). Maybe Siri isn't really suitable for the purported purpose. Another broken search engine? --- Jura02:59, 13 August 2018 (UTC)
┌────────────────────────────────────────────────────────────────────────────────────────────────────┘"We haven't identified the source of the error." Poppycock. In the case of the definition of "male", the error is ours. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits07:49, 13 August 2018 (UTC)
When creating items with names that are shared by many other items, I have been creating an alias by appending the description to the name (for instance, Twilight (Q7857976) has the alias "Twilight video game developer"), which has allowed me to easily recover the item when linking claims to it. I thought perhaps this was OK given that the aliases don't have references attached and seem like an informal way to help users identify an item. However it occurs to me now that perhaps I should check to see if I should stop doing this. I just found the page Help:Aliases and on the discussion page there is a topic where people seem to agree that aliases can be used to expand the name of a geographic location, which seems somewhat similar to what I am doing, except names like "Portland, OR" could still be considered a valid name whereas "Twilight video game developer" not so much. So with this knowledge I think I will stop doing this. Should I try and go through my contribution history to remove all of these aliases?
Assuming using aliases to disambiguate item names is not acceptable, are there any plans to make finding items with very common names less tedious than clicking "more" several times? On the "Help:Aliases" page I saw that a search engine upgrade was planned to add "fuzzing searching" - that edit was made four years ago, is this still planned? It seems ideally that descriptions would be added to the search engine. I know that I can do a query if I wanted to find the correct item, but that also seems cumbersome. Rampagingcarrot (talk) 05:20, 13 August 2018 (UTC)
@Rampagingcarrot: I'm not sure if there are any actual policies about this, but I think a disambiguation as an alias would be fine if it seems like a plausible and correctly-spelled Wikipedia article title or redirect; e.g. "Twilight (video game developer)". Jc86035 (talk) 16:51, 13 August 2018 (UTC)
I disagree. Aliases are pretty much useless as disambiguators, not least since they are not for the most part surfaced in search results. The Description field is the place to provide sufficient information to disambiguate - see the first sentence of Help:Description. In the instant case, Twilight's description preumably identifies that they are a video game developer. "Twilight (video game developer)" is not an alternative name for "Twilight"; it is a disambiguator, and I cannot see any support in Help:Aliases for using aliases to disambiguate. Should you go through your contribution history to remove all of these aliases? Yes. --Tagishsimon (talk) 17:01, 13 August 2018 (UTC)
Aliases affect the search results, descriptions don't - the descriptions help you distinguish the items you're shown, but they don't affect whether the item you want will appear or not. I don't think it's unreasonable for there to be a unique-enough alias which will return an item on the first page of results, so that people don't have to repeatedly load more results every time they want to use an item. - Nikki (talk) 18:28, 13 August 2018 (UTC)
I see where you're coming from, but A) you're 100% wrong about descriptions not affecting search results - try this search, for instance: https://www.wikidata.org/w/index.php?search=&search=Film+von+Ridley+Scott+%281982%29 or do your own experiments on searching for things found in descriptions; and B) that way lies madness of a couple of different sorts: 1) it is not our business to start coining new aliases for people/things, but to reflect the aliases already used by others and 2) if we are going to invent new aliases to satisfy the plurality of searches that are made, then we must gear up for the countless variants one can conjour up for those people and things: for hyphens instead of parenthesis, for 'game developer'instead of 'video game developer', dev instead of developer, &c &c &c. --Tagishsimon (talk) 20:55, 13 August 2018 (UTC)
@Tagishsimon: It looks like there is some confusion about what is meant by "searching". Up until now the way I've looked up items is by typing the item name in the search bar in the top right and looking through the suggested results WITHOUT pressing enter or the question mark first. Then I would click on the suggested result I was looking for, never having reached a search result page. I suspect this is also what @Nikki: was talking about. In my experience, typing parts of the description in the box doesn't show the desired results, while typing the alias does. I never realized that pressing enter gives different results than the suggested results that are shown. I don't know if this is made clear to new users.Rampagingcarrot (talk) 22:13, 13 August 2018 (UTC)
Yup, autocomplete - which is what I understand the drop-down suggestions are called - pays no heed to descriptions (as far as I know ... which is not terribly far.) But search results do. And I tend to think nothing is made clear to new users; it's rather like rule 1 of Fight Club (Q190050). --Tagishsimon (talk) 22:25, 13 August 2018 (UTC)
@Tagishsimon: Yes autocomplete seems like the correct name. Now that you've shown me that the actual search engine does take descriptions into account, I can see why putting descriptions in the aliases is even less appropriate. Rampagingcarrot (talk) 22:31, 13 August 2018 (UTC)
Help:Alias is somewhat outdated so it might not be of much help. If the company is referred to such, I think it is a useful alias. --- Jura17:13, 13 August 2018 (UTC)
Ships
Is it possible to add a property class to ships of Ship launch date, this is the key disambiguation datum for ships of the same name in en:wiki. Thanks. Broichmore (talk) 21:51, 13 August 2018 (UTC)
Hi, everyone! September 17th we will have the tenth Wikidata Lab. This time the event will discuss how to develop bots on Wikidata. The event is free and open to all, but by limit of vacancies, prior registration is required. If you are near São Paulo, Brazil, please join us! If you can't come, it's fine! There will be live streaming and the presentation will be available at Neuromat's channel on Youtube.
This is the tenth activity of a series of trainings for the integration of the projects Wikidata and Wikipedia. The presentations, photographs and impact reports of the first nine activities are available for consultation in Wikidata Lab I, II, III, IV, V, VI, VII, VIII and IX, respectively.
The "Wikidata Lab X: Writing Wikidata bots" will occur in the University of São Paulo, on September 17th (Monday), from 9:30 am to 4:30 pm BRT
When adding items for companies I have been using the name given in whatever source I am primarily referencing as the label. Sometimes this source has Ltd. (for British companies) or Inc. (for American companies), and other similar suffixes for other countries at the end of the name; sometimes there is no suffix. If the item for the company already exists but the label has the suffix and the source does not or vice versa, I have been adding the source version as an alias. Is this the correct approach? Rampagingcarrot (talk) 22:05, 14 August 2018 (UTC)
@Mahir256: Thank you, I will put names with suffixes as aliases. Do you happen to know what the proper procedure is for "official name" in regards to region-specific suffixes? Or should I ask as a new topic? Rampagingcarrot (talk) 04:17, 15 August 2018 (UTC)
We had a productive “strategy meetup” at Wikimania with a group of about 20 people, to talk about the future of WikiCite and a roadmap for source metadata in Wikidata more generally. The motivation for this meetup was a set of concerns around scalability and “growing pains” around bibliographic and citation data in Wikidata, as well as the need (that many in the community have expressed) for a clearer goal, value proposition, and scope for WikiCite. The result is a series of notes fleshing out 4 possible scenarios for the future of bibliographic data as structured data—from a centralized scenario to a fully federated one—discussing their possible risks and benefits at the technical, social, and governance level. The question these notes try to address is whether Wikimedia should aim to build its “bibliographic commons”, and if so, what it would look like, and where it should live. This document is not a formal proposal or an RfC open for a vote, but a conversation starter to evaluate what type of future makes most sense for this data and the communities and stakeholders that will benefit from it. A shared understanding on what we’re building towards is also going to help us inform the program of the upcoming WikiCite 2018 conference in November (the application process will open in a few days). If you wish to share your thoughts on these four scenarios, please chime in on this page.--Dario (WMF) (talk) 03:10, 15 August 2018 (UTC)
Property constraints – music releases
Should both AllMusic album ID (P1729) and AllMusic song ID (P1730) be allowed on the same item? AllMusic has album identifiers for singles, and currently items about songs released as singles seem to be about both the song (the musical work) and the single (the product/release containing that musical work and maybe another). The alternate way of modelling this would be to treat all singles as albums and prevent songs from being also classified as singles, which would require the creation of a lot of items. Jc86035 (talk) 19:39, 5 August 2018 (UTC)
@Valentina.Anitnelav: Yes, I thought so too. I've been trying to model (and source) Eastside (Q55975144), and so far I've made items for the song itself, its music video, and the single releases on iTunes/Apple Music and Spotify. To me it would be logical to have items for each copy of the video (on YouTube and iTunes), each release of the single (on iTunes, Spotify, ...), the music video, the single and the song. I'm not sure whether the release on YouTube Music counts as a video or an audio track. Jc86035 (talk) 07:19, 6 August 2018 (UTC)
(I've assumed it makes sense to have separate items for the iTunes and Spotify singles because they have different metadata and the iTunes single is stated to have been released a day later. Same reasoning for the videos; the YouTube video is also slightly longer. The iTunes music video is pending a property proposal.) Jc86035 (talk) 07:25, 6 August 2018 (UTC)
@Jc86035: I never thought about the modelling of different distributions and at this moment it does not seem necessary to have different items for these, unless there are other differences (sometimes releases differ in the tracks they contain). I think one could just put the release dates on the single item, maybe with qualifiers (as done for most films).
@Valentina.Anitnelav: I considered merging the iTunes and Spotify items, but I think I didn't know how to model "iTunes" and "Spotify" into the qualifiers for date and so on (I had to undo the merge). It does make sense to make additional qualifiers for separate recordings of a song, but how would this be structured? Would the original recording be the canonical version (with modified version of (P5059)), or would the song itself be a separate entity from both versions (with edition or translation of (P629))? I think either approach would make sense depending on when each version is recorded/released; if they're released at the same time then edition or translation of (P629) might make more sense. Jc86035 (talk) 09:02, 6 August 2018 (UTC)
Okay. I've merged the items for the single into one. I don't think there's anything wrong with it right now. Jc86035 (talk) 12:47, 8 August 2018 (UTC)
┌────────────────────────────────────────────────────────────────────────────────────────────────────┘ @Valentina.Anitnelav, Mahir256: Let's say I want to represent the release history of Popular Song (Q7229734), which currently has one item representing the song (sung by Mika (Q186329) and Muni Long (Q7245622)) and the re-recording (sung by Mika and Ariana Grande (Q151892)), as well as the single (which consists of the latter). I'd also want to model Popular (Q3908430) (the song from the musical Wicked), since it's sampled prominently in "Popular Song".
I'm guessing there should be one item each for "Popular Song", the re-recording, and the single, modelled as described above.
@Jc86035: Some preliminary thoughts on your questions (numbered for clarity):
Yes; a single release may have B-sides or multiple remixes or edits of a song that themselves may be included in other albums or compilations.
The canonical version should be the composition by Stephen Schwartz, since those who perform Wicked on stage may not necessarily wish to base their stylistic choices on a specific performance of the musical.
If it is uniquely identified by a given ISRC (P1243) and MusicBrainz recording ID (P4404), then yes. (We probably should reserve P31 "song" for the composition and not its recordings.)
In this case we may wish to add a "samples" property to make this relationship clear (something especially useful for hip-hop tracks; this also could permit a "whosampled.com ID" for individual songs).
I would defer to what the official credits for a given song say, in an effort to express relationships similar to how, for example, Stan (Q312122) was 'written' in part by Dido despite the relevant portion of that song being ripped from Thank You (Q1635376).
@Mahir256: Treating e.g. "Popular Song" as a composition separate from the original recording (instead of just as a "song") really complicates things, I think, since covers which aren't note-perfect facsimiles would also have to have an item for their arrangement. Is the cover then a work based on the recording or the composition? I think arguably they should be represented as one and the same because the production of a song might not be "recorded" in the literal sense of being based on a real recorded sound. Jc86035 (talk) 16:02, 8 August 2018 (UTC)
@Jc86035: I'm quite torn between the ideas of collapsing the levels of work and recording for (most) songs in popular music and keeping them distinct. There are definitely advantages concerning cover versions (for some it is quite a stretch to call them just recordings of an existing work, as there might be considerable changes (genre, tonality, etc.) and it is actually hard for most songs in recent popular music to refer to a "work" apart from the recording (the vagueness concerning those features as tonality, rhythm, etc. (if abstracted from a certain recording/performance) and the relevance of improvisation contributes to this)). But there are also disadvantages: We would need to treat classical music (and maybe songs with a traditional songwriter) and popular music differently and we would need to call the recording of a live performance of a band of a song previously recorded by themselves in studio a "modified version". There is also the question of the identification of composer/lyricist: would we want to indicate them at every version - then it might be better to have an own composition item? Or would we just indicate them at the original version, maybe indicating arrangers at the other version via adapted by (P5202)? As to your questions:
I agree with you and Mahir256: There needs to be at least an item for the recording-work ("canonical version"), the rerecording ("different version") and the single (depending if we want a uniform way of modeling classical music and popular music there might be the need of a work/composition-item)
There should be definitely a separate item for the composition by Stephen Schwartz and each recording/performance should be distinct from this item. In most cases a separate item for the performance of a song in a musical might not be needed, as it is sufficient to have an item for the performance of the whole musical with cast information and to indicate the character featured in this song to deduce that a performer performed a song on stage. But if an item for a recording is needed (e.g. to specify that this was sampled in another work or to link it to a release) it should be created.
Popular Song (Q7229734) is currently a single. If one creates an item for the song one could think about collapsing this with the recording.
There is has melody (P1625) that could be used to indicate that Popular Song (Q7229734) (on the composition level) uses a melody from the composition of Stephen Schwartz (but this would be just on the composition level). To indicate that Popular Song (Q7229734) (the recording/performance) uses actual sound from the recording by Kristin Chenoweth (Q231811) another property might be needed (If a new property is created this could also be used to indicate that a certain track is used in a film, when there is no soundtrack album) - Valentina.Anitnelav (talk) 19:58, 8 August 2018 (UTC)
@Valentina.Anitnelav: I agree with you and Mahir256 on points 2 and 4. For point 3, I would keep all the interwiki links on the item about the original song recorded with Renea, and split the single and the version with Grande off into new items, since in my enwiki experience (from substituting Template:Infobox single (Q8999420) with a bot account) it's very rare for an article to be about the single release rather than the song unless there are two or more A-sides (which is also quite rare outside "single albums").
I think it's interesting to note that "a song by" usually means the songwriter only in enwiki articles for songs up to about 1960, or otherwise for songs where the sheet music isn’t just a partial transcription of the recording and/or was released before any recordings (e.g. Palladio (Q19059210)). There would probably be some edge cases (I think classical pieces published as recordings first would have to be both recording and composition).
If nothing else, recordings of live performances by the original artist could use a specific Wikidata item ("live recording of a song"?) which is classified differently in some way to cover version (Q155171) and live album (Q209939). For performances themselves, I think maybe a new "setlist" property (like tracklist (P658)) could be used to avoid creating new items for performances which weren't commercially released and don't have their own identifiers. Jc86035 (talk) 08:26, 9 August 2018 (UTC)
Before I begin: Classical music is "it's own thing" and we shouldn't try to model pop songs after how classical music is treated or vice versa. There are obviously some gliding, overlapping areas between the two but I think that belongs in a separate discussion (or discussions). I come from the world of popular music, and will try to limit my comments to that field. Musical works, recordings and releases are three well established, well understood, separate things. That WD often conflates song and single I view as a legacy issue stemming from Wikipedia, and something that slowly will disappear over time as we the editors properly split these items up into their proper parts. What I personally would really, really like to see is the inverse of recording or performance of (P2550), something like a "has recordings" property to use with musical works, to properly model both ways of the relationship. Should both AllMusic album ID (P1729) and AllMusic song ID (P1730) be allowed on the same item? The answer is no in my opinion, they clearly represent different things. We should encourage entering data "properly" and strive to correct historical entries. I apologize for not answering everything but I need some more time to parse through this large (but very interesting!) discussion Moebeus (talk) 15:55, 14 August 2018 (UTC)
Inverse properties
On the whole thing about using explicit inverse properties: can someone explain to me why there is so much of this in Wikidata? Is it just that it would be so computationally intensive to find the incoming references to an item? If so, couldn't we still only explicitly maintain one side of such a relationship and have a notion of certain properties having an inverse, so that when the property is added, modified, or deleted the inverse is automatically maintained, rather than someone needing to do so explicitly? (An approach like that would also guarantee consistency.) - Jmabel (talk) 20:45, 14 August 2018 (UTC)
I think it could be just part of a broader norm/tendency to add as much data as possible to prettify items. For example, at time of writing Alexandria Ocasio-Cortez (Q55223040) has somewhat unnecessary qualifiers for place of birth (P19), since in its existence the Bronx has always been part of both New York and the US.
As another example, MusicBrainz recording ID (P4404) was originally supposed to be used as a qualifier for tracklist (P658) (and I have done so as well), even though this shouldn't actually be necessary since each song/track item should have at most one (barring MusicBrainz errors), or at worst zero (which makes its use as a disambiguating qualifier sort of useless); similarly for series ordinal (P1545) on part of (P361) qualifiers for songs/recordings, which duplicate tracklist (P658) qualifiers on albums, singles and their versions. Jc86035 (talk) 05:14, 15 August 2018 (UTC)
Let's also assume that I'm not going to bother doing any more tracklist (P658) things because it takes too long and MusicBrainz doesn't even have identifiers for all four of Adele's single releases.
Wikipedia says the Joel single was actually the first release, as Joel's album was released on 19 August 1997 (no single date given), before Dylan's album was released on 30 September 1997 (single was apparently released in 2010). Is the song still a Dylan song?
How is "Make You Feel My Love" (Live At Hotel Cafe) classified? Is it more than a version, edition or translation (Q3331189)? Do we need to create an item for tracks which were recorded live and are versions of another song (i.e. tracks on a live album (Q209939))?
If "Make You Feel My Love" (Live At Hotel Cafe) and the studio version were released at the same time, is the studio version automatically the canonical version by virtue of not having a parenthetical disambiguator on its track name?
Does Wikidata need to record track names like "Make You Feel My Love (Live At Hotel Cafe)", or should they only be in subject named as (P1810) qualifiers to references while the actual name stays in title (P1476)?
There are no credits on Adele's music video (at YouTube, at least) that say that it uses the live recording. I had to figure it out based on the differences between the tracks. If I can't find a source for this, will it eventually have to be removed?
Is a vinyl record (Q178588) automatically assumed to have a certain diameter and rotational speed? There is an item for the 12" single (Q918887), but not for any other sort of record. Should this information be in qualifiers for distribution format (P437) statements, or should separate items be created?
If one takes the definition of a cover version (too) literally one might get the idea of Joel's version being the original and Bob Dylan's the cover, but I actually don't find anybody claiming this and this seems rather counter-intuitive to me (he wrote the song as a singer-songwriter with intending to perform it himself), but this is still an interesting case.
There might be some special cases but normally (also in this case, as it seems to me) the studio version should be the canonical version as this was developed under controlled circumstances, best representing the artists/composers/producers conception of the song.
For #2, I think it is best to distinguish an arrangement from a recording: the two Adele recordings (the "cover" and the "live version") are much more similar to each other than they are to the Dylan recording (intro, key, melody, singer's timbre, tempo, instrumentation, ...), so the "live at hotel café" version should probably be stated to be a recording of the cover.
For #4, adding the parenthetical doesn't really help because there would still be at least thirteen items related to the song titled "Make You Feel My Love" (the song itself, three covers, four singles, the four distinct versions of the Adele single, and one music video). The parentheticals can also vary across services (e.g., on Spotify the track on the single is named "Make You Feel My Love (album)"), so there would have to be multiple entries for differing bits of titles which should really be in a separate metadata field.
For several years, I'm still wondering that why this is a pure string property, instead of a monolingual text property? Why someone says that the value of that should only be in one language (or even English only), even used in Switzerland related items, where Switzerland is clearly one of multilingual countries? Can we please create a new monolingual text property with same translations of this, and replace it? --Liuxinyu970226 (talk) 22:34, 9 August 2018 (UTC)
@Liuxinyu970226: I agree that this property should be monolingual text. Quite a few countries and territories don't monolithically use one language or even one writing system. Jc86035 (talk) 11:19, 15 August 2018 (UTC)
Head up: Mix'n'match sync
Mix'n'match sync to Wikidata seems to have been inactive for a while/some catalogs. It's now starting up again, so we might see a large (10-100K) influx of statements under the User:Reinheitsgebot account. Each catalog sync gets its own batch ID, you can revert the entire batch at EditGroups if necessary. Please let me know about any major problems, however, please keep in mind that the occasional erroneous edit is not something I can scale to fix. --Magnus Manske (talk) 13:36, 10 August 2018 (UTC)
What exactly does a sync do? I thought doing a match on Mix'n'match added the ID to Wikidata immediately. Is this going to re-add all the bad matches we already removed? - Nikki (talk) 19:24, 10 August 2018 (UTC)
@Magnus Manske, Nikki: Yes, I saw a number of errors I had made and then reverted reintroduced just because I hadn't made the fix on Mix'n'Match too (which I would eventually get to by checking the reports and synching. I'm afraid I don't see the purpose of this task - IMO if MnM and wd differ long after MnM was used to assign a value, a human should evaluate. --99of9 (talk) 10:52, 15 August 2018 (UTC)
Oh great. I've spent hours and hours removing IDs added to the wrong items (which usually come from Mix'n'match, since it seems to be full of absurd suggestions which people accept anyway) and now they're going to be re-added? What a complete and utter waste of my time... - Nikki (talk) 16:14, 15 August 2018 (UTC)
"Error: The two items cannot be merged because one of them links to the other using the properties: P1441, P674." I don't know how to deal with that. —— FireFeather (talk) 19:04, 14 August 2018 (UTC)
In zhwiki, the article content includes both the story and character, and it seems that no language separates these two into different articles. That why I want to merge them. —— FireFeather (talk) 03:21, 15 August 2018 (UTC)
Sitelink zhwiki to the story, which is presumably the more general topic. But just because we don't have a wikipedia article about something doesn't necessarily mean it isn't worth a Q-code. - Jmabel (talk) 05:27, 15 August 2018 (UTC)
It's a rather frequent question. If the article needs data from two or more items, you need to set the QIDs explicitly. --- Jura16:28, 15 August 2018 (UTC)
Could not save due to an error. The save has failed. This action has been automatically identified as harmful, and therefore disallowed.
It seems the previous link redirected to the link you wanted to add so I have changed it. Your edit was disabled because it was considered harmful by the abuse filter. Esteban16 (talk) 23:57, 20 August 2018 (UTC)
AmaBhungane is an investigative journalism organization in South Africa. The name was chosen, because it means “dung beetle” in at least one of the dialects of isiZulu, the primary indigenous language of South Africa. As of 2018-08-21 Google Translate rendered it as “butterfly”. That's not the translation intended by the founders of amaBhungane. Does it make sense to try to enter this information about amaBhungane into Wikidata? Or just enter it into the Wikipedia article I'm writing about it? Thanks, DavidMCEddy (talk) 06:18, 21 August 2018 (UTC)
E.W. Hobson duplicates but there is a third option?
There are duplicates: E. W. Hobson (Q436096) (British mathematician, 46 statements, 10 sitelinks) and Ernest William Hobson. 1856-1933 (Q47475543) (Biographical article, 9 statements, 0 sitelinks). I think it would be better if there was an entry just named Ernest William Hobson? I'm not sure how to go about this, perhaps someone can step in please. Deadstar (talk) 12:32, 21 August 2018 (UTC)
No. The second item is an item for an academic journal article. It is a distinct thing; academic journal article <> person. They should stay as they are, linked, as they are, by a main subject (P921) statement. Their descriptions provide the necessary disambiguation. --Tagishsimon (talk) 12:38, 21 August 2018 (UTC)
Bodies of water connected by watercourse not notable in itself
If there are two lakes, for example, connected by a river that shows up on maps as a blue streak, but that is not shown on any map with a name or any other form of identification, how would one indicate that the two lakes are connected by a river?
I note that point three of the notability guidelines states that items can be created to fulfill a structural need, so creating an item for this river is an obvious solution, but is it the preferred solution? Is there some property that could be used instead?
If creating a new item is the preferred solution, I imagine the label of the item should follow a regular pattern and be something like "river connecting lake A and lake B", and that some coordinates should be added to the item in order to give it a more firm identification.
You can, yes, but should you? Probably not. Titles are important. The pattern "river connecting lake A and lake B" reflects common usage, as recommended by Help:Label. --Tagishsimon (talk) 20:58, 13 August 2018 (UTC)
So when the the drainage basin is not notable in itself, I create an item for that instead of for the river(s)? That has the advantage of requiring fewer items for terrain features not notable by themselves, but it does not provide the information about which water bodies are directly linked, nor about whether a water body is upstream or downstream of another. The latter is also a disadvantage of using connects with (P2789), unless there is a qualifier that can provide this information (e.g. using upstream (Q2440529) and downstream (Q15873020)). --Njardarlogar (talk) 19:30, 15 August 2018 (UTC)
How to record Oricon anime[BD/DVD/VCD/VHS]/comic/novel/music/etc sales data in Wikidata?
Oricon is a Japanese company that is responsible for collecting sales data from various different channels for various different media in Japan.
For different media, there are different way for them to release their collected data, and are often of interest to different people.
Take anime as an example, for number of discs sold, currently they would release weekly/yearly data about sales performance of each individual anime blu-ray disc volume, and weekly data for dvd discs, in addition to by-series weekly/monthly data release.
For each anime title, the first week sales performance of each individual blu-ray/DVD release are of interest, and then the latest total amount of disc being listed in the oricon data are also of interest. An example list of data for a single anime production could be re-represented like this on other websites:
○Title of Series
Vol First week Total Date
BD(DVD) BD(DVD)
01 15,000(*2,000) 18,000(*3,000) 10.09.11 ※Total 21,000
02 14,000(*2,000) 17,000(*3,000) 11.09.10 ※Total 20,000
03 10,000(*2,000) 13,000(*3,000) 11.10.09 ※Total 16,000
---
Avg 13,000(*2,000) 16,000(*3,000) ※Total 19,000
And then if there are special versions or sales of things like boxset then those data would also be added in. How should such data be stored inside Wikidata?C933103 (talk) 17:24, 21 August 2018 (UTC)
My first concern would be intellectual property & licences. This sounds like valuable commercial data. Have you given any thought to copyright & database rights? Under what licence, if any, is it released? --Tagishsimon (talk) 19:52, 21 August 2018 (UTC)
Humm indeed the original data release channel requires a paid subscription to access some of these data and the user agreement also stated that content should not be transferred to third parties. And the nature of the data is also more like an approximation based on collected data instead of direct reflection of collected data so there might also be some creative elements here that might make it eligible for copyright. It would probably be better to avoid meddling with those data here. C933103 (talk) 23:44, 21 August 2018 (UTC)
The Watchlist (top right link in the user interface) lists recent changes to entities you have "watched" - click the star between "View history" and the search box to add an entity to your watchlist. ArthurPSmith (talk) 20:36, 21 August 2018 (UTC)
What is the current situation with interface admins?
I have seen some discussions here and there, and I believe I participated in some of them, but I am not sure whether we have come to some consistent policy? As a frequent editor of MediaWiki:Gadget-markAdmins.js, I need to make sure that the workflow is not broken, and that the script still is editable by a number of people after the flag gets introduced next week. Also it would be great to have a policy on who and how gets the flag. Are we anywhere close?--Ymblanter (talk) 07:18, 22 August 2018 (UTC)
@Pigsonthewing: Vojtěch Dostál said in the second proposal "abandoning original request at Wikidata:Property proposal/Red List status of species" Thank you --David (talk) 05:20, 15 August 2018 (UTC)
I saw that. It does not explain why the first proposal - with all its discussion - has been withdrawn, and a second, duplicate proposal launched (other than "as advised by Succu"). Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits10:28, 15 August 2018 (UTC)
The first proposal started more or less as a RfC. After some discussion we have now a clear proposal to vote on. Thats the whole story. --Succu (talk) 19:58, 15 August 2018 (UTC)
"The first proposal started more or less as a RfC." No, it was posted as a property proposal, like many before and since. We don't "vote" on proposals, we discuss them. Note that the standard heading used is "Discussion" ({{int:Talk}}). It appears that your explanation is far from the "whole story". Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits15:59, 16 August 2018 (UTC)
I think we vote on property proposals. Preferable voting with remarks and comments which are leadinge sometimes to longer discussions. --Succu (talk) 20:11, 16 August 2018 (UTC)
LinkedIn personal profile URL
It appears that country-specific URL values for P2035 (P2035) now redirect to a more standard form; for example:
If it was accepted and returned, are start time (P580) and end time (P582) sufficient? Could the reason for the return be stored in some way?
Refused
I wouldn't like the awards to be removed from entities who refused, because the institutions giving them out deemed the recipient most worthy, so that's something which should be recorded in any way. But I think the refusal to accept should be noted as well. Ideally, the reason for the refusal could be stored.
Not Present at Award Ceremony
In addition, is there a way to record whether a person did not show up to accept the award in person? With a reason why? For the more prestigious ones, that might be of interest, e.g. Nobel Peace Prize laureates sitting in prison.
For "not present at ceremony", or anything else like it (eg "awarded posthumously"), I'd recommend a "significant event" qualifier on the award. I wouldn't worry about recording "speech given by someone else" but if you really want to, then you could use the same approach. Andrew Gray (talk) 09:41, 11 August 2018 (UTC)
Whether somebody is present at a ceremony or not doesn't sound particularly significant. If they are in prison, that can be recorded in another way. As for refused or returned awards, I'd say it would depend on whether the issuer of the award still considers it awarded to that person. The receiver doesn't have any control, beyond refusing to take a prize, even dead people can receive awards. Ghouston (talk) 11:06, 11 August 2018 (UTC)
I think there have also been instances where awards were withdrawn by the issuer. E.g., see the disclaimer at the end of [6]. In that case, I suppose the award would have a start and end date. Ghouston (talk) 11:10, 11 August 2018 (UTC)
I add many, many awards and typically it is from lists where details like these are not given. These lists ae revisited occaisionally and therefore additng qualifiers are fine. Removing awards is not practical without changing the sources. Thanks, GerardM (talk) 09:14, 18 August 2018 (UTC)
adopted child / adoptive parent : various practices, clarification needed
Hello all
I'm trying to figure the current state of affairs regarding adopted children and adoptive parents.
Those qualifiers are not used much (yet) but the use often does not respect this rule. The qualifier adopted child is sometimes found on child, and the qualifier adoptive parent on properties father or mother, which somehow makes sense, but may lead to a break of unicity constraint for P22 or P25, in the case where both biological and adoptive father or mother are present in the data. There are even cases where a parent is linked twice, once by father and once as relative/adopted parent.
This figures that the use of those qualifiers need clarification, and maybe the properties father, mother and child themselves. Their current definition don't explicitly restrict their use to biological parenthood, but they don't explicitly include adoption. This point should be clarified, my view being as follows:
If only one mother (or father) is asserted, the data are generally agnostic about the fact that she/he is a biological/legal/adoptive parent, and no qualifier should be used generally, although many legal parents are not biological parents, even outside the case of adoption. But the qualifier adopted child/father/mother should be allowed on the properties child/father/mother.
If more than one mother (or father) is documented, a qualifier should be used to make distinct the biological and adoptive parent. The biological parent should use the direct property (P22 or P25), the adopted parent being linked as relative with qualifier.
[edited] : The unicity constraint on P22 and P25 is in fact problematic in case of e.g., homosexual parents. This would need another thread?
Bvatant (talk) 21:03, 16 August 2018 (UTC)
I disagree with your second bullet; there's no good reason why a legal but non-biological parent should be shunted off to relative (P1038) just because a biological parent happens to exist in P22/P25 statements. That said, for better or worse, we do have stepparent (P3448) pointing to stepparent (Q19822352) for its definition, in EN: "non-biological parent or guardian". P3448 is gender-neutral, predicated on the observation that gender can be discovered from their item; spouse (P26) and child (P40) are also not gendered. P22 & P25 are gendered. And it is unclear where a gestational-parent would be fitted in - biological parent does not distinguish between gamete-parent and a surrogate mother.
I agree this is an area that should be improved, possibly in part by redesign or redefinition, and certainly by documentation, but I fear that wikidata torpor will inevitably run us into sand. To start: do we know if there is a documented model anywhere on wikidata from which we can work? Do we know of any ontologies outside of wikidata that might be useful? We surely cannot be the only group working with family relationships in RDF. Can we?--Tagishsimon (talk) 23:55, 16 August 2018 (UTC)
I'm not sure I agree myself completely with my second bullet :-) An alternative is to relax the unicity constraint on P22 and P25, but in case they have more than one value, should be made distinct by qualifiers. In this option, allow "type of kinship" on P22, P25 and P40, with allowed values in subclasses of kinship such as "child type" and "parent type" with values "biological", "legal", "adopted", "gamete-parent", "surrogate" ... and whatever might be invented by genetics wizzards.
Regarding stepparent (P3448) and stepparent (Q19822352), I don't find this definition "non-biological parent or guardian"? This is a more extensive definition than "spouse of parent" (which is the definition I read). Both are different from adoptive parent, who is often BTW another family member (uncle, aunt, grand-parent). Agreement with Jura's below comment.
Regarding other genealogy ontologies in RDF, unfortunately there is not much available. Searching e.g., https://lov.linkeddata.es/dataset/lov/terms?q=mother&type=property does not yield much. Either very general models such as Proton, or too convoluted ones such as CIDOC-CRM, using extra nodes such as Birth etc.
As for the torpor, well ... the bright side of it is that the use of qualifiers such as type of kinship are still rare in WD, so if a clarification is made now, they all could be curated manually quite quickly. Bvatant (talk) 10:18, 17 August 2018 (UTC)
I think P22/P25 is generally used for that, at least for countries were adoptions are considered equivalent. P3448 is for something different (spouse of parent). --- Jura05:14, 17 August 2018 (UTC)
WHERE {?parent p:P40 [ ps:P40 ?child ; pq:P1039 ?type ].}
ORDER BY ?type
Yields 213 results, with 17 distinct values of ?type. Some are clearly errors to be corrected (I will), such as "grandson" or "mother-in-law" (sic). Other are borderline such as "stepdaughter". "son", "daughter", "adopted child", "illegitimate child" are the most frequent and seem OK to me, although "son" and "daughter" can be derived from the child gender.
Replacing P40 by P25 or P22 yields less than 100 results, which I can look at and curate as necessary. Bvatant (talk) 10:51, 17 August 2018 (UTC)
Per [7], it's confirmed that this station is to be opened its service at 23 Sep, but after changing the date of official opening (P1619) value, it says that The value for $1 ($2) should be in the past, but not before $3., means that there probably have bug that treats that that date as "was in Gregorian 80"?! --Liuxinyu970226 (talk) 04:24, 17 August 2018 (UTC)
Liuxinyu970226, it would be *really* useful if you would stop using {{Q}} in subject headers on this page. It is not possible to navigate from the watchlist nor from the page history to the header, since the watchlist & history URLs are https://www.wikidata.org/wiki/Wikidata:Project_chat#Bug_in_{{Q|2130136}}? whilst the actual URL is https://www.wikidata.org/wiki/Wikidata:Project_chat#Bug_in_West_Kowloon_railway_station_(Q2130136)? (or, presumably, whatever the item label is in other languages). It makes it a right pain in the arse to find your threads. Better by far to put the {{Q}} in the body of the posting and, if you can think of no other wording, a plain Q2130136 in the header. --Tagishsimon (talk) 04:59, 17 August 2018 (UTC)
In answer to your question, if you look at Property:P1619#P2302 you'll see it is not a bug but a feature, which is noted on Property talk:P1619 - "Range from “+0080-00-00T00:00:00Z” to “now”: values should be in the range from “+0080-00-00T00:00:00Z” to “now”. Exceptions are possible as rare values may exist." Meanwhile date of official opening (P1619) is about the date something opened ... past tense. It is not, strictly, for making predictions about future events. --Tagishsimon (talk) 05:08, 17 August 2018 (UTC)
While it makes sense to forbid dates in the future to avoid data rot - how can be add the fact that a project under construction is set to be finished/opened at a given time? Would we need a new property to make sure that the item gets updated when it is really opened? Which leads to further question - at what step such items becomes an instance of "railway station". When its officially opened? While under construction? When project officially announced? And what about projects which were abandoned before opened, what to use as instance of (P31) for them? Or just using start time (P580) as qualifier set to "no value"? Ahoerstemeier (talk) 08:00, 17 August 2018 (UTC)
It seems the API prevents to remove more than 50 statements of the same property in an item within a single edit. Can this somehow be circumvented? Steak (talk) 21:41, 17 August 2018 (UTC)
At some (to me random) pages, like Q*s about asteroids, I get a symbol with 2 arrows left of the "Read" button asking me to merge with the chess player Q55229637. There seems to be an unsolved merge request. How can I get rid of that? --Gereon K. (talk) 07:30, 18 August 2018 (UTC)
It's also when you choose the "select for merging" option and then go to another item, the little icon appears to select the other item for the merge. Hovering over it, the tooltip says "process the postponed merge". --Kam Solusar (talk) 12:18, 19 August 2018 (UTC)
@Jeluang Terluang: I think this would require a software change; it looks like the URL formatter is hardcoded to do this because (a) most websites don't include spaces in their URLs, and (b) outside Wikidata items, you can't include a space in a URL because it breaks the wikitext formatting, so it's assumed by the software that you would encode any spaces in URLs as %20. Jc86035 (talk) 14:41, 19 August 2018 (UTC)
@Jc86035: Ah, but there's the problem. Instead of encoding it with %20, it instead encodes it as + which does not work. The current situation is any space in the field will be converted to + instead of the correct %20. Jeluang Terluang (talk) 15:39, 19 August 2018 (UTC)
@Jeluang Terluang: I think the best solution for now would be to have a bot convert all spaces in that property to %20 (@Pasleim: does DeltaBot do this already?). You should file a bug on Phabricator about the issue if one does not already exist. Jc86035 (talk) 17:35, 19 August 2018 (UTC)
For folks interested in Books and Citations
Since we still can't ping large projects, I'm dropping a note here that I have posted a refined proposal for modeling books (and the like) at Wikidata talk:WikiProject Books. The question of a new model for books has been an open issue for years, and there are two different proposals on the table (the other by Snipre), both of which differ from best practices as published today. Comments from interested Wikidatans are welcome. @Snipre: FYI. - PKM (talk) 19:51, 18 August 2018 (UTC)
Areas of Cyprus Republic (de jure)/Areas of Northern Cyprus (de facto)
Hello. First of all I want to say that I am a Greek Cypriots (Q245794). But I believe that I have Neutral point of view. Sorry about using words like "occupied", "free" etc. It's easier to explain that way.
There is no problem having items for the villages and municipalities of Northern Cyprus (Q23681). We can do the same as any other village or municipality in the planet. They claim that they are a country (even though not recognized), they have their own administrative territorial entity.
As you can see, these districts do not identify with the "occupied" districts of Cyprus, perhaps with the sole exception of the province of Kyrenia (Girne), without being absolutely sure ... We have a separate item for each district of Cyprus (Q229) and for every district of Northern Cyprus (Q23681).
The main problem is about municipalities and communities (villages). Cyprus (Q229) elects mayors, municipal councils, community councils for all "occupied" municipalities, has Geographical codes of the Republic of Cyprus (Q55963047) for all of these areas, etc. With always the footnote that concern areas belonging to Cyprus (Q229) but not are controlled by Cyprus (Q229) because of the presence of the Turkish army. Of course, Northern Cyprus (Q23681) also elects mayors, has their own codes etc.
Problems:
Kyrenia (Q206760): Who is the mayor? The Greek Cypriot or the Turkish Cypriot? In English Wikipedia article of w:en:Famagusta has both mayor with de jure and de facto in brackets.
That what Cyprus (Q229) define as Keryneia Municipality, the area that the mayor has the has the administrative responsibility, is not identical for Cyprus (Q229) and Northern Cyprus (Q23681). That is, the Greek Cypriot mayor is the mayor of a particular territory, but the Turkish Cypriot mayor is the mayor of a larger or smaller territory. We both call them Kyrenia/Girne! But they are not the same.
coat of arms image (P94): Keryneia municipality has different coat of arms that Keryneia/Girne Municipality.
Websites: Famagusta Municipality for Greek Cypriot [8] and Famagusta Municipalities from Turkish Cypriots [9]. (English Wikipedia article gives both w:en:Famagusta).
I am not sure about any solution. The only I have though:
Kyrenia (Q206760): item only about the city. Not informations about municipalities. That way there is no problem, because city history continues regardless of who are living to the city and who are managing. The city has no statutory (legal basis) limits, area, emblem, website, quarters. The municipalities have. The municipalities do not identify with the city, as Limassol (Q185632) is no longer identified with Limassol Municipality (Q28870916). The only problems here are that it is necessary to clarify in which district the city is located (different for Cyprus (Q229) and Northern Cyprus (Q23681)). Perhaps there are others problems that I did not think.
Keryneia municipality: item about Keryneia municipality from its foundation at 1878 as today. It will be the item for the municipality that continues to exist as an entity but with the administration and the residents outside the municipality (considering that for Cyprus (Q229) refugees from Kyrenia and their descendants have electoral rights in that municipality. They are considered voters of the municipality and citizens of the municipality in general, according to Cyprus (Q229)).
Girne Municipality: an item for the municipality that undertook the administration of the area in 1974 and which in 1983 became the municipality of the newly established Northern Cyprus (Q23681)). It will essentially be the item for the municipality with a year of foundation in 1974.
However, that solution did not solve the problem of the villages. Are we going to have 2 items for each village? It that useful? Again, however, it is not certain that each village is terribly identified as Cyprus (Q229) and Northern Cyprus (Q23681) mean it, even though they have the same name (translated between to Greek and Turkish). I cannot find a solution.
The problem you describe is not unique. The solution is in essence simple. When you talk about "Territorial and administrative entities", they exist in relation to each other. Actual human settlements are part of either structure. Officials have their office in the entities. Thanks, GerardM (talk) 08:57, 18 August 2018 (UTC)
The solution is that there are two distinct and separate structures in place. A Greek and a Turkish structure with only overlap where the structures indicate physical objects like human settlements are in both. Thanks, GerardM (talk)
Not for the settlements but for all the "administrative and territorial units", they are "per country". The lowest level links to the human settlements Thanks, GerardM (talk) 04:41, 19 August 2018 (UTC)
For some of the labels to make sense, wouldn't you need to add "genre" instead, if you drop "film"? --- Jura05:58, 19 August 2018 (UTC)
I disagree. For example, I'd venture that no one ever referred to a heist film as a "heist". A "heist" is an actual crime, not a film about one. - Jmabel (talk) 07:03, 19 August 2018 (UTC)
It seems like it really depends, Like you would never call a Western film (Q172980) a "western film" you would just call it a "western". But to be fair, the diseimbiguation is implied (since that's the only usage of western as a noun I can think of). If you were instead listing a list of genres you might say: "science fiction, western, crime, comedy". On the other hand, their are some notable exceptions like silent film (Q226730) and 3D film (Q229390) which should always have the "film" suffix, but in my mind these are more film types than genres. I could go either way on heist film (Q496523) since it would fit either way in a list of genres, but then again, it really depends if you see a genre as a noun or as an adjective. I think as far as Wikidata is concenred, it doesn't matter because of Help:Label#Disambiguation_information_belongs_in_the_description, but like I said, there are probably some exceptions to that rule. U+1F360 (talk) 13:29, 19 August 2018 (UTC)
The more I think about this, the more I think this is a problem with English. Some of our genres are nouns (“western”, “comedy”, “science fiction”, “thriller”, “action”, "horror") and some are adjectives (“crime”, “heist”, "teen"). Indeed, on Wikipedia you see these genres get the "film" suffix (in the article body) when they are used as an adjective, but the suffix is dropped when being used as a noun (i.e. "teen film" is a type of "comedy"). For the pruposes of Wikidata, I think the labels should be consistant, even if they don't make a lot of sense in noun form. It also seems that in general, the more traditional genres (that perhaps existed before film did) do not get the "film" suffix, but newer genres (or subgenres) do. However, I am excluding silent film (Q226730) and things like ti because I'm not convinced it's a film genre (Q201658) as it has to do with the technology used on the film (or lack thereof), not the film itself. U+1F360 (talk) 13:49, 19 August 2018 (UTC)
The thing isn't so much that some are adjectives and some are nouns as that some are used as' an adjectives and some as nouns. English, much more than other Western European languages, allows nouns to be used as adjectives; for example, there is a community near where I live called "Lake Forest Park". Or, also, "western" is certainly originally an adjective but has become a noun for a film genre (also for a genre of novels, by the way, though it's not as popular as the film genre). "Science fiction" is already an adjectivized noun ("science") plus a noun ("fiction") and all on its own, without "film", is much more likely to refer to novels or short stories than films.
"Western" usually means a film, so I could see shortening that, but absolutely would not do that with "heist" and would lean against with "science fiction" if we are specifically intersecting it with film. No one says, "I'm going to see a science fiction" the way they'd say "I'm going to see a western". As a genre, it spans literature, film, artwork to some extent, etc. If we are specifically talking about science fiction film, I think pretty much everyone would include the word "film" or "movie". - Jmabel (talk) 16:10, 19 August 2018 (UTC)
Or perhaps a better question... how would one query for a list of film genres, and get... a properly named list? U+1F360 (talk) 16:44, 19 August 2018 (UTC)
@Jura1: it's just an example, but how would you propose this done in a multilingual fashon? Of course in english I could just strip out "film" but that's not the same for every language (or even the same structure). And I'm still wondering how "film" is not disambiguation information. Could you please explain how it isn't? U+1F360 (talk) 18:24, 19 August 2018 (UTC)
┌────────────────────────────────────────────────────────────────────────────────────────────────────┘ I suspect it's infinitely debatable whether "film" is a disambiguation term in "comedy film", but the same could be said for "hat" in straw hat (Q204390) or "work" in reference work (Q13136). If we're going to make rules based on grammar, then the rule is probably to allow "loose compounds" (very common in English) and not allow phrases that aren't accepted as loose compounds, but then we have to pick which dictionaries or usage manuals we choose to follow. Being a pragmatic sort, I think we should include "comedy film" and "comedy" as label and alias, and the preferred label usage should be agreed to as a standard by the appropriate Wikidata project where there is one. - PKM (talk) 19:33, 19 August 2018 (UTC)
@PKM: Perhaps a solution could be to split these items into one that is a subclass of (P279)film (Q11424) and another that is a instance of (P31)film genre (Q201658)? I think perhaps the confusion is over if this thing is a type of film or if it's a genre. The former would have the suffix, the latter would not. This would be consistant with your examples as well (i.e. "Staw hat" is a type of "hat" and "reference work" is a type of work). Would this be an acceptable compromise? U+1F360 (talk) 20:00, 19 August 2018 (UTC)
wikt:heist#Noun has "heist" as slang for "heist film". Accordingly, it would be a reasonable alias for the concept, but shouldn't be the label. If you need shorter names, you could try adding "short name" statements to the items. --- Jura20:15, 19 August 2018 (UTC)
Well I didn't know about short name (P1813) so perhaps that's the best place for the shortened version. Are we good with adding those to genres in order to remove the "film" suffix? If so, that works for me (I realize that "science fiction" would become "scifi" but I think that's fine). U+1F360 (talk) 20:25, 19 August 2018 (UTC)
@U+1F360: This looks like an attempt to find a problem where none really exists. There's enough real work to be done around here, without fixating on stuff like this, where there is no problem, and no action either required or desirable. I strongly oppose all of your proposals so far, as either removing information, disrupting valid relationships, or misusing P1813. Nothing to see here, move along. Jheald (talk) 11:30, 20 August 2018 (UTC)
@Jheald: If this isn't something that's important to you, then that's fine, but being dismissive of what's important to me is rude and hurtful. Regardless, the examples on short name (P1813) give adding "Adair" to "Adair County" (it was even the justification for the original proposal), so based on the examples, it's a perfect use of short name (P1813) and that sounds like the way forward. Thanks everyone for your help! U+1F360 (talk) 12:55, 20 August 2018 (UTC)
OH! I'm sorry, I totally read that wrong. Yeah that's fine, as long as the shortened version exists somewhere I think that's an acceptable compromise. I just thought of another idea, which is that I could just take the alias that is the shortest (which, should probably be the same as the short name anyways?). Which would you all prefer, having a short name of "hiest" or having "hesit" as an alias? (I don't think you could make an alias shorter than "scifi" so I think that works too). I'm totally cool with either solution. U+1F360 (talk) 13:32, 20 August 2018 (UTC)
I personally prefer alias, which is more flexible and searchable than "short name" (P1813). It's totally acceptable to add phrases without "film" as aliases to film-genre items.--Neo-Jay (talk) 13:49, 20 August 2018 (UTC)
Quelq'un qui n'a pas de "bagages" informatique peut-il aisément espérant apprendre plus?
– The preceding unsigned comment was added byYobouekouadiojeanluc (talk • contribs).
I am looking at publication date (P577) of She and Her Cat: Everything Flows (Q54957839) and the value being displayed in the box to me is "4 3 2016", without any delimitor other than just spaces, nor any indication on month/date order. Is it possible to let users change it into a format they are more familiar with, or can it be localized to a date format that user would understand? C933103 (talk) 10:31, 20 August 2018 (UTC)
Bovlb I see two options available there. One of them is "default", the other one is "2018-08-20T17:49:47". How can I, a.) change the default for all people using this locale in Mediawiki software, and b.) add another options there? c.) Also, I have selected the "2018-08-20T17:49:47" option but the value in the entity I mentioned above is still displaying as " 4 3 2016". What's missing? C933103 (talk) 17:54, 20 August 2018 (UTC)
@Bovlb, Jc3s5h, C933103: This is a bug, phab:T63958, not something that can be fixed in user preferences. (The bug is from February 2014 and is not fixed.) For Chinese there is only one option in preferences, which is not the one being displayed. Jc86035 (talk) 17:56, 20 August 2018 (UTC)
Wikidata weekly summary #326
Here's your quick overview of what has been happening around Wikidata over the last week.
This is the Wikidata summary of the week before 2018-08-20.
JOSM/Plugins/Wikipedia/GSoC 2018: 'Floscher' has been working to improve the Wikidata/Wikipedia-related functionality of the 'JOSM' editor for OpenStreetMap
Design Challenges in Named Entity Transliteration - a paper discussing "fundamental design challenges that impact the development of a multilingual state-of-the-art named entity transliteration system", with reference to Wikidata
Quicksilver: Training an #ML system to generate draft Wikipedia articles and Wikidata entries simultaneously, slides by John Bohannon and Vedant Dharnidharka
How do our other ontologists feel? Should "history of Europe" be a <subclass of> "history of the world", or <part of> "history of the world"? There are enough items built both ways that harmonizing them will be a big effort assuming anyone thinks we should do so. I am leaning toward <part of> myself, but I could change my mind. - PKM (talk) 01:43, 15 August 2018 (UTC)
I lean the other way, because wdt:P31/wdt:P279* tends to be the first tool I reach for when wanting to report across items having different P31s (for me, recently, mostly films and airports); and also because there's no expectation that a reciprocal has part(s) (P527) is required. That said, films & airports are not part of a divisible whole in the way that Hof Europe is assuredly part of the HofWorld. That being the case, has part(s) (P527) aside, I don't have a great problem in having a redundant part of (P361). --Tagishsimon (talk) 02:18, 15 August 2018 (UTC)
"Part of" seems more natural to me, there's only one "history of the world" or "history of humanity", and I'm not sure what a subclass (or instance) would be. The history of Europe is a part of the history of the world, but it's not a history of the world in its own right. Ghouston (talk) 02:53, 15 August 2018 (UTC)
But you can say that about any subclass: a part of the definition of any subclass is that it does not contain all of the class. Aviation example: 747-200 is a sublass of 747, but does not contain any of the other 747 place types. Or from the subclass of (P279) definition: "this item is a class (subset) of that item." The History of Europe is a subset of the History of the world. --Tagishsimon (talk) 03:05, 15 August 2018 (UTC)
Not really, since a particular plane of type 747-200 is also a 747: the instances of 747-200 are a subset of the instances of 747. The history of the world doesn't have instances like that, the history of Europe isn't the history of the world. An alternative would be to make it an instance of something like "history of a region", like history of Africa (Q149813) is an instance of regional history (Q1802210). Ghouston (talk) 03:13, 15 August 2018 (UTC)
How do I indicate that a school served only black or white students during segregation?
I am trying to create structured data for a number of segregated schools in the American south in the first half of the twentieth century, when Black and White students were separated. What property do I use? The same issue arises with schools attended by only male or female students. Leofstan (talk) 20:18, 20 August 2018 (UTC)
@Leofstan: It's an interesting question. I've had a quick poke around and I can't find any single-sex schools/colleges which have relevant Wikidata properties, either. instance of (P31) doesn't seem like a good approach given that many of these will have been changed between segregated and mixed over time. One approach might be to use legal form (P1454) with a value like "segregated school", but that seems a bit hacky. Either of these would work for a quick-and-dirty solution we could improve later, though.
Alternatively, we might need to create something - preferably not just for schools, as there are also things like racially segregated military units (eg 92nd Infantry Division (Q4645846)), and all sorts of other organisations with some kind of restriction on who can be associated with it (eg freemasonry (Q41726)). Can't think what to call it, though. Andrew Gray (talk) 22:24, 20 August 2018 (UTC)
Related issues: How to structure items for schools that are for particular religious groups, or particular age groups, or for speakers of a particular language? (These three might be somewhat different as all usually come with differences in content or method of instruction as well.) --Yair rand (talk) 22:39, 20 August 2018 (UTC)
Fetch only specific "instance of" content from Wikidata
Hello. In the context of improving a wikidata-supported infobox on enwiki, is it possible to fetch the "instance of" parameter from Wikidata, but to only display it if the field contains specific content... For example, lets say I want to add a wikidata infobox for Nesjavellir Power Station (Q693330) on Wikipedia. In that infobox, I want to show that Nesjavellir is one of three types of geothermal power stations (i.e. Dry Steam, Flash Steam, or Binary Cycle). But as you can see in Nesjavellir's wikidata item, further types of "instance of" could be loaded for various reasons. How do I filter those out? Thanks in advance! Rehman04:59, 9 August 2018 (UTC)
Use lua code to filter the data extracted from WD, or create a dedicated property in WD with constraints to be sure that only a limited set of value is used for one characteristic of the item. Instance of will be the less reliable value in WD as there is no unique way to classify an item. Snipre (talk) 09:03, 9 August 2018 (UTC)
Thank you, Snipre. If you are willing, may I ask for your help to do that at this sandbox please? Or maybe show an existing example? I need to update | data13 so that it:
Filters from Wikidata to "only show one of the three geothermal power plant types"
Use the local value, if a local value is added
If I can see an example of how that's done (particularly #1), I can work out how to implement the same in other areas as well... Rehman10:49, 9 August 2018 (UTC)
A non-Lua way to do it is {{#ifeq:<search string>|{{replace|<search string>|<search text>|}}||<display text>}} - where <search string> is the info from P31, <search text> is the text you are looking for, and <display text> is what you show in the infobox if it's found. Thanks. Mike Peel (talk) 14:45, 9 August 2018 (UTC)
True-ish. But redundancy makes report writing more easy. I wonder if symmetrical warnings such as this could be extended, maybe by gadget, to provide a 'do it' button, which, if clicked, does it? --Tagishsimon (talk) 15:53, 20 August 2018 (UTC)
I came across a <no value> in a date of death (P570), which surprised me; and talked to a user who suggests that it is a reasonable way of encoding "is still alive" against a person item. As far as I can see, 69 items at the time of writing this have such a value. What do we make of this? Good, bad, indifferent? Thoughts please; thx.
A good reason to have the statement is data completeness and that in this way we have recorded the fact, that said person is indeed currently living, because they haven't died at the time when the statement was made. Not to have any statement can lead to the confusion whether the person is currently living or the date of their death just has not been recorded in Wikidata yet. That means less confidence in the actual status from the data at hand. On the other hand, reasons against the statement could be, that some living people, who have Wikidata items could find it creepy, to even have a statement related to their death at all. --MB-one (talk) 11:18, 21 August 2018 (UTC)
How about people who have been mistakenly reported as dead? We have the wrong date depreciated. Should we have a normal statement with <novalue> or only have the depreciated statement? – Máté (talk) 11:25, 21 August 2018 (UTC)
I think it would be misleading to use that property for living people, since it implicitly says that while we don't know when they died, they did. – Máté (talk) 12:36, 21 August 2018 (UTC)
I guess my question is what have I done wrong, can I fix it, and why would these publicly funded entities and products be considered spam/non-notable given the number of users from within their field?
@WEP11: All items look absolutely fine to me - notable, now interlinked; no problem I can see. User:Esteban16, do you still have a problem with them? They're clearly not spam, so it's a bit odd that you marked them as such. (WEP11 - I refactored your post to get better links to the items in question.) --Tagishsimon (talk) 13:57, 21 August 2018 (UTC)
@Tagishsimon: I noticed that having a Wikipedia page is important, but that won't appear until it makes it through Wikipedia's review process.
@WEP11: That helps by satisfying criteria 1 on Wikidata:Notability. Your items clearly satisfy criteria 2 (references always help) and criteria 3, since they will be linked from academic & other papers, people items, etc. But, you know, deletionists versus keepists (?)--Tagishsimon (talk) 14:10, 21 August 2018 (UTC)
user:WEP11, sorry about the anti-organization bias. best to find a friendly neighborhood wikimedian to move your article, and avoid Conflict of interest claims. as you see, some editors can be dismissive of what they do not understand. we would be happy to help with an editathon, if you had content you wanted written about. Slowking4 (talk) 20:08, 21 August 2018 (UTC)
@Slowking4: I started a Wikipedia draft with my conflict of interest stated in the talk page. My hope was that I provided the bare minimum of information that would get someone's attention at least to contribute more. I also added a section to the National Centers for Environmental Information page that provides information about the governing program.. But I guess this is getting too far off topic from WikiData? Sorry, I'm new to all this, but I don't want to violate any of the conflict of interest requirements either. WEP11 (talk) 18:14, 22 August 2018 (UTC)
That is correct. We're using "subject" and "object" to refer to two of the components in a statement: <subject> <property> <object>. --Yair rand (talk) 04:19, 22 August 2018 (UTC)
Is it true that disambiguation pages and proper article pages should not be linked to the same wikidata item? If that is true, Annam (Q564450) should be split into two. For example, entries on German, Japanese and Russian WP's are articles, but English and Chinese ones are disambiguation pages.--Roy17 (talk) 14:29, 18 August 2018 (UTC)
I'm afraid it has to be split. Annam (Q564450) is a former name of Vietnam, which at some point in history was a protectorate French Annam (Q10828323), so some WP's write it as an article explaining the name itself, some write it as the protectorate, and some use it as disambiguation, pointing to the protectorate and other things with the same name Annam. The tricky thing is this problem involves many WP's. I don't know if there is a quick way to split an item (like an inverse function of merging items? :D), otherwise manually assigning them one by one is a little cumbersome.--Roy17 (talk) 16:46, 18 August 2018 (UTC)
@Roy17, Jmabel: Mostly done, except eowiki, frwiki, ruwiki and viwiki, all other non-disambiguation pages are moved from Q564450 to Q10828323, and merged Q16766542 (another disambiguation pages item that has jawiki, kowiki and zh-yuewiki links) to Q564450, now I need suggestions on how to handle the rest four. --Liuxinyu970226 (talk) 23:27, 18 August 2018 (UTC)
pinging all non-bot contributors of rest four to help us on handling:
@Liuxinyu970226: Thanks a lot! I'd say those remaining pages are examples of articles explaining the name Annam itself, which should be split into a new item. Annam, a historical name of Vietnam....--Roy17 (talk) 23:47, 18 August 2018 (UTC)
I am reading the manual for the template https://en.wikipedia.org/wiki/Template:Coord on English Wikipedia, in which it mentioned that, the source: parameter can be used to specify the data source format, with one of the possible format being Ordnance Survey National Grid (Q1967305). As coordinate location (P625) specify that only WGS84 coordinate should be included in the property, if someone accidentally included Ordnance Survey National Grid (Q1967305) coordinates as value for this property, it might result in errors in the resultant product (possibly up to 120m difference according to its article on wikipedia). And then as bots/users would copy value from Template:coord into wikidata for coordinate location (P625), it seems like it is possible for coordinates to be mistakenly copied over. Is there anyway to find out if there are any/how many of such cases exists in wikidata, and then locate/fix them?C933103 (talk) 23:56, 21 August 2018 (UTC)
So tree of elections would be extended down to what Americans usually call a race, that is, the part of an election that fills a single office, or a group of identical offices. Then the properties mentioned in Template:Election properties are added to the race. Have I got that right? Jc3s5h (talk) 19:29, 22 August 2018 (UTC)
I don't know if that would work. It seems to be trying to say "one child as of this date, two as of this one" - but we wouldn't normally say she stopped having had one child when they died. It's a bit odd. Andrew Gray (talk) 21:08, 22 August 2018 (UTC)
I am starting to understand the meaning of the number now! Until now I believe that it meant “first child”, “second child”, and not “one child”, “two childs”... is that the meaning? Anyway the child born in 1939 is death in 1979 and in somehow must be added the death date... --151.95.10.8521:24, 22 August 2018 (UTC)
Bot account limited
I was just using my Pigsonthewing-bot account to manually fix up a few P31s, like this, and was blocked with an error message: "As an anti-abuse measure, you are limited from performing this action too many times in a short space of time, and you have exceeded this limit. Please try again in a few minutes."
I have a request on behalf of WikiProject Newspapers of English Wikipedia, which is an initiative to build a number of short articles by December 15, 2018. We are in need of a convenient way to measure our progress and, on that date, our overall level of success.
The most essential thing is to be able to measure:
A1 and A2 are the number of Wikidata items as of that date:
that are instance of (P31) newspaper (Q11032)
Where country (P17) is USA (Q30).
B1, B2, C1, and C2 are the respective numbers of Wikipedia articles associated with the Wikidata items on the first line.
Nice-to-have
The chart above, with all values filled in by a bot or query or other technical means, would be tremendously helpful in itself. Several other features would make it even more helpful:
Each cell links to a list of all items/articles, similar to how the WikiProject assessment charts on English Wikipedia link to pages on wmflabs
In addition to the three rows above, we also measure:
Number of articles that do, or do not, contain warning banners (e.g., notability)
Article is assessed higher than start class (or, separate rows for stubs, start, C, B, GA, and FA class)
Ability to generate tables like this for each state
The tools built to meet this need can be readily adapted to assist other, similar content-building initiatives in the Wikimedia space in the future
I don't know if this is the best venue for this request; please advise if there is a better place to inquire.
Okay let me start to provide some input here. First of all, it is difficult to figure out numbers for the 1st of July, as historical data is very difficult to evaluate. I thus give numbers as of now:
I have made a small modification in the query, so that it also considers instances of subclasses ofnewspaper (Q11032). Numbers are not very different, as most items use the newspaper item directly.
The query links always provide up-to-date numbers and lists, thus the numbers will change as soon as your project makes progress. Since I’m not very familiar with enwiki’s article assessment system, I’d like to hand over those queries to someone else. It should not be so difficult to evaluate most of your requests with petscan in a similar way like in the last row of the table above. —MisterSynergy (talk) 09:25, 23 August 2018 (UTC)
@MisterSynergy: Thank you for this! Very helpful. I especially appreciate how you provided queries both for the raw count, and for the links to the actual articles. Also, thank you for fixing my request, I had not considered the "subclass" issue -- good catch!
I realize that going "back in time" is a more challenging request, but it's important, so I hope there's a way to do it. It's not just laziness, i.e. not completing a baseline; it's also important because the quality of the Wikidata items is dramatically improving during the course of the project. Thanks especially to the efforts of @99of9:, we've been doing a great deal of data cleanup...matching, merging, adding missing statements, removing incorrect matches, and so on. So:
Number of articles matching the query on July 1, 2018 ≠ Number of articles matching the query on December 15, 2018 that existed on July 18, 2018.
I am thinking that perhaps a bot or wmflabs tool could handle this part? That is, take the results at the end of the project, crawl the articles, and record whether or not the article existed on July 1, and if so, whether or not it contained the infobox on that date. Does that seem possible? Prohibitively difficult? -Pete F (talk) 17:10, 23 August 2018 (UTC)
Phew. From my experience “prohibitively difficult” is a suitable description for this task, but maybe someone else can come up with an idea. My first guess would be to download a database dump from around the 1st of July, if available, and evaluate it offline, which probably requires some powerful hardware and lots of installed software (I never tried this by myself). We can follow the history of individual items quite nicely and tell for any time in the past which state it had back then, but over entire Wikidata this is pretty difficult to achieve as Wikidata (and Wikipedia) are not really designed to query historical content development data (both Petscan and the Wikidata Query Service cannot access page histories). A similar task in Wikipedias would be to evaluate (the number of) category members of a given category at some time a year ago. There might have been articles removed from the category, which is extremely difficult to track, and additions would have to be looked up article by article as well. So if ever possible, I suggest to arrange a way to deal with this incomplete data … (maybe data extrapolation helps to close the gap later). —MisterSynergy (talk) 17:27, 23 August 2018 (UTC)
What if we think about providing the most automated assistance possible for a human effort? For instance, I'd imagine it would be pretty easy (like, a regex search-and-replace) to take the lists you generated above, and convert the URLs to display the view history screen with the oldest revisions of each article. Maybe a fairly simple text processing algorithm could then extract the date of the oldest revision, and determine whether it was before or after July 1, 2018...or, maybe it could use the diff ID in order to make that determination?
I could imagine in some cases, this won't be very good information because the original creation was a redirect...but, I think that's something we could correct for with manual inspection...
I couldn't find an link to an help portal, so I will ask it here. Please feel free to move it to a better place if needed. I just created: David Sambissa (Q56192990). Could you please check if its done correctly as its my first.
Is it possible to append something to the URL of an item, such that, when viewed, the "All entered languages" box is uncollapsed? Or to display all the labels in some other way? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits21:33, 23 August 2018 (UTC)
Thank you, but I was thinking of sending URLs to third parties, when they ask a question like "what are all names, in different languages, for a sunflower?". Apologies that my question was ambiguous. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits15:21, 24 August 2018 (UTC)
The easiest thing to do would probably be to tell them to click the "All entered languages" link on the page. Other than that, all I can think of is to create a SPARQL query such as this one . - Nikki (talk) 17:24, 24 August 2018 (UTC)
Distinct value constraint issue
Consider the situation where I am having two wikipedia articles 1970 Stockholm Open and 1989 Stockholm Open. Since both these articles contain same information in different years, the official website has not changed. In such a case, when I add the same official website to two wikidata items, it shows distinct value constraint issue. How can it be solved? Adithyak1997 (talk) 06:10, 24 August 2018 (UTC)
The previous discussion on this subject got bogged down in discussions of user interface and search recall and ended without any definite conclusion. I thought I would have another go at resolving it, focussing on two specific points here.
The Union List of Artist Names (ULAN) is a useful source, but the "additional names" have a number of problems:
They are not tagged with any language;
Most are not common names for the entity in any language;
Many of them are common misspellings or variations in word order;
Many of them actually identify related entities instead; and
Some of them are marked as being of unknown provenance.
The bot operator has indicated that the bot will re-add these aliases if other users remove them and that that this is intended behaviour. It's not clear to me that we want bots to always override user decisions or to get into edit wars.
I think you might want to ping users who work on or with artist items for this discussion. Project chat is not always followed by everybody. That said, I think it is important to note that users are free to add aliases and the bot will not remove any aliases that are added. It just updates the item with missing aliases from ULAN. So it is not true that user decisions are always overidden by the bot. Next the reasons for the bot are clear, namely to improve search for artists and reduce the creation of duplicate items for artists. It is still not clear to me however what your specific objection is besides wanting to ban certain information coming from the ULAN. I guess I don't see the point of your argument. Jane023 (talk) 10:01, 15 August 2018 (UTC)
I take your point that the bot override is one-way. It allows addition of aliases, but not removals. Nevertheless, it could still lead to edit wars with users trying to follow policy.
My objection here is we have a fairly clear policy on aliases that is not being followed for ULAN import. I agree that making the Wikidata search engine useful is important, not least to avoid duplicates, but it is not the only (or even an explicit) goal of aliases. Importing ULAN additional names into English aliases results in many alias claims that are false according to policy. This is not a purely theoretical concern as this reduces the practical utility of aliases for many purposes, such as building a lexicon to match Wikidata items in text.
So, a specific proposal:
Create a new property for ULAN additional names that allows qualification to reflect the attributes defined in ULAN.
Switch the bot to populating that property instead of aliases.
Augment search to drawn on that property in addition.
Do you have a specific proposal for changing it? If aliases are just to support search then we could throw in anything that might increase recall and let the search ranking deal with precision, but I think I've explained above how I think aliases have other uses beyond search where the current policy is more helpful. Cheers, Bovlb (talk) 17:21, 16 August 2018 (UTC)
I think it would help if the help page would describe current practice. Writing it down would be sufficient. The main point that worries me from what you mention is "Many of them actually identify related entities instead". Do we have samples from Multichill's bot additions for that? Oddly, your proposal doesn't address this. --- Jura07:07, 17 August 2018 (UTC)
Most aliasses imported from ULAN are okay but there are some outliers which can exasperate people [10]. I already suggested few months ago, it would be pleasant if the bot could keep track of its own edit and therby not undo edits of humans. Adding a small fraction of bad data once is much less problematic than adding it over and over again. --Pasleim (talk) 12:51, 17 August 2018 (UTC)
If BotMultichill[T] is going to add ULAN aliases that are removed from the list of English aliases back to items, it would do well to not just check that they were removed from the English alias list, but also check that those aliases have been moved to the list of aliases for another language—before deciding to add them back to the English alias list—so that those aliases which better fit in the German/French/Italian/etc. alias lists remain in those alias lists only. Mahir256 (talk) 14:27, 17 August 2018 (UTC)
Having as many labels for an entity is very useful in NLP tasks such as Named Entity Recognition.
I'd like to comment on your claims one by one:
"They are not tagged with any language"
"Most are not common names for the entity in any language": Just because a name doesn't look "legit" to you, doesn't make it useless in art research.
"Many of them are common misspellings or variations in word order": "misspellings" are useful. Word order variations (name inversions) could be omitted, hoping that a NLP software could compensate for that. But I don't know how to disinguish them. Eg "Jean Breughel, dit, De Velour" ("Jean Breughel called Velvet") doesn't have a name inversion.
"Many of them actually identify related entities instead": please give examples and try to quantify this claim
"Some of them are marked as being of unknown provenance". Getty is quite scrupulous in recording provenance: all names must have occurred in some source to be recorded (ULAN has 61k sources). Let's see at http://vocab.getty.edu/sparql: 857929 ULAN names have a source, and only 140 don't have a source (change the query to "filter not exits):
I'm not saying that uncommon or misspelt labels are useless for, say, art research; I'm saying they don't meet our (current) definition of aliases, and there are better ways we could represent them.
To inform possibly interested parties and consider thoughts already brought up on the problem a chronological list of previous related discussions I recalled and collected (also because User talk:Multichill unfortunately gets archived quit quickly).
I want to be able to add a list of politically relevant publications (articles, books, blog posts) to entities that are instances of business. How can I achieve this?
Date of death 'unknown value' for people who can be assumed dead?
I came across the statement that there are currently 72 living Medal of Honor recipients in Wikipedia. I wanted to check if a WD query comes close to that number:
It matches more than 250 people, but a lot of them have dates of birth in the 1800s. Is it alright to define a date of death for all of those with the value set to "unknown value"? Obviously they are dead, but maybe the date of death is known to somebody. Is there a community consensus on when to use "unknown value"? If it is alright, could someone run a bot to add dod=unknown to all people with a date of birth at least 130 years in the past? Then again, the date of birth may be wrong, and adding that dod automatically would just decrease the quality of data further. Interested in opinions or pointers to WD rules or past discussions. Thanks!--37.201.29.13621:40, 25 August 2018 (UTC)
We also know they have a place of birth, a place of death, a mother, a father, one or several places of residences .. not sure what it helps if we set all these to unknown (unless we insert a reference that states it's unkoown). Obviously, it would be helpful to insert actual dates, especially when Wikipedia might have them. --- Jura21:45, 25 August 2018 (UTC)
I see your point about avoiding clutter, but the attributes you mention are certain to exist, a date of death is not. Deriving the fact that somebody is dead from the date of birth is shaky but would keep people from having to add the kind of guessing logic I mentioned (dead if dob at least 130 yrs in the past) to all their queries to make them more accurate.--37.201.29.13622:03, 25 August 2018 (UTC)
Well, it would cloud up cases where we don't actually know and I'm afraid it might deter people from adding actual dates. I don't think it's problematic to do it in the query only. --- Jura08:35, 26 August 2018 (UTC)
It's inherently "clouded up". An unknown or imprecise value could become a known or more precise value if new sources are found, even if it was the best information we had at the time we entered it. - Nikki (talk) 09:47, 26 August 2018 (UTC)
In this case, no attempt is made to seek such sources. It's just that no import has been done yet. --- Jura09:52, 26 August 2018 (UTC)
That's how it's always worked. There's no feasible way to track who is still alive, so all we can do is track who is dead and assume that anyone who isn't dead is still alive. - Nikki (talk) 10:09, 26 August 2018 (UTC)
Well, not really for people born how would be older than 100 or 110 years. Anyways, obviously, if it's thought that merely adding "unknown" is preferred .. would make it easier for many things. --- Jura10:13, 26 August 2018 (UTC)
Date of death set to "unknown" where we are absolutely confident the person is dead (ie, "would be older than the oldest known person", or "is known to have died per source X") is pretty common and I don't see any problem with it. However, I agree with Jura that we should avoid adding it if it's possible to do a quick import of dates first to get a better value (or even a vague 'circa' one) - it's best used for "known unknowns". Andrew Gray (talk) 12:06, 26 August 2018 (UTC)
Quick addition - I've had a look for those who aren't noted as date-of-death unknown on WP, and noticed that one of these is Frederick Hayes (Q24962680), who seems to be one of the "revoked" 1860s awards. I'm not sure how you want to handle these but thought I'd better flag it up ... Andrew Gray (talk) 12:34, 26 August 2018 (UTC)
Heat map for IDs of geographical items
Hi. Do we have a tool that displays in a certain range on a map all the items with coordinates and gives them a different color based on the number of IDs, possibly the usual "green to red" scale? Or should we use a query for that?
Since IDs are a good clue of quality and depth of long-term maintenance, it would be useful to spot the worst areas in order to boost specifically their situation for an edit-a-thon or WLM or another wikimedian event, so the overall situation is kept balanced.
we could do the same also with properties but they seem to have bigger inter-item variance. There are instead many geographical IDs that are valuable both for churches, palaces or parks so the background is much less variable in this case, IMHO. Properties can also be imported from wikipedia but the number and type of IDs seems less affected if such items have an article or not. There are also many single-ID massive import that I would like to spot better on the map as recently occured with (too) many items of school and museums in Italy.
In any case, I wanted newbies to work specifically on IDs to boost some literacy under this aspect, that's why I also would like such heat map to be based on the number of IDs and not properties. I would do this before a campaign for better sourced statements.--Alexmar983 (talk) 21:42, 25 August 2018 (UTC)
Such a map would be misleading. While there are of course countries or areas which have very few geocoded items, there are others which at first look have a lot, but then 95% of them are un-curated imports from geonames via the bot-created ceb-wikipedia. These items even more need a edit-a-thon to do the necessary fact checking, duplicate checking (but also fact-adding). IMHO the wrong data of such unchecked imports hurt Wikidata much more than missing data. Ahoerstemeier (talk) 23:10, 25 August 2018 (UTC)
Ahoerstemeier this is in the end relative heat. cebwiki imported stuff also here in italy (and we have some good ID of Belgian IDs of theatres if I remember correctly, and so on), it all averages out to what is the core human effort, which makes the difference.
To make a possible example about my region... imagine a map like one I saw in another tool of Manske that I cannot find now, a map where you can enlarge and gives you a global hue. So, you see this general averaged shade when you enlarge. Now the province of Pistoia will currently look globally red, Florence is orange, Pisa is yellowish and Massa-Carrara is greenish. This would happen besides the import from here or there, which is averaged out at that point. You are not comparing different countries but different regions in the same country
Also, you are considering the presence of bot activity stronger than users' work. This is true in certain areas maybe, where you are free not to use such heat map... I expect countries with limited support of human editors to look globally the same, but this is not true in all the areas. And yes, I would like to put the users to work on the items of Botswana but since I have to teach them I prefer the neighboring province. Once they learn, they can make fruitful expertise of their holidays, for example. Or a trainer can use this approach in another "virgin" country.
Otherwise, if we are doomed to be constrained mainly by bot activities, why bother teaching and coopting new users? and creating tool to find the best way to organize this work.
Finally, again, this is for manual work. This means that if you keep finding a recurring problem in an area, you explore it and set up a cleaning process globally, as it should be natural, manual work is designed for long-term results, not fast results. We cleaned up Pistoia intensively and as a result, we had three new IDs for Italy and counting and we "issued" a warning about cebwiki parks items to be revised in Italy.
Just to say, you shouldn't avoid to create an analytical tool because some people might not read it or use it correctly. I'd like to calibrate my next effort for WLM next year both where the situation is better or worse depending on my time and the newbies/volunteers/experience users I have, for example.
Even in very crucial events like San Diego Comic-Con International (Q279385) with a lot of users editing the page over the years we don't really put in the metadata what they are about. Maybe we use a much more specific item for the "instance of" but that's it. How can a machine read that the San Diego Comic-Con International is for example also about.... cosplay?
Can anyone work out why Commons:Creator:Karl Parsons shows (at the bottom) "This Creator template is relying on a Wikidata item, which is missing the Commons Creator page (P1472) property. Please click the Commons to Wikidata QuickStatements.svg icon above to add it!"? The item Karl Parsons (Q6372141) has the appropriate P1472 property. - Jmabel (talk) 17:41, 26 August 2018 (UTC)
Ah, so it would be a stale cache server side? (I'd already done a CTRL-F5, so I know it wasn't stale on my end.) - Jmabel (talk) 18:42, 26 August 2018 (UTC)
Yes, putting "?action=purge" in the URL tells the server to rebuild the page rather than sending a cached version. Thanks. Mike Peel (talk) 19:10, 26 August 2018 (UTC)
Pages_not_connected list not updating completely
I work with "Special pages > Pages not connected to items" for several under-served African-language WPs to identify and remedy connections to Wikidata items. Often I'll "make the connection" for the item, see that the interwiki links display on the connected page - but the page name remains on the list. Recent test: of ten templates in the ZU WP, five disappeared from the list upon refresh, but five remain (after several days!). What might be done about this phenomenon? -- Deborahjay (talk) 06:24, 26 August 2018 (UTC)
I've experienced the problem on different Wikis too. I didn't know ApiSandbox, that could be interesting: how is it possible "to make null edits to relevant pages" using it? Could you @Matěj Suchánek: explain? Thank you very much, --Epìdosis09:47, 26 August 2018 (UTC)
Whoever you think is capable of handling this. But in fact it is not too technical, you will just reach your project, open the tool, fill in according to my suggestion, submit and then after each batch hit continue. Or you can just open each page and save it without any change. Matěj Suchánek (talk) 08:58, 27 August 2018 (UTC)
Next IRC office hour on September 25th
Hello all,
The next IRC office hour with the Wikidata team will take place on September 25th, from 18:00 to 19:00 (UTC+2, Berlin time), on the channel #wikimedia-office.
As usual, we will present you some news from the development team, the projects to come, and collect your feedback.
It's possible to add somevalue and novalue using QuickStatements, and to use the tool on test.wikidata.org as well.
More properties can be searched using haswbstatement. This includes all properties with "external identifier", "string", "item", "property", "lexeme", "form" and "sense" datatypes, except published in (P1433) and cites (P2860). In addition you may make queries without a specified value.
The page creation and edit rate limit imposed in April will be removed in the next days. However, users who are not bots or administrators are throttled to 90 edits per minute (which was set up on all Wikimedia wikis in May).
These new languages are now supported for monolingual text strings: mid, mnw, gmy, rar, haz, uga. The language doesn't show up in the suggester, but you can save the statement and it will be stored properly.
R. Krishnan was a Tamil politician and former Member of the Legislative Assembly. He was elected to the Tamil Nadu legislative assembly as a Communist Party of India (Marxist) candidate from Vasudevanallur constituency in 1977 and 1980 elections.
He was a politician who represented the Alathur constituency in the first two, three, four, and Kerala legislatures . Krishnan (08 May 1914 - 28 January 1995). The CPI He came to the Kerala Legislative Assembly in three and four terms. He was a member of Madras Legislative Assembly from 1952 to 1956 .
Hi. I have used a lot Google Maps Customer ID (P3749) so far but what is about yahoo, bing, baidu and yandex map? I live also in China and I can access those maps, but not google maps.
It seems to me that their ID are difficult to extract without instruction, but maybe someone has some links or studied the problem in the past? Also the google maps CID is not easy but I found the way to do it in the end.
Was it discussed to make an effort in that direction or they are considered not very useful in the general framework so it will never be a "priority"?--Alexmar983 (talk) 05:41, 26 August 2018 (UTC)
I don't know much about these other IDs, but if they are stable, can be readily extracted, and can be put back into a formatter URL, I think they would be welcome as properties. So maybe they just need a good proposal. --99of9 (talk) 23:32, 26 August 2018 (UTC)
99of9 I can make the proposal only if I can prove they can be extracted in some way. Google Maps is not easy to extract, but we know how to. But does anyone know how to do so for at least one of them? I don't. I am sure they use some sort of IDs in any case, that's probable... and they keep it far away form the regular user, as google maps does. I am just surprised it was not a priority to find it, they are probable universal IDs in the end if we can access them and quite useful especially in less covered areas of the world. Maybe google maps CID was not used a lot so it was not considered a target to have another similar one? There was really no discussion at all?--Alexmar983 (talk) 05:19, 27 August 2018 (UTC)
My instinct is to go the other way and make an item called "protein single molecule". When we talk or write about proteins, we hardly ever mean a single protein molecule. So when the Wikipedia article attached to our item writes "QQQQ is a protein that does YYY", I think we would be better to follow that ontology. Also, whatever we do we should coordinate with Wikiproject Chemistry, because "chemical compound" similarly also almost universally applies to a whole lot of identical molecules (but at least they're not usually part of a family tree!). Regarding isozymes of Myoglobin, there is already a family item myoglobin (Q24783412) - does that encompass the variants you need to describe? --99of9 (talk) 23:27, 26 August 2018 (UTC)
@Ivan A. Krestinin: Before starting to modify a classification, we should check the definition because people often have a personal definition which is not the one in the item. Then we need to check the definition of the upper class in order to be sure that the inheritance can applied. If we take protein, we can see that this item is related to chemical substance which is defined as "matter of constant composition best characterized by the entities (molecules, formula units, atoms) it is composed of". Can a molecule be a subdivision of chemical substance according to the above definition ? I don't think so.
For chemical compound, we try to use the structure to define what is a chemical compound or not. Perhaps a similar rule for protein is necessary in order to define what is a subclass of or an instance of protein. Snipre (talk) 19:55, 27 August 2018 (UTC)
What is the item's definition? From what I can see (I don't have proficiency in most of these listed languages and thus relying on machine translation to read these articles which therefore could made mistakes),
The Japanese article say it is about all technology that transmit audio signal via radio, including like mobile phone or radio broadcast but also introduce a certain device.
The Korean article talk about wireless phone as in telephone without wire, including mobile phone and cordless phone
The German article defined it as transmission of speechs via radio/stations, but seems like it focus on more traditional radio communication and do not appears to count things like mobile phones
The Dutch/French/Finnish article includes all form of speech transmission over radio including mobile phones
The Italian article seems to include mobile phones in the article but also separated mobile phone from the subject of the article saying the article's subject is predecessor to mobile phone
The Romanian/Polish article specifically say that it is generally "portable device" for radiocommunication" over "generally short distance"
The English/Greek/Indonesian/Malay article seems to be talking about a particular kind of communication system for transmission of speech that is "very rarely interconnected with the public landline telephone network"
The Esperanto article seems to be talking about a specific type of device that have a role between walkie talkie and telephone
Which of them describe the concept of the item most accurately? Which concept should be used for a new language version article in this wikidata entry? Is there any of these entries that need to be separated?
Also, in various wikipedia language versions like English/Japanese entry, there are still some legacy interlanguage links linking to articles in entry of radiotelephony (Q1799494), what's the relationship between the two concepts? For that concept,
The Korean article talk about wireless phone as in telephone without wire, including mobile phone and cordless phone→radiotelephone (Q2313348)
The Malay article seems to be talking about a particular kind of communication system for transmission of speech that is "very rarely interconnected with the public landline telephone network"→radiotelephone (Q2313348)
The Telugu article talk about something related to human speech generation mechanism and mobile phones
And then, the concept "Marine VHF Radio" mentioned in some of the items above actually also appear in Marine VHF radio (Q1530859) in wikidata, in which:
The English/French/Norwegian Bokmal/Swedish/Portuguese/Sundaese article talk about the Marine VHF Radio
The Dutch article seems to be describing a particular type of it?
The Estonian/Japanese article talk about the entire marine communication system around it
The German article talk about the VHF frequency band use for such service
The French article talk about not just the marine communication around VHF band but also all the other frequency band being used?
It seems quite hard to define things, especially when multiple languages are involved. E.g., the description of wireless communication (Q249) in Wikidata doesn't match the corresponding enwiki article. The latter includes radio, optic and sonic communications as well as power transfer, while the Wikidata item claims to be about only communication using electomagnetic waves, which would be radio and optical communications. Meanwhile, Commons has c:Category:Radiotelephones as a subcategory of c:Category:Mobile phones, which has no correspondence at all with the linked enwiki articles. Ghouston (talk) 03:39, 28 August 2018 (UTC)
Editing of sitewide CSS/JS is only possible for interface administrators from now
(Please help translate to your language)
Hi all,
as announced previously, permission handling for CSS/JS pages has changed: only members of the interface-admin (Interface administrators) group, and a few highly privileged global groups such as stewards, can edit CSS/JS pages that they do not own (that is, any page ending with .css or .js that is either in the MediaWiki: namespace or is another user's user subpage). This is done to improve the security of readers and editors of Wikimedia projects. More information is available at Creation of separate user group for editing sitewide CSS/JS. If you encounter any unexpected problems, please contact me or file a bug.
Better suggestions for constraint values on properties
Hello all,
Currently, when adding a new value in the constraints section of a property, there is no suggestion to fill the value or the qualifier. We’ve been improving this by a few changes that are going to be deployed this week.
When adding a new value in the property constraint statement, a list of suggestions will be displayed and, all the relevant constraint items will be showed first. They will be selected among the list of qualifiers present in the statement “property constraint -> one of constraint” of property constraint (P2302).
Of course, you can still type anything you want in the field to find a value. The full text search has been improved (when typing “none constr” you will also see “none of constraint” in the suggester)
In a very near future, we will also make the following happen:
Same behavior for qualifiers inside the constraint statements. The suggester will pick up the values from “allowed qualifiers constraint”
When clicking on “add value” in the property constraint statement, a suggester menu will directly appear (without having to click on the value field)
The first two changes will appear on wikidata.org on August 30th, the following ones in the next weeks. Feel free to make some tests, and let us know if you find a bug or something that doesn’t behave as expected.
Moveable dates are dates like Easter or Hanukkah that change each year. There are many examples, all of the Jewish and Muslim holidays because they use a different calendar system. Moveable dates present difficulties on Wikipedia to update each year manually, so I created a template en:Template:Calendar date which reads in a JSON file en:Template:Calendar date/holidays/Hanukkah.js .. there are currently about a dozen of these JSON files but there could be 100s of even 1000s of moveable date holidays and events. Thus it's not practical to host the data files on Wikipedia. It seems like a logical fit for Wikidata. I don't know anything about WD. Are moveable dates something that would make sense for WD, and if so how to get started? I'm ready and able to write bots that generates/scrape the data and upload it to WD. -- GreenC (talk) 13:30, 25 August 2018 (UTC)
Jura, that looks fantastic! I just created Q56251842, my first item creation, following your example. Now I need to figure out how to upload via bot.. and retrieve via Lua on enwiki. I will definitely be using your Easter data. -- GreenC (talk) 02:41, 26 August 2018 (UTC)
I think to do Hanukkah that way the user application would need foreknowledge that Hanukkah occurs in November and/or December, and for a given year search every day in those months until it finds 25 Kislev (Q11013530). Then the application would need to calculate 8 days hence (ie. foreknowledge of how long the holiday runs for) to find the last date of the holiday. Or am I misunderstanding how it's setup? I was hoping the application can be dumb about holiday specifics, with the intelligence loaded in the database 1-time so the application only needs to provide a year and a holiday name and get back the dates. -- GreenC (talk) 14:51, 26 August 2018 (UTC)
Jura, This is what is available. I don't see a way to do queries. The way that looks possible (have not tried) is with mw.wikibase.getBestStatements( entityID, propertyID) .. like for Easter, it would be entityID Q51224536 and propertyID P585 - it would return all statements including date of Easter (Q51224536)point in time (P585)1 April 2018 -- GreenC (talk) 21:49, 26 August 2018 (UTC)
Hi Jura, it doesn't seem possible for Wikipedia Lua templates to use Wikidata for moveable date holidays/events, unless they are structured like the Easter example, as a separate entity. I don't know how Wikidata culture works, if it's flexible to end-user application requirements, or it be done a certain way within a Wikidata best-practice. I wouldn't want to upload the data as an entity, and change the template only to have the entity later removed. Can you provide any guidance or suggestions? -- GreenC (talk) 15:29, 27 August 2018 (UTC)
@GreenC: I think there are several tickets in phabricator that try to solve the underlying issue. As I wouldn't count on them working in the medium term, the solution you suggest it is feasible and if it was just for one holiday, it would be fine, but we might end up doing that for all of them. I might be able to provide an other approach in a few weeks, but in the short term, I can only suggest the following: you could generate a Listeria report on Wikipedia with the holidays you are interested in (e.g. like this one) and read that from the template. Alternatively, the MediaWiki date function might allow to get the date directly. --- Jura23:01, 27 August 2018 (UTC)
@Jura1: thanks. I think the template will take a multi-solution approach. There will be 3 ways to get date data. First a calculator that determines the Gregorian based on calendar offsets. The calculators will be third-party applications or plugins, anyone can write, the template doesn't do calcs itself. One already exists for Jewish holidays. If a calculator is not possible, such as with Easter and a few Jewish holidays, it will get the data from an entity such as the 'date of Easter'. If that is not available, it can fall back to using local JSON files to host the dates as last resort. Maybe later add a 4th option for Listeria bot, will keep it in mind for future expansion. -- GreenC (talk) 01:08, 29 August 2018 (UTC)
@GreenC: Sounds good. Just to be sure which ones would be using something like Q51224536? I will try to get the ones with offsets working with queries. --- Jura06:11, 29 August 2018 (UTC)
For Jewish holidays, this is what someone told me: "if you were not aware, between 1 Adar (or 1 Adar II in a leap year) and 29 Cheshvan, the number of days is always fixed. Therefore, any date can be derived as the addition or subtraction of a fixed number of days from the Rosh Hashana (or Passover) within the interval. Since Hanukkah, 10 B'Tevet and Tu B'Shvat do not fall within that interval, the calculation does not work for them. But it works for most everything else." (emphasis added). Adar I and II occurs during February (I) and March (II) and Cheshvan between October–November - you can see why I didn't want to write calculators :) Anyway, a holiday outside that window, such as Hanukkah can't be calculated, though maybe it can using Wikidata? -- GreenC (talk) 13:27, 29 August 2018 (UTC)
w:Hanukkah has "25 Kislev / 2018 date Sunset, 2 December" and {{#time: xjj xjF | 2 December 2018 }} gives "24 Kislev", so it seems to be possible with MediaWiki. Obviously, it don't know how reliable any of that is, but I think anything that moves with a calendar should be doable without storing all dates explicitly. --- Jura13:39, 29 August 2018 (UTC)
Well if you believe it is possible to make a calculator, for now I will use local JSON files storing the dates, and wait for someone to make a calculator, or until I figure it out. Here is the cfg file, you can see some holidays are using "datatype:calculator" with "datasource" being the calculator. Hanukkah uses "datatype:jsonlocal" and "datasource" points to the JSON file Hanukkah.js (which needs to be fixed work in progress). If at some point it's determined a calculator is not possible, then the data can be uploaded to a WikiData entity like Easter, "datatype" set to "wikidata" and "datasource" the entity "Q" number. -- GreenC (talk) 15:43, 29 August 2018 (UTC)
Q56222412
Asia Carra is a young italian singer born in Beijing so she is half Italian and half Chinese. It is exact for the English grammar write “Italian Chinese”? Or there is a different mode? --151.95.37.11218:47, 27 August 2018 (UTC)
If she's a citizen & resident of Italy, we'd definitely say "Chinese-Italian" or "biracial Chinese-Italian". Similarly, if she's a citizen & resident of China, we'd definitely say "Italian-Chinese" or "biracial Italian-Chinese". Other combinations get trickier to express so succinctly. - Jmabel (talk) 19:48, 27 August 2018 (UTC)
Do you have a reference that says she's both Italian and Chinese? You don't necessarily get the nationality of the place you are born or live. I can't find anything on the web except for a few youtube videos. Ghouston (talk) 06:09, 28 August 2018 (UTC)
Ah, yes, I didn't even consider that. If her parents are both Italian, and they merely happen to have been living abroad when she was born, she's simply Italian. - Jmabel (talk) 15:36, 28 August 2018 (UTC)
Category:Swedish words prefixed with ytter- (Q56276138) was recently created, and now has categories on English and Norwegian Wiktionary linked to it. However, the English category is named "prefixed with" while the Norwegian category is named "som starter på" (beginning with). Being prefixed with something is not the same as merely beginning with something. The definition used in the English Wiktionary is that "prefixing" means that the word is formed by prefixing, not the mere presence of some prefix. The English word bevel begins with be- but it's certainly not prefixed with be- the way become is! Rua (talk) 11:48, 28 August 2018 (UTC)
The name could possibly be clearer, and we'll keep that in mind for further changes (it concerns a number of categories, and changing the naming convention isn't a one-minute job). However I can confirm that the intention of all som starter på categories on no-wikt is corresponding to the English prefixed with, and equally som ender på is corresponding to the English suffixed with. In this case forgetful would be categorized in no:wikt:Kategori:Ord i engelsk som ender på «-ful». A possibly better name would be dannet ved prefikset (rp. dannet ved suffikset) maybe ... but when changing the naming convention we'll make sure to move all categories so the Wikidata entry should be correct anyways. Mewasul (talk) 10:49, 29 August 2018 (UTC)
several job openings in or around the Wikidata dev team
Hey folks :)
We have a number of open positions at the moment at WMDE that are relevant for Wikidata including a program manager for Wikidata who will work closely with me. I'd love to see many applications from people who already are a part of the community and understand and love what makes Wikidata special. Find the full list including job descriptions here. If you have any questions please let me know.
@Thierry Caro, VIGNERON: As far as I'm aware, singles, songs and music videos should all be separate entities (see the music video items linked to from Apple Music music video ID (P5655), for example – I also did some work on separating song and single for those topics). Music videos may have director (P57), songs may have tonality (P826), and songs and singles may both have charted in (P2291) depending on the methods of particular record charts. (Sitelinks should normally be hosted by the song's item, and songs, singles and videos can all have has edition or translation (P747), as far as I'm aware.)
@Jmabel: I think it was because it was done this way on Thriller (Q380825), which is one of the example items, and it made sense to me so I used it on the items that I worked on. I think the use on Thriller (Q380825) is actually incorrect, though, since the version of "Thriller" used in the music video is different to both of the versions for which Wikidata items currently exist. The audio of the entire music video (not just the part comprising the song) might also have been separately released (I don't know if this is the case). For the music videos for which I created items, all contain the entire uninterrupted audio track of the song. Jc86035 (talk) 16:51, 29 August 2018 (UTC)
@Jc86035: I'd rather see a more specific pair of properties for this. It's common enough that we should probably eventually have tens of thousands of these, so I think the specific relationship of a piece of music being used in a video specific to that piece of music merits distinct properties. - Jmabel (talk) 22:35, 29 August 2018 (UTC)
@Jmabel: I think it would be useful to have a property or two for the use of a full audio track in a film/video, as well as for a non-repetitive but truncated use of an audio track in a film/video. However, since songs can have multiple music videos and music videos can contain multiple songs, I think having properties for the relationships "the music video for this song" and "the song of this music video" would be too specific. Jc86035 (talk) 07:50, 30 August 2018 (UTC)
Duplicate creation by GZWDer_(flood)
There is a discussion about the ongoing creation of duplicates by GZWDer_(flood) at Wikidata:Administrators'_noticeboard#Please_block_Special:Contributions/GZWDer_(flood). Essentially, the user chose to change the approach with PetScan and create new items even if there are already existing items with the same name. This instead of using Duplicity. The user is known for leaving it to others to clean up after their sprees. Please comment there if you don't mind (or if you do mind). --- Jura21:20, 29 August 2018 (UTC)
Put mw.loader.load( '//www.wikidata.org/w/index.php?title=MediaWiki:Gadget-dataDrainer.js&action=raw&ctype=text/javascript', 'text/javascript' ); // [[MediaWiki:Gadget-dataDrainer.js]] to Special:MyPage/common.js, and there will be a new dialogue linked from the “More” tab next to the search field. (Not sure why I can’t find it in the gadgets section of the preferences.) —MisterSynergy (talk) 14:15, 30 August 2018 (UTC)
For Nijinsky (Q26798) and other instances of racehorse (Q10855242), the property mother (P25) is flagged as an error because racehorse (Q10855242) is not a subclass of any allowed class for subject of this property. Among allowed classes is listed eukaryote (Q19088), which in fact is an instance of taxon (Q16521). I've hard time to figure where this is breaking, and I don't want to meddle in the metaclasses structure.
@Succu: Because from lingustisc view is it problematic. When you says about man that is male organism (Q44148) in czech language, is it considered as near vulgar (he is very sexual active). And the same situation is in many other languages. JAn Dudík (talk) 20:10, 28 August 2018 (UTC)
I would propose limiting the scope of father (P22) and mother (P25) to mammal (Q7377). The mechanistic definition of biological parent which applies to humans only goes as far as mammals. Even though most other animals have biological sex, the mechanism of gender determination isn't the same as that in humans and the labels "male" and "female" are assigned by partial similarity with human biology. For example drone (Q650665) (traditionally considered "male") only have one parent (traditionally considered "mother"). Hippocampus (Q74363) "males" get pregnant. Since the human definition of "mother" and "father" doesn't fit these other animals, we shouldn't use the same property. Deryck Chan (talk) 22:42, 30 August 2018 (UTC)
I'm not sure the constraint route is the right way to go. Irrespective of how "female/male" is defined or determined in a species, it is often defined for non-mammals. Our definition of mother is "female parent", which apparently still stands in the examples you gave. But it may be worth opening up another option: for asexual reproducers or those without a male/female definition, it would probably help to have a non-sexed Property:Biological_parent, of which mother and father could be supproperties. --99of9 (talk) 00:13, 31 August 2018 (UTC)
Please add Project:Reference desk (Q3907025) to your watchlist, especially if you speak Spanish. The Spanish-language description has been changed to something irrelevant twice recently. I'm not sure how it's happening, but if it were on a few more watchlists, then perhaps these accidents would get reverted sooner. WhatamIdoing (talk) 15:41, 30 August 2018 (UTC)
Fixing the search to be useful again; aka making the simple, simple again
In times gone by, one used to be able to do a simple search for a person by partial name and partial data field, eg. throw in a year of birth. Great functionality if you were looking to match someone, or knowing to create someone, etc.
If you want simple people to help, you have to retain the simple and useful features. Sometimes the geniuses here aim for the high end, and forget the simple tasks. How are people meant to prevent duplicates when the simple things are made so hard? If the simple things are hard, people will just walk away. — billinghurstsDrewth00:15, 28 August 2018 (UTC)
You still have that simple search capability, but you need stay in Wikipedia and use Wikipedia search, or on Commons, and if you scroll all the way to the bottom you will see the clickable Q results. However, in the example you gave, you first pull up all the objects and research papers on Bonaparte. Wikidata has not only changed search, but has grown so exponentially that you get lost in the results. You can better run your own simple query script. In my field of interest (paintings) I use the same search time and again, so I have become a fan of the SQID tool to look up museum paintings like this. Jane023 (talk) 06:59, 28 August 2018 (UTC)
If that's the correct answer, that's a bad situation. The ordinary search ought to be something readily used by newbies and give reasonably accurate and naively expected results. - Jmabel (talk) 15:38, 28 August 2018 (UTC)
There is no "correct" answer, but personally I haven't used search much the past few years and prefer to use Google to search Wiki(p/m)edia projects, since Google shows me stuff not labelled in my search language. Jane023 (talk) 16:18, 28 August 2018 (UTC)
Yes of course this is open to interpretation. I suppose the most succinct correct answer then is: No, something like "Bonaparte 1769" would probably not have found Napoleon (Q517) a few years ago, because the birthdate was not findable as text, either in that shortened form, or with the birthday included. Jane023 (talk) 08:03, 29 August 2018 (UTC)
If the basics in search cannot be done, then what is the point? Sending someone to a third party search function is just weird. Searches as described used to work, and now they don't. I was going to give examples, but in the end, it is just bashing one's head onto a wall, and of that I have had enough. I don't see it working well for the sister wikis any more, with little benefits for lots of input. I feel that it matters NOT to others here. Some of us want to do our work at the other wikis, and have it work well here. I don't see that the reverse is the case. Have fun, it seems that the mission has changed. — billinghurstsDrewth08:15, 30 August 2018 (UTC)
@billinghurst: I just checked through the page history for Napoleon (Q517), and on the wikidata side, 1769 was NEVER in one of the labels or descriptions. The only way wikidata search ever would have found it with your query would be if one of the sitelinks had "1769" in its title - which is possible. But wikidata search has only ever searched labels, descriptions, and sitelink titles - it has never searched the values of statements. See Phabricator ticket T163642
for an effort to improve that recently, although in my opinion it went astray - but it's trickier than it sounds to do what seems obvious here for a variety of reasons. ArthurPSmith (talk) 18:51, 30 August 2018 (UTC)
It was an example only, one picked out of the blue. Don't worry, I have had enough and I am predominantly gone, and no longer involved in the politics here. If simple search isn't important to you on site, that is your business. — billinghurstsDrewth01:40, 31 August 2018 (UTC)
Hmm, I think you are misunderstanding my comment? Simple search is very important to me - but it's much harder here than on the wikipedias because the "pages" we see are not plain text, they are data which looks different in different languages. I suggested an approach on the phabricator ticket to handling this, but it was tried and breaks for at least some identifier cases. It's not clear what the right method is. If you have a specific suggestion here, I'm sure it would be very welcome on the phabricator ticket! ArthurPSmith (talk) 15:58, 31 August 2018 (UTC)
Dutch bot descriptions generally include year of birth and year of death. So many items about people could be found thanks to that. Q517 doesn't have that though. --- Jura16:09, 31 August 2018 (UTC)
Ha! That would nice but I also feel that in another 5 years there will be other even larger datasets added, so you will keep having to turn of more and more search areas or "ontology trees". But maybe by then Wikipedia search will have improved so that when you search for specific things not in Wikipedia (yet) it will pick up all related articles, including related Q numbers, with or without matching search text. Jane023 (talk) 17:48, 31 August 2018 (UTC)
@Epìdosis, Jane023: I've suggested something along those lines too - basically supplementing the namespace filtering (i.e. Items vs Properties, Lexemes, Wikidata proper, etc) that wikimedia software naturally does out of the box with the ability to filter on classes (P31/P279 category) - but technically I think that's also very hard to do... Note wikibase IS open source software that anybody willing to work on this could try contributing to! ArthurPSmith (talk) 18:14, 31 August 2018 (UTC)
Equivalent to e.g Property:GND for Wikidata Object IDs?
What is the property that holds Wikidata Object IDs? I would like to use a property Wikidata Object ID on an external website but I am not able to find the equivalent property here that is defining the URI formatter or the regexes etc. Obviously you are not using it here since this would be a self-reference however Wikidata is big and important enough to get a respective property I guess. Thanks for your infos and help. Cheers --[[kgh]] (talk) 14:15, 29 August 2018 (UTC)
New user script: browser/desktop notifications for watchlist and recent changes
Hi all! I wanted to announce a useful user script I’ve written recently: User:Lucas Werkmeister/changeslist-notify.js translates live updates on your watchlist or recent changes into browser/desktop notifications. To use it, add the following line to your common.js (or even your global.js – it should work on any Wikimedia wiki):
Then, open a new tab with your watchlist or a recent changes page (e. g. lexeme changes), click the “live updates” button, and then just leave the tab open while you go about your business. If there’s an update, you should receive a notification within the next ten seconds. To disable the notifications on a particular page (e. g. on an unfiltered recent changes page, where updates are extremely frequent), press the “stop notifications” button right next to the “stop live updates” button (only visible when live updates are enabled).
Please keep in mind that while this means you may see new messages on talk pages more immediately than before, no one is entitled to an instant response by you, and there’s no such thing as “read receipts” on-wiki, so don’t feel pressured to react to notifications as fast as they come in. Personally, over the past few days I’ve found this user script very useful because it relieved me of the urge to constantly check my watchlist, but if you find that it stresses you out, please don’t use it! I’m sure this is not something for everyone, I just hope it will be useful for some people.
part of (P361) seems fine to me - I'm not convinced we need constellation (P59) in the first place, given the relatively small number of them, but I suppose there are plenty of stars they could apply to. But there aren't more than a handful or two of asterisms, are there? You can tell the child is a star, and the parent is whatever its instance of (P31) property says, so the meaning of part of (P361) should be perfectly clear. ArthurPSmith (talk) 13:50, 31 August 2018 (UTC)
OK, thanks for your input! I've now entered the Big Dipper stars as part of (P361) Big Dipper, and I also added the inverses, has part(s) (P527), for Big Dipper. I actually did this using the QuickStatements tool that I just learned about from your user profile, so thanks for sharing that information too! :) And yeah, Wikipedia only recognizes 25 asterisms, so a new property would seem like an overkill. --ZeroOne (talk) 19:35, 31 August 2018 (UTC)
Question given the various comments of the participants, what outcome would you consider appropriate? 
Notified participants of WikiProject Video games
Comment For which elements on that site would you want a property? The only elements I see are the character (some ~100 of them), but I don’t think this warrants a property. Jean-Fred (talk) 09:59, 31 July 2018 (UTC)
InfoWikidata:Property proposal/Match Score--Mad melone (talk) 06:06, 3 August 2018 (UTC)
Done --Tagishsimon (talk) 15:58, 17 August 2018 (UTC)
⟩ located in the administrative territorial entity (P131) 
and polyploid (Q44445817) 
WikiProject Books has more than 50 participants and couldn't be pinged. Please post on the WikiProject's talk page instead.
humbly veto any qualifier (e.g.language of work or name (P407)) suggestions, as they really give nothing helpful, and, as of now, not all wikis support that way. --Liuxinyu970226 (talk) 22:43, 9 August 2018 (UTC)
