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.
I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. Moebeus (talk) 00:03, 5 March 2020 (UTC)
Suspended, deleted and deactivated website accounts
Property proposals for PoS blockchain cryptocurrencies
Hi, I recently discovered that while PoW currencies like Bitcoin are reasonably describable using our existing properties, I found that more were needed to accurately model PoS coins. I went ahead an created a series of property proposals. You are very welcome to comment on them and support those you like. Thanks in advance.--So9q (talk) 16:07, 1 March 2020 (UTC)
What is going on? At Property talk:P5972 there is a really weird thing happening (see https://archive.is/gHEDf). The name of the page is shown as "Internal error", and the content of the page shows as "[XlxbPQpAMFkAABjowJAAAABC] 2020-03-02 01:02:53: Fatal exception of type "InvalidArgumentException". The code of the page seems fine; I tried pasting it into my user page and previewing and this didn't happen. Trying to preview it on the actual doesn't work right. What is going on here?DemonDays64 | Talk to me01:06, 2 March 2020 (UTC) (please ping on reply)
Importing movie box office data of a single small distributor into wikidata
Hi! I was discussing this on my home wiki (huwiki) and I was advised to ask for your opinion. There is a small Hungarian indie movie distributor Mozinet who regularly publish the box office and attendance data for the films they distribute on their website. For Hungary, this is a niche market and most of these data don't make it to the big box office reviews of other sources. to be able to use this as a source/reference for articles in huwiki, the page at http://www.mozinet.hu/hu/beveteli_adatok needs to be archived since they refresh the data weekly. The issue is that they publish the data in Google Docs, which archive.org or other archiving pages are not able to handle properly (tabs...). The question is:
is it okay to use something like Openrefine to import the csv
and if so, what kind of structure do you recommend to use?
I believe that copyright-wise this database qualifies as spin-off, so the data would be okay to copy.
If the above cannot work, would you be able to advise for a solution?
Have you tried to engage them and ask them to publish the data in a format that archive.org can actually archive? It seems to me like it should be in their interest to have people engage with their data and their current solution is likely not intentional. ChristianKl ❪✉❫ 18:46, 2 March 2020 (UTC)
@ChristianKl: It's a two person company, unlikely they will invest in finding another way of publishing the data just for our sake. (Not by chance they use free Google Docs for that). Especially that they would have to regularly be able to update that, hence the Google Doc solution, without having to pay a webmaster to do that work every week. Teemeah (talk) 21:47, 2 March 2020 (UTC)
Wikidata weekly summary #405
Here's your quick overview of what has been happening around Wikidata over the last week.
July 2-4, Lisbon: WikiData Days 2020. Call for proposals is open until April 15.
Program submissions for the Celtic Knot Conference are open until March 30th. Submissions about languages on Wikidata, GLAM or supporting minority languages are very welcome.
Wikidata graph builder is a front-end on top of the Query service, allowing to easily build graphs.
Other Noteworthy Stuff
The Wikidata Languages Landscape dashboard provides insights into the ways languages are organized and used in Wikidata and across the Wikimedia projects that reuse Wikidata.
A quick question: would it be okay to add interwikis like this: [[hu:incubator:Wn/hu/Luxemburg ingyenessé tette a tömegközlekedést]] to Q86746352, i.e. in the case of Hungarian, pointing to the Incubator? - Xbspiro (talk) 20:00, 2 March 2020 (UTC)
I think it's frivolous and unnecessary. Wikidata is not meant to "look good" like an article. It's primarily a repository for machine-readable data, some of which gets digested and regurgitated by infoboxes and other outlets. The slap-dash hodge-podge of items, properties and qualifiers need not be grammatically correct, nor aesthetically pleasing. -Animalparty (talk) 03:01, 4 March 2020 (UTC)
Is there any indication at http://www.unece.org/cefact/locode/welcome.html that it is restricted to cities? If so, is the definition of city used for locode absolutely compatible with every other authority's idea of what is & is not a city. If you have any doubt about either of these things - and you should - then you should think again about your constraint. In general, you're best off announcing your intention on the talk page for the property & seeing if anyone has anything to say. --Tagishsimon (talk) 19:09, 3 March 2020 (UTC)
When was the following alert added: "An entity with date of baptism in early childhood should also have a statement religion." Seems like a really bad idea. After all the religion that someone was baptised in as a baby may not be the one they espouse as an adult, and currently there is a standard that Religion should have at least two references, so it seems like there should be a degree of caution about urging people to add more Religion statements. Levana Taylor (talk) 02:30, 1 March 2020 (UTC)
You are baptized into a religion, so yes, you should add the religion you are baptized into. If the religion changes you can add a start and end date. --RAN (talk) 04:46, 1 March 2020 (UTC)
Now, with that last part, I agree. Definitely it would be good to state what religion that baptism was as a qualifier. But I am still reluctant to think that that baptism should be used as a Religion statement, not, at any rate, unless you also know something about the person's adult religion & can set an end date to their baptismal religion if they no longer followed it. Otherwise, you are at risk of letting unwarranted assumptions be made about a person -- should never assume their baptismal religion was their adult religion. Levana Taylor (talk) 05:41, 2 March 2020 (UTC)
If I remember correctly, there were periods in UK history (in the 17th century?) when parents were obliged to have children baptised in their local parish church (and even attend regularly?), even if they did not subscribe to the Anglican religious denomination (eg Quakers, Catholics, other non-conformists?). So it may be that religious denomination (as a main statement) cannot necessarily be deduced from a baptism record, even for the point of time of the baptism, without additional evidence. On the other hand, noting the denomination as a qualifier for the baptism statement seems harmless; though I am not sure I would go so far as to make it a requirement Jheald (talk) 12:40, 4 March 2020 (UTC)
The property is mostly relevant when the date of birth can't be found. If the date is referenced, I'd expect it to include a church's registry. --- Jura10:20, 4 March 2020 (UTC)
The large game and natural history of South and South-East Africa
Thank you! Unfortunately the error message was completely uninformative: I should have checked for another version earlier. -- The Anome (talk) 22:40, 2 March 2020 (UTC)
@The Anome: Did you receive the error message: Could not save due to an error. The save has failed.? I know, it isn't the most informative messsage ever. However for future reference keep in mind that a common cause of this message is exactly this situation, the link already existing in another item. Either ways no worries, and thanks for finding this duplicate! --Kostas20142 (talk) 22:47, 2 March 2020 (UTC)
Yes, that was the message. I was sure I did a check first, but I think what must have happened was I was outraced by a bot which created the Wikidata item almost immediately after I created the Wikipedia article, and before I tried to make the item. -- The Anome (talk) 22:55, 2 March 2020 (UTC)
I think it is bad practice for a bot to create a new Wikidata item immediately (in this case 45 minutes) after the creation of Wikipedia article. In many cases there may already be an item for the subject, or, as here, the creator of the article will create the item themselves. The bot should have waited. It seems this have given problems before according to the bot permission --Dipsacus fullonum (talk) 07:05, 3 March 2020 (UTC)
It also happened to me when moving sitelinks at a time of high WD load. Often then the original link is deleted but the new one is not created. This was claimed to having been fixed but it still happens. --SCIdude (talk) 17:01, 3 March 2020 (UTC)
Would it be worth trying to encourage Wikipedias to show a pop-up to encourage article creators to look for a Wikidata item (perhaps with any suggestions), whenever a new article is created? Jheald (talk) 12:28, 4 March 2020 (UTC)
What wikidata project should I contact for queries regarding UN/LOCODE (P1937)? Should there not be some property assigned to properties to link them to wikiprojects?
I want to clarify the use of UN/LOCODE (P1937) but I don't know what project owns the property or who to ping in the question. Is there some simple way to find the project that owns the property? Could such a relation not be captured in the property itself? Iwan.Aucamp (talk) 15:48, 3 March 2020 (UTC)
at the moment I think about how Wikidata can be made more userfriendly for people who are not so deep in the structures of Wikidata. I think forms are a way to create items easier and with the structure as needed for that. A tool to create forms is Cradle. In Cradle it is possible since a few weeks as far as I know to see the properties in the language of the user inferface. So it is now easy to use it also for people who speak another language. I for myself do not know all English Labels of the Properties and so this is very helpful. I think it were great if some of you would create forms for topics where the structure of the items is clear. I think this is helpful. After there are a lot of things who dont have a clear structure a page where the structures how the items should look like are documentated in a way what is easy to understand would be helpful. --Hogü-456 (talk) 20:20, 3 March 2020 (UTC)
@Jmabel: Cradle is a Tool. So as far as I know this is independent of the settings for the language in Wikidata. From my point of view the change in Cradle improved the possibilities to create items. -- Hogü-456 (talk) 20:55, 3 March 2020 (UTC)
If there are fields that you think could benefit from a form, you could ask at the relevant WikiProject. --- Jura10:12, 4 March 2020 (UTC)
B. J. Elder: duplicate items
Hello! Items Q86837071 (B. J. Elder) and Q3631677 (B.J. Elder) are about the same person. The second one (Q3631677) was already present, while the first was automatically created when I created the corresponding article on en.wiki (here). These items should be unified – also, the second item includes a categorization of Elder as a volleyball player, for unknown reasons, and cites a wrong birthdate. Since I've never actually edited on Wikidata before I prefer to write here instead of potentially making a mess! Thank you! --Triple 8~enwiki (talk) 23:40, 3 March 2020 (UTC)
I have noticed that one the many problems and headaches caused by The Peerage import is that, on rare occasions, birth & death dates seem to get conflated if an item is a redirect to another on a Wikipedia. Like if an unremarkable son on en-Wikipedia is a redirect to his famous father, the vital dates of the father may be applied to the son, or vice versa. I have not investigated this closely, but it's something to watch out for. -Animalparty (talk) 21:47, 5 March 2020 (UTC)
Is there someone at the Peerage that we report errors on their side to? Or do we have a page setup to keep track of errors from their side, like we do for VIAF? --RAN (talk) 05:08, 6 March 2020 (UTC)
It looks like there has been more than one import. Sometimes names and dates may be corrected, but they also have pages with warnings saying that they are under construction and probably contain errors, and these have been used as sources, both in the import and in Mix'n'match, which seems to be older data. Also when the site removes an entry the same identifier may be used for a different person, so Isabella Fane (Q76213523) is "named as Charles Roger Lupton" because #609239 was previously a conflation of Charles Lupton (Q5080395) and Charles Athelstane Lupton (Q75864909) - I don't know if this only applies to "under construction" pages, or to all pages. Peter James (talk) 11:16, 6 March 2020 (UTC)
There is only about 100 current errors but that is using the detection method of people who died before they were born. I am sure there are more subtle errors. --RAN (talk) 21:56, 6 March 2020 (UTC)
@Richard Arthur Norton (1958- ): Worth noting that if you make very poor decisions when "correcting" data, as you did here when you presumed that the DoB and DoD were transposed (in fact the problem was an erroneous DoB by 1 century), then you just make things worse. --Tagishsimon (talk) 22:08, 6 March 2020 (UTC)
I have corrected over 1,000 of these errors and occasionally I will make a mistake, so there is really no need for the sarcastic tone and the scare quotes with "you make very poor decisions when 'correcting' data." That is 999 errors corrected and one error transformed into another error. No data set is error free and correcting errors can introduce new errors, that is just how information theory works. --RAN (talk) 22:23, 6 March 2020 (UTC)
It suggests you're working on the basis of your assumptions, rather than checking the data against sources. That will not end well. --Tagishsimon (talk) 22:31, 6 March 2020 (UTC)
Suggested metrology properties: "quantity symbol" and "unit of measurement"
Hello. Physical quantities and units are an important aspect of scientific communication as they are the means to represent quantitative results. As a simple example, the physical quantity force, with quantity symbol, can be defined by the formula , with mass and acceleration . The number of physical (and chemical, biological, ...) quantities is in principle unlimited, because any relation between quantities gives rise to new ones. Having one item for each quantity mentioned in the literature - with label and description in each language - with sufficient information to determine with certainty which quantity the item represents, will help with the unambiguous communication across language barriers in the scientific context.
Take for instance the quantities luminous flux, luminous intensity, luminance, illuminance, ... Those are all quantities related to light, but each has their own, distinct definition. They have "radiant ..." counterparts, covering more than just the visible spectrum of light. Properties that are important for correctly identifying a quantity include a formula, the quantity dimension, the quantity symbol and the unit of measurement. The representation of the latter two in Wikidata is currently suboptimal, in my opinion.
I'm writing to encourage people interested in metrology and science to participate in the discussion and creation of the following two properties:
unit of measurement: That one is more tricky, please check the extended motivation as to why this is useful.
(The latter has one "oppose", but since then I have extended the motivation based on literature research and my experience so far in improving quantities and units in Wikidata.)
I'm with a large U.S. university library and we are looking to set up a WikiBase instance for experimentation. We are currently assessing hardware and software requirements. Thanks to the excellent and simple documentation on the WikiBase site, we are aware that we will need a minimum of 4GB of RAM and 6GB of disk space for an for an empty base.
However, what are the recommendations for a sizeable sandbox? We know of several initiatives in the US and in Europe that have experimented and even adopted WikiBase but we cannot determine their storage specs. Any idea on what these might be? Even better, is there a rough estimate of the storage reqs for record anyone can provide? Thanks in advance for any advice.
Kindly,
Digitalibrarian
Wikidata with close to 100 million items seems to have a total of about 1.5 TB of changes since it started, based on this graph. From this email. the RDF dump for Wikidata (as of almost a year ago) is about 350 GB. That suggests sizes around 10-20 kB per item; presumably you need a small multiple of that to have things run comfortably. ArthurPSmith (talk) 17:55, 6 March 2020 (UTC)
Open sourcing "wiki-slurpee" a tool for scraping data from google
Awhile back I posted here about a browser extension I was working on for scraping Google's knowledge graph data into Wikidata. Now it's open source and available.
@MisterSynergy: sure, I can do it like this but it seems that's not possible to do for freebase ids (it yells at me when I try to add a Freebase ID (P646) to a reference). But for now at least I am already adding the freebase/gkg ids as a primary property. BrokenSegue (talk) 20:42, 6 March 2020 (UTC)
Ah, ok yeah I can do that. Yeah sometimes for technical reasons I cannot get the actual ID but I add it whenever it's available along with a timestamp. BrokenSegue (talk) 21:26, 6 March 2020 (UTC)
Would it be possible for you to add support for Spotify, Play Music, Deezer and TuneIn? Those are the ones you typically see when you google musicians. --Trade (talk) 23:22, 6 March 2020 (UTC)
Is there a Wikiproject that concentrates on error detection and tracking? I want to group all the error detection searches in one place and have one place that tracks all the lists of errors that need to be corrected at the source, and who to contact at the source. We have searches for people with the same first and last name, people who died before they were born. Dopplegangers that need to be merged and a few other error detectors. If there isn't one already I will create one. --RAN (talk) 21:59, 6 March 2020 (UTC)
It seems to be the correct ID. If there is only this one, it should probably be added as an exception to the constraint. If there are many more, a change to the constraint would be better. Peter James (talk) 17:28, 6 March 2020 (UTC)
What's the right way to handle a Wikipedia article that basically refers to one of two different things, ambiguously?
The English Wikipedia article Calorie refers to either the "physics" calorie or the "food" calorie (= 1000 physics calories). That ambiguity is reflected in the label of the corresponding entity here, small calorie (Q130964), which says "unit of energy with different values".
But Q130964 is trying to be two things at once. Its description says, and it's tied to a Wikipedia article that says, that we might be talking about one thing or the other. But most/all of the rest of the claims at Q130964 are clearly talking specifically about the "physics" calorie -- that is, the rest of the entity is not something with "different values", it's trying to talk about just one. And clearly we ought to have distinct entities that each unambiguously talk about exactly one kind of calorie.
one disambiguating between the two, which can be properly tied to w:Calorie.
It's probably safest for (1) to remain Q130964, and for (3) to be a new entity, and for the w:Calorie sitelink (and some/all of the other sitelinks) to be removed from Q130964, and attached to the new entity. I suspect there are existing entities linking to Q130964 as the "physics" calorie, but I don't know how to track them down. I suspect there are few if any existing entities linking to Q130964 as a deliberately ambiguous item.
What I don't know is how to link (3) to (1) and (2). I could give it two different from (P1889) claims, but that's not quite right, because it's more the same as them than it's different from them. I could give it two has part(s) (P527) claims, but that's not quite right, either. Really I want a property that says "One of the things this ambiguous name might refer to is...", but I don't think we have (nor am I suggesting we add) that strangely specific property. But do we have anything that's close?
(Note, by the way, that there's actually another sort of disambiguating entity lurking in the wings here, Calorie (Q16768157), but that's clearly a different animal, not usable for (1), (2), or (3).)
This situation is similar to (but not really the same as) the "Bonnie and Clyde" problem, I guess. (It's sort of "Bonnie or Clyde".) —Scs (talk) 18:03, 6 March 2020 (UTC)
@Tagishsimon: I'm not sure how the fact that a bot is scraping it makes it obviously useful, especially because the bot is finding no countries to associate with it. Am I missing something? - Jmabel (talk) 23:47, 6 March 2020 (UTC)
does not find Q2195. What am I doing wrong? (Or is there a query service help page I should be asking this question on?) —Scs (talk) 12:33, 7 March 2020 (UTC)
@Scs: That'll be to do with statement ranks on the various P31 values. Conflation is marked as preferred, meaning it will be the only response to a wdt: query. (see also this.) Try:
Oh! Wow. Thanks. I shall have to study this. (I confess I don't really understand that syntax p:P31 [ps:P31 wd:Q21484471].)
Actually, I guess I need to learn more about "rank" in general. (And it occurs to me to wonder: could rank be part of the answer to the ongoing question of whether Atlanta is located in the administrative territorial entity of DeKalb County, or Fulton County, or the state of Georgia?) —Scs (talk) 13:40, 7 March 2020 (UTC)
@Scs: I've recycled a couple of talk page responses, dealing with qualifiers and references, and truthy/non-truthy statements, which might help you understand both the weird syntax, and the absence & means of finding Q2195, at User:Tagishsimon/WDQS. Meanwhile Wikidata:Request a query is probably the best place to bring WDQS questions. hth --Tagishsimon (talk) 15:28, 7 March 2020 (UTC)
I'm reminded of the old story about the blind men and an elephant (Q1218005). You've got this nice explanation (which I do hope to wrap my head around) of the rather complicated situation from the perspective of RDF and SPARQL queries and Wikidata's actual data structures. Me, so far my own -- no less complicated but rather different -- understanding of all of this comes from working almost exclusively with the JSON representation in the dumps, which give a very different perspective...
@Scs: Done that ... it's the up & down arrows to the left of the value field. Good luck with SPARQL; don't think it'll take you long, but you must grok the data model. :) --Tagishsimon (talk) 18:33, 7 March 2020 (UTC)
Automatically add and fill qualifier of a property based on content of the item in it?
The property translation (P5972) is controversial (to understate it), but it's to my knowledge been at least for now agreed upon (see deletion discussion) that it is a mildly acceptable way to link lexemes because we don't have a good consensus on stuff like whether verbs should have items to link them together.
So translation (P5972) is a property I end up using a lot when making Spanish lexemes that already have English equivalents. However, there is one problem that makes it a lot slower to use and add to many lexemes quickly: to make the translations say what language they are in without having to click the links and navigate away, one must manually add a value for language of work or name (P407) each time, slowing down the process of adding this already rather unwieldy property to lexemes.
As each instance of translation (P5972) contains a value that is a lexeme sense, is there some way that this property could be changed to automatically show a value of language of work or name (P407) on it that is equal to the language of the lexeme linked in it? I am not technically knowledgeable enough to understand how this would work or why it could not, but if it did this this property would be a lot faster to use to help to make navigation between lexemes meaning the same thing easy. Thanks! DemonDays64 | Talk to me23:09, 7 March 2020 (UTC) (please ping on reply)
I'd rather try to tweak to the GUI for property values that are senses. Even if no glos is present in the user's language, the lemma should be displayed. Not sure if anything will happen, as I'm not aware of any recent development for lexemes. Maybe the qualifier could be a solution. --- Jura08:50, 8 March 2020 (UTC)
Bishop Otter College was founded in 1839. In 1977, it merged with Bognor Regis Training College and became the University of Chichester. This history is discussed on the Wikipedia page for the University, and w:Bishop Otter College redirects to w:University of Chichester. I think, though, it would be best to have separate WD items for the institutions. My questions are: 1. Should I link the item for BOC to the WP redirct page in some way? 2. What is the correct way to write the statement "merged with X to become Y?" Levana Taylor (talk) 17:48, 8 March 2020 (UTC)
Yes, Bishop Otter College should have its own item, as a distinct entity, not simply a name change. Unfortunately, it seems there are no "correct ways" of doing anything on Wikidata: it's the wild west with little guidance and lots of chaos so whatever works best, I say do that. Relevant properties for former institutions include merged into (P7888), replaced by (P1366) and followed by (P156). A redirect on Wikipedia can be attached to a Wikidata item by temporarily removing the redirect (e.g. delete the "#REDIRECT") and then linking the page to its corresponding Wikidata item. -Animalparty (talk) 18:06, 8 March 2020 (UTC)
Wikimedian photographer details on Structured Commons
Hello. I'm trying to add creator (P170) to photographs which I uploaded to Commons (as shown here). Then I realised that, in order to do that, we need an item about the particular user. I've looked at WD:N and also stumbled across this discussion from 2018, but cannot decide if creating items for users is allowed. Looking at this list, I've concluded that it is probably allowed. But when attempting to create the item, I get a warning "Per Wikidata notability policy adding links to userpages or userspace is not allowed". What am I missing? If creating a user-item is not allowed, any suggestion on how to add P170? The goal here is to fill structured data on Commons. Appreciate any clarity on this. Rehman10:56, 8 March 2020 (UTC)
Thank you for the link, GZWDer. It's a pity that discussion was mostly inconclusive. It was going somewhere, and then basically died down without feedback. Rehman02:54, 9 March 2020 (UTC)
(Copied from the Wikimedia Commons village pump as it would be more relevant here.) I have seen several Wikimedia Commons photographers add a creator link, for example this Wikidata page of Benoît Prieur which they use for authorship information on Wikimedia Commons. I see no reason to exclude Wikimedians who wish to include themselves like this on Wikidata, as this would actually make it easier for them to include authorship information and attribution information on Wikimedia Commons, this would make it within Wikidata's scope as it fulfills a structural need for another project. -- Donald Trung/徵國單 (討論 🀄) (方孔錢 💴) 18:47, 9 March 2020 (UTC)
Call me a pessimist, but I think eventually, inevitably, the Wikidata "notability" dam will break and anyone who has ever posted a photo, tweeted a tweet, made a Facebook account, or had their name in a yearbook, phone book, or website anywhere will be eligible for inclusion, whether they want it or not. Data demands ever more data. -Animalparty (talk) 19:02, 9 March 2020 (UTC)
Querying the URL datatype with haswbstatement is now possible (phab:T243693). It will take two to three months before URLs are indexed for all Wikidata items.
Wikidata Bridge: more style fixes, preparing a prototype to show how we will display references
Fixing various production errors
Monitoring the run of wb_terms migration
Fixing an issue with the Commons files search field (phab:T196165)
Somewhere last year I noticed on this page that there were about 60,000,000 (sixty-million) Wikidata pages but this has grown with around 20,000,000 (twenty-million) pages in a relatively short amount of time, is there a specific reason for this explosive growth? And where can I find more detailed statistics about the content created and how many items belong to what type of data? I honestly find it curious how quickly Wikidata grows in the millions surpassing even Wikimedia Commons in number of pages in such a short amount of time.
How many of these pages are bot-created Vs. Human-created? It's quite interesting and exciting seeing Wikidata grow so quickly in such a short time span. -- Donald Trung/徵國單 (討論 🀄) (方孔錢 💴) 20:26, 9 March 2020 (UTC)
This Grafana dashboard is a good entry point for detailed statistics around Wikidata; there are also valuable links to external tools provided.
Most activity here is automated in some way, and this also applies to item creations. I'd be surprised if more than 5% of item creations were done manually.
Particularly astronomical object (Q6999) items have been amassed since early November (from 400k to 5.8M). Not sure whether scientific article items are created in large numbers currently as well.
Hi. For some reason, I can't connect item Q87411076 with French WP article fr:Forte (film). Maybe it's because it's already connected to an item but I couldn't find it and that's why I created that one. Any help? --TwoWings (talk) 21:13, 9 March 2020 (UTC)
There don't seem to be any anti-bot provisions on the Mathematics Genealogy Project / North Dakota State University site, however. Ghouston (talk) 22:57, 8 March 2020 (UTC)
Such a property could be useful when you know someone's zodiac sign but not their exact date of birth. What do you others think?--Trade (talk) 22:00, 8 March 2020 (UTC)
It would make it infinitely easier to query people with specific zodiac signs. After all, many people are interested in astrology. --Trade (talk) 23:44, 8 March 2020 (UTC)
Yes. Ascribing completely bollocks information to items normally makes it easier to query for the completely bollocks data. Compelling argument. --Tagishsimon (talk) 02:02, 9 March 2020 (UTC)
We should be using reliable sources. I wonder if Trade can give an example of any person where there is a reliable source for the person's zodiac sign, but not for the person's date of birth. Jc3s5h (talk) 23:25, 8 March 2020 (UTC)
I tend to agree with this, although it does raise an interesting point, if we wanted this derivation to be represented in the data then the zodiac sign items themselves need to be associated with the date ranges somehow? It looks like currently they're not, see Libra (Q134394) for example. --SilentSpike (talk) 23:29, 8 March 2020 (UTC)
I think at the moment we don't have any capacity to include "single days" as dates - things like day in year for periodic occurrence (P837) have to use items, so Valentine's Day (Q37587) - day in year for periodic occurrence (P837) - February 14 (Q2332). So there isn't really a data model (or data type) suitable for representing this (though we could probably hack something together with qualifiers). Having said that, I'm on the side of "we probably don't want to have a property for this" - we can calculate it for the vast majority of people if we ever have a reason to, and there's only a small fraction where we can't calculate it, would want to know, and can find a reliable source saying what it is. Andrew Gray (talk) 21:59, 9 March 2020 (UTC)
In addition to the points made by Andrew Gray, there is no exact correspondence between zodiac signs and dates. It varies from year to year, and also depends on the time zone. The transitions do not occur at midnight, so there will always be days that are one sign in the earlier part of the day and the next sign later in the day. Jc3s5h (talk) 23:19, 9 March 2020 (UTC)
Didn't there used to be a page for property proposals where the suitable data model have yet to be created? --Trade (talk) 23:41, 9 March 2020 (UTC)
Request for feedback: Research on use of lists and queries for Wikidata maintenance work
Hello all,
A few weeks ago, the Wikidata UX team talked to nine people about how they improve and enrich data on Wikidata using queries and lists. The research shows that work practices, tools used and goals are very diverse. We noticed that communication and code examples are key in learning SPARQL and that people often worked on closing gaps in data and on finding problems in data structure and modeling.
Before moving on with the research, we would like to have your feedback on the existing results, to check if they adequately represent the goals, activities and problems when using queries and lists. All comments are very welcome, even if we won’t necessarily be able to cover all the possible variations of workflows in the final research.
To give feedback on the slides, feel free to comment directly in the document or to leave your comment below, indicating the slide number and/or quoting the text.
On top of that, if you are an experienced user of Petscan (for list building as well as statement creation based on Wikipedia Categories) we would love to hear more about how you use the tool. PetScan use was mentioned several times in the interviews but we could not cover the process in detail.
It could be interesting to see a short table with four columns:
what problem the user is trying to fix
what tool they are using
which query they use to find items
where that query is stored
PetScan used to be a great tool for many, until it broke sometimes last year. Technically, I think it didn't "break", but WMF cut its resource allocation, so it stopped working.
United Nations Security Council sanctions permanent identifier
The United Nations Security Council assigns a Permanent Reference Number to every individual and entity listed for one of their sanctions list. This is often useful for Wikipedia articles such as en:List of designated terrorist groups ([3]). These identifiers follow a fixed format where first two letters refer to a sanctions committee (e.g. QD for Al-Qaeda and Daesh), next letter is for indivuduals or entities (i or e) and then a unique number within the corresponding list. Each entry has a corresponding page at UN.org with a Narrative Summary ([4]) but, as far as I know, there are no direct URLs to get them from Permanent Reference Numbers. Do you think it would be reasonable to add a new identifier property? --MarioGom (talk) 09:04, 9 March 2020 (UTC)
Yes I think it would, identifiers do not need to have a formatter URL, it's just very common that they do and makes life a bit easier when they do. --SilentSpike (talk) 10:49, 9 March 2020 (UTC)
I would suggest that you contact them to ask whether they have or plan to have links to the narrative summaries based on the permanent reference numbers. I don't even see a way to get to the narrative summary from the consolidated list, and searching for the PRN does not work well. Bovlb (talk) 15:48, 10 March 2020 (UTC)
Creating data-tables to show in Wikipedia - eg. for Corona virus
Hi, I am completely new to WikiData, so I just want to understand what is possible and maybe get some links. Now during the Coronavirus outbreak there are many tables summarizing number of confirmed cases and deaths. These tables are both globally, per country and even per region. Wikipedia is a poor tool for this, as each cell needs to be maintained individually and there are no links or calculations. For example it would make sense to be able to sum the bottom row automatically, it would make sense to have calculated columns (eg. death ratio) and it would make sense to link one country total on the global page to the sum of regions on the country page. All in all, this is data. Can this be organized in WikiData and then linked into Wikipedia tables? Thanks! Osram (talk) 13:35, 10 March 2020 (UTC)
I think Commons works for tables that are stable. If data cells change one by one, that should work better here. Obviously, editing starts getting slow when there are many statements. --- Jura14:45, 10 March 2020 (UTC)
Surely Wikidata is not the right place to try to track every person who has contracted a particular illness. If someone finds wikibase the right tool for that, I would think they would need to set up a separate instance. - Jmabel (talk) 17:18, 10 March 2020 (UTC)
Help with splitting a conflated person item
Unless he completed his PhD at age 13, Christopher Turner (Q5113349) seems to conflate an author (DOB unknown, but likely before 1980) and a footballer (DOB 1987). I read over the help pages and I'm still not sure how to best resolve this - do I create only one new item or two? Alex Cohn (talk) 16:29, 10 March 2020 (UTC)
Twitter (X) username (P2002) is not a stable identifier, thus the data should be dated so we know when it was true. The constraint is a suggestion constraint, not a requirement. The reason point in time (P585) was chosen (over start time (P580) and end time (P582) which also date the data) is because of the complications around use of the qualifier number of subscribers (P3744) which can change rapidly in time unlike other qualifiers/the main value (which change at specific points in time). Now that I've been running a bot on this data for a bit, I tend to agree with Jura1 (talk • contribs • logs) that adding the followers as a qualifier is not a good solution because that data itself really needs to be qualified with P585 separate from the username statement. --SilentSpike (talk) 19:49, 10 March 2020 (UTC)
I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. SilentSpike (talk) 15:25, 16 March 2020 (UTC)
Sorry, I created this page by mistake. Please delete it. Thank you. – The preceding unsigned comment was added by霜奶仙 (talk • contribs) at 2020-02-27 10:58 (UTC).
I am not absolutely certain that I understand what you are doing to get this error. Can you please list each step in your procedure carefully? I would be happy to read a response in French if that would be easier for you. Bovlb (talk) 13:56, 11 March 2020 (UTC)
It seems that there was no input on the page itself. You could close it yourself as "malformatted". --- Jura15:51, 10 March 2020 (UTC)
@Jura1: like it says at the top, the discussion was on the talk page and then we adapted the document based on feedback from the 30 people who took part. John Cummings (talk) 20:55, 10 March 2020 (UTC)
s:Author:William Pinkerton contains one story and one magazine article, and there's a note on it, saying that nothing is known about this author and "It is not completely certain that the two pieces on this page are by the same person." Therefore, out of an abundance of caution, I created two items, one for the unknown author of each piece of writing. I've asked on the wikisource community portal whether they should be combined on a single author page, and consensus is that yes they should. That leaves me with a dilemma: what item do I link the author page to? Levana Taylor (talk) 14:50, 11 March 2020 (UTC)
That's a thought: "William Pinkerton, magazine contributor" linked to WS and cross-referenced to the two single-contribution items using "said to be the same as" (Curran Index is a reference, tentatively suggesting they are by the same person.) Makes it easy to merge later. Levana Taylor (talk) 16:46, 11 March 2020 (UTC)
What label should be entered when there are multiple correct spellings?
I'm asking because of Azadeh. In Farsi this is آزاده. In English it is often written as Azadeh, but Azade Shahmiri (whose first name is also آزاده) is absolutely Azade without h, see here: "First of all my first name is misspelled! Azade without 'h' at the end is correct. This is how my name is written on my ID, email and all the other official references. (I know it's more common for Farsi-speakers to put an h but it is what it is!)"
Azadeh is for people with that spelling in Latin script, e.g. US born persons. Ideally, you'd create an item for the name in Farsi and add there all possible transliterations. If you want you could also make one for "Azade" in Latin script and add that as well. --- Jura17:27, 11 March 2020 (UTC)
@Little pob: I have reverted my edit since I am not very much experienced in adding statements which has relations with preferred rank. I haven't dealt with it yet. Adithyak1997 (talk) 13:15, 16 March 2020 (UTC)
@Adithyak1997: No worries, it's not such a big deal in this instance, but generally in Wikidata you should never remove/replace a statement if it was valid at any given point in time. Instead using start/end time qualifiers and set preferred rank on the preferred value (can use a qualifier to indicate why it's preferred). I've gone ahead and done so with the formatter URL here, though note I've not added an end time to the old format since it's still valid. --SilentSpike (talk) 14:47, 16 March 2020 (UTC)
I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. SilentSpike (talk) 15:03, 17 March 2020 (UTC)
Policy for ID/schema change?
After some discussion, there is an issue that I remember cropped up before.
We have a property P with external IDs.
The IDs are used on site X, the primary source for these IDs.
Site X changes their ID schema, using new IDs.
Site X invalidates the old IDs, they can't be queried/linked to/seen any more.
Do we
Change the property P to use the new IDs, removing the old IDs
Change the property P to use the new IDs, keeping the old IDs as deprecated (note: QuickStatements can not currently mass-change statements to deprecated)
Create a new property for the same site X for the new IDs, keeping the old property P with the old IDs, maybe label/describe as "deprecated"
If so, can we fast-track the property creation, arguing that since property P is relevant, so is the new property
In general I would prefer #1 or #2 for the situations where the old IDs are of little or no practical use (ie they don't resolve to anything). Otherwise it's a bit confusing to end users if we have two different properties for the same ID, with different values. Andrew Gray (talk) 21:10, 11 March 2020 (UTC)
Question is what's the "practical use" here. I argue that the primary goal of identifiers is identification, and the identifiers (in Wikidata actually "formatter URL"+"identifier fragment") are just names for the things described, in the form of URLs. Following basic linked data principles, these names do not require that they resolve to anything—although in most cases they do, fortunately. When connecting Wikidata with other linked data services, matching items by identifiers is a pretty common process. It is useful to keep older names (identifiers) for things available in Wikidata, as this might be the only way matches can be found if the other linked data service has not been migrated to the new schema yet. From the practical point, I think we are limited by our approach how we technically implement the identifiers as "formatter URL"+"identifier fragment". If one of these components changes, one can take a practical approach and add the new value, either the formatter URL in the property page, or the identifier fragment on item pages, with preferred rank, keep the old one with normal rank. Old and new names (identifiers) would still be available, but the new ones that actually resolve to external resources would be much more visible to data users. However, once both formatter URL and the identifier fragment format change, I think we should really create a new property and keep the old one (both property and all statements). —MisterSynergy (talk) 22:40, 11 March 2020 (UTC)
Somehow I doubt two different formats for values can work well in the same property. Property constraints can't differentiate between rank when checking for a given format. Maybe at Commons it can do that? --- Jura09:11, 12 March 2020 (UTC)
Continued use of the same property when either the URL formatter or the identifier fragment format change is a pragmatic approach; I would not oppose if it was done more rigorously, i.e. with a new property as soon as there is any substantial change in the identifiers. However, I suppose that there will not be a consensus for such a rigorous approach. —MisterSynergy (talk) 09:29, 12 March 2020 (UTC)
I don't really see a consensus for repurposing the existing entity nor for repurposing Wikidata from a database into a webdirectory. How would MARC standards handle this? --- Jura09:54, 12 March 2020 (UTC)
Technically no, it's not CC0, which is a specific public domain grant and license. But it's likely to be public domain in the US anyway, combined with the author disclaiming rights, it seems reasonable to call it public domain. Ghouston (talk) 23:39, 11 March 2020 (UTC)
Well, the basic list of words would be public domain in the US, maybe not when combined with the descriptions. Ghouston (talk) 23:41, 11 March 2020 (UTC)
Ghouston the translation is the interesting part, I am using Whitaker's WORDS for the words themselves and their morpholigical data, which according to Wikilegal is not copyrightable.Uzielbot (talk) 08:02, 12 March 2020 (UTC)
Is it misleading adding coordinates to the items about the COVID-19 outbreak?
Hello. I have started a discussion with @Neo-Jay: at COVID-19 pandemic in Basque Country (Q87452683) about adding coordinates to items related to national/regional outbreaks of the COVID-19 disease. I understand the idea of adding coordinates: we can map things. But as the coordinate is placed in the geographical center of the country where the item belongs we have a point in a place that is not related to the actual outbreak point, and can be really misleading. I would understand adding a point for the COVID outbreak at Wuhan, but not at the geographical center of China, as this is not the main point I want to visualize in any external app (a map, a Wikipedia article...). I noticed this in the article eu:COVID-19_gaixotasunaren_2019-2020ko_pandemia_Euskal_Herrian, where we load the coordinates of events in the template automatically (for example, for an earthquake). As the discussion is not about COVID-19 pandemic in Basque Country (Q87452683) (where the point is in a mountain) but more general, I would like to know your opinion on this. I ping
If the outbreak occurred throughout the said country/region, it's the proper to use the geographic center of that country/region as the coordinate location. If the outbreak occurred only in part of the said country/region, the coordinate location may be changed to a more relevant place. If Theklan is discussing how to determine a best point as the coordinate location (Property:P625) for a COVID-19 outbreak item, that's fine. But it seems that Theklan opposes adding any coordinate location to a COVID-19 outbreak item. If so, I disagree. Coordinate location can be improved, but should not be removed. --Neo-Jay (talk) 11:17, 12 March 2020 (UTC)
Yes, I'm suggesting that the coordinate should give information about where something is happening, and the geographical center of a country is not accurate. -Theklan (talk) 11:45, 12 March 2020 (UTC)
It's a case-by-case issue. The geographic center can also be accurate under some circumstances. If you think that the geographic center is not appropriate for the outbreak in Basque Country (Q87452683), you may change it to a proper point, not simply remove it. As for the specific issue about Wuhan, I think that Wuhan is of course the coordinate location for the outbreak in Wuhan (Q85872051), but may not be the coordinate location for the outbreak in Hubei (Q87402404) and in mainland China (Q83872271). As the coronavirus outbreak occurred throughout every city in Hubei and every province in mainland China, it's proper, in my view, to use the geographic centers of Hubei and mainlan China as their respective coordinate locations (this is not about an earthquake epicenter). Using different coordinate locations for these three items can also make it easy to find the three items on the map in Wikidata query. --Neo-Jay (talk) 11:59, 12 March 2020 (UTC)
What I think is that using the geographic center of an entity for an event is not accurate. We don't use the coordinates of the city a person was born in the Wikidata item of the person, because we can get them easily using a query. If we want to build a map of countries where the outbreak is happening, we can query it easily having countries and then the coordinates of that country. But the event didn't happen in that precise point. -Theklan (talk) 12:41, 12 March 2020 (UTC)
Can coordinate location (P625) be used for an event item? Of course it can. Are you arguing that P625 should not be used for an event? I have put it very clearly, if you think that the geographic center is not appropriate for a specific item, you can change it, but please don't remove it. OK? Can you answer me: what do you think is the appropriate coordinate location for the outbreak in Basque Country (Q87452683)? Please feel free to add it. Please. And you cannot categorically claim that the geographic center should not be used for an event. Again, this is a case-by-case issue (why cannot the geographical center be used as the coordinate location for the outbreak in a small city, e.g., Macau or San Marino?). And we cannot find the the COVID-19 outbreak locations simply by country, because there are many non-country items. For example, the country (P17) of the outbreak in Hubei (Q87402404) is China, but we need to find Hubei as the location on map. And we also cannot find the outbreak in Europe on map by country (P17), because it has many values. --Neo-Jay (talk) 13:19, 12 March 2020 (UTC)
Do you mean that at least adding a coordinate for the geographic center to an item about COVID-19 outbreak in a relatively small area (e.g., as small as that in Normandy landings (Q16470), such as outbreaks in Macau (Q85872060) and in San Marino (Q87123561)) might make some sense? So the geographic center can be used as the coordinate location for at least some COVID-19 outbreaks? Is that correct? Frankly speaking, I don't know how to build a query to show the center point of the places where COVID-19 happened if not using coordinates (P625). For an item about the "outbreak in place X" (place X can be a continent, region, country, province, city, etc.), although the item's label is very clear to a human, how to let machine understand what place X is? It cannot be determined by "location" (P276), which may include place X's subdivisions. For example, how to build a query to get California for Q87455852 (COVID-19 outbreak in California)? Although P276 (location) of Q87455852 is California now, it may also have other values (e.g., counties in California). If coordinate location (P625) should not be used, can we use coordinates of northernmost/southernmost/easternmost/westernmost points (P1332/P1333/P1334/P1335) for outbreak items? It seems inappropriate to me. If you could teach me how to build a query to show the outbreak place that the item is about without adding coordinates, I would sincerely appreciate your help. --Neo-Jay (talk) 17:22, 12 March 2020 (UTC)
You use P17 (country) or P276 (location ) to locate an item. But as I put it, P17/P276 of item "outbreak in place X" may have many values other than place X. For example, your first query generates more than 100 locations for item Q81068910 (global outbreak of COVID-19), whose P17 (country) has many values. And P276 of Q87455852 (COVID-19 outbreak in California) may also have many values other than California (e.g., counties in California) and can generate many locations for this single item on map. I need to locate only one entity for an item, e.g., only Asia for Q86521237 (outbreak in Asia), and only California for Q87455852 (outbreak in California), not multiple countries for Q86521237 or multiple counties for Q87455852. Furthermore, your queries may make different items share a same point on map. For example, both Q87455852 (outbreak in California) and Q83873577 (outbreak in the US) have statement "country (P17): United States". What I need is to locate each item at a single and unique location on map so that we can layer the map by the item's other data, e.g., P1603 (number of cases) and P1120 (number of deaths). Unfortunately the two queries above cannot achieve it.--Neo-Jay (talk) 23:29, 12 March 2020 (UTC)
If you have data sufficient to enable you to add a P625, then you have data sufficient to choose a value or values for P276 of granularity appropriate to the locus of the outbreak - a county, or counties, in the US, rather than a state. --Tagishsimon (talk) 23:40, 12 March 2020 (UTC)
You are right. P625 (coordinate location) may also have more than one value. Although P625 has a single value constraint, its different values can be separated by separators. My query, which uses P625 to locate each item at a single and unique location on map (for now) and can layer the map by level of cases/deaths/CFR, is as follows:
Now such a mission seems impossible if we cannot locate item "outbreak in place X" solely at "place X" (not place X's country or subdivisions). Can we add "P518 (applies to part): place X" as a qualifier for item outbreak in place X's statement "instance of: disease outbreak (or its subclass, e.g., pandemic)"? I don't know whether it is appropriate and whether qualifier P518 can also have more than one value in this case. --Neo-Jay (talk) 00:14, 13 March 2020 (UTC)
There are some property proposals that are not being transcluded in Wikidata:Property_proposal/Authority_control - it seems it is related to the amount of transcludes. There are many properties marked ready there and some that are already handled. Should I move the completed ones to archive by hand and can someone create the properties marked ready? Iwan.Aucamp (talk) 07:30, 11 March 2020 (UTC)
Maybe I can move the properties that are created/done to the bottom of the page and move ones who are ready above them and under discussion to top? Iwan.Aucamp (talk) 08:15, 11 March 2020 (UTC)
Proposals should be marked as "ready" (which simply flags them for attention of a property creator or admin) by anyone, once there is either a clear consensus emerging, or discussion has died down. There is no sound reason to exclude the proponent from this. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits00:33, 14 March 2020 (UTC)
Yes my understanding is that taxonomic synonyms are being kept separate and we associate sources using different names with different items. --Jarekt (talk) 00:13, 13 March 2020 (UTC)
Since about 0415 UTC (a little after midnight US EDT) this morning, the dispatch lag from wikidata to client wiki's has been averaging about 5 minutes continuously. This has basically blocked all bot edits on Wikidata (all the ones that pay attention to maxlag). This is not the usual WDQS lag problem - for once the lag on the query servers has been fine (less than a minute except for a brief period about 2 hours ago) - probably because there have been almost no bot edits. We haven't had this sort of lag to client wikis for over a year - does anybody have any idea what could be causing it? It will become very hard to keep Wikidata up to date if bots are blocked like this for long. ArthurPSmith (talk) 14:29, 13 March 2020 (UTC)
According to this dashboard, dispatch lag fell sharply around 14:53.
I see the following bots heavily active during the busy period:
It appears that service is non-functional at least since November, as some batches I requested then are still not done. OTOH the translation to QS still is usable. Question: are these the only services that take a PMID or DOI and create a WD item from it? --SCIdude (talk) 07:20, 19 March 2020 (UTC)
I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. SCIdude (talk) 14:25, 19 March 2020 (UTC)
I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. SilentSpike (talk) 14:15, 19 March 2020 (UTC)
Is "former hospital" and intersection of "hospital building" and a "former building or structure"?
This seems wrong - a building that was once a hospital but is now a block of flats is a "former hospital", but it does not follow that it is a "former building or structure". It can change its purpose but still be in existence. Andrew Gray (talk) 21:15, 11 March 2020 (UTC)
Perhaps in some cases, but probably not all. There's not much (if any) external difference between modern apartment, office and hotel buildings, and they can be converted between such uses. A building also has an architectural style, which can be represented with architectural style (P149) instead of a class hierarchy. I know for example of a 19th century building, which was constructed as a bank, but the bank never occupied it and it was used as a chemist shop, then a bank about 10 years later, and is currently a cafe. It may have some stylistic characteristics that would indicate its original use, but perhaps it's confusing to be calling it a bank building now? There may also be buildings that were never used for their intended purpose. Ghouston (talk) 01:13, 12 March 2020 (UTC)
Bank building have generally a safe (Q471898), wich is probably present in the building. If we desire a more general purpose, we can also use office building (Q1021645) or commercial building (Q655686). All the usage of your exemple enter in the second one. But I agreed that older institutional buildings, like school, hospital or convent often look the same and are often interchangeable. If it could help, in my line of work, we divide buildings in 5 great famillies: Residential, agricultural, commercial, industrial and institutional building. At the difference of the other national museums of Ottawa Canada Science and Technology Museum (Q1163464) is a located in a former bakery. It's probably of good idea to put industrial building (Q12144897) in the item of the building even the curent use is a museum. --Fralambert (talk) 02:00, 12 March 2020 (UTC)
Possibly, although it's hardly unknown that barns or industrial buildings or schools are converted into residences, and it may still seem odd to describe a block of apartments as an industrial building. Museums can be housed in any random left-over structure, as well as new build, so having a museum "building type" wouldn't work well. Ghouston (talk) 02:02, 15 March 2020 (UTC)
PAWS
PAWS is a great tool, except for WikiData. When you develop a script, it waits 11 minutes due to database lag to only read an item. Then you get an error because you made a small error in your script. This makes the tool useless. Edoderoo (talk) 15:51, 12 March 2020 (UTC)
You can overwrite the standard pywikibot configuration in PAWS by creating a file "user-config.py" (file ending .py, not .ipynb!) in the main folder of your PAWS account. Maxlag can be deactivated by adding "maxlag=0"—this is okay for testing when developing a script, just do not run any production batches with this configuration. It might also be worth to add "put_throttle=1" permanently to that config file, as this increases your edit rate up to 1/sec. —MisterSynergy (talk) 16:07, 12 March 2020 (UTC)
The put throttle is acceptable for me (I have an Ubuntu box running pywikibot for production), but it is a good tip. I'll try out the max-lag parameter, as without this it becomes too annoying. Edoderoo (talk) 17:05, 12 March 2020 (UTC)
The other thing you might want to do is temporarily turn off your bot! :-)
Worth noting that when bots overwhelm the capacity of the RDF export by ignoring maxlag, wikidata - which is to say you and your fellow wikidata users - appear to suffer a disproportionate penalty seen, most days, in the Edit & Creation rate chart - a gap-toothed pattern the area beneath which is much smaller than that beneath the more rare within-capacity periods. --Tagishsimon (talk) 12:26, 14 March 2020 (UTC)
For your first point, we have Help:Label which is unambiguous "The label is the most common name that the item would be known by." So that would be X.509, not a mouthful of official title. Also known as is for all the other names you can come up with. --Tagishsimon (talk) 12:37, 14 March 2020 (UTC)
@Tagishsimon: From Help:Label "Because the aim is to use the name that an item would be known by to the most readers, labels should reflect common usage. When it comes to scientific names, for example, of a species, labels should use a species' common name, however items must always also have the scientific name listed as Alias."
To me this would rather suggest quite the opposite of what you suggest - or at least would favour "C++ standard" for ISO/IEC 14882 (Q66084444) , "Pulse code modulation (PCM) of voice frequencies" for Template:Q952898 and "Public-key and attribute certificate frameworks" for Template:Q1065865.
My own feeling about Wikidata labels is that if you are arguing about them, you are probably doing something wrong.
Arguments about the "right" names for things are for the Wikipedias, not Wikidata.
Labels here are for convenience, for people who don't remember that Q100 is Boston.
As long as the label is easily recognizable, it's fine, and no one needs to worry about whether some other label might be "better". (In my opinion.) —Scs (talk) 13:28, 14 March 2020 (UTC)
@Scs: There is already guidance on labels here which is quite specific on some matters and at least to one person very clear on this specific matter. The lack of norms will lead to edit wars and needless argument - it won't avoid argument. For example if someone goes and changes the label RFC 1323: TCP Extensions for High Performance (Q47263971) from "RFC 1323: TCP Extensions for High Performance" to "RFC 1323" and the bot changes it back. Why not just avoid this by putting some guidance down on the matter? Easily recognizable is a subjective attribute - to me "EN 301 347" and "ISO/IEC 14882" for example is not easily recognizable - but this is the preferred label in Tagishsimon's view.
Empirically - or, at least, anecdotally based on my experience - lack of norms does not, in fact, lead to edit wars and needless argument. My advice is that in most cases, there is a common name. It may well be C++ standard rather than ISO/IEC 14882, but it'll equally be Z39.50 rather than Information Retrieval (Z39.50), or Information Retrieval (Z39.50):Application Service Definition and Protocol Specification. And where there's not an obvious common name, then either a) be happy with a label and an Also Known As or several; or else b) curate a set of items such that the labels and aliases follow a consistent pattern. We cannot give much more useful guidance for a broad set of items, beyond the common name pattern, since for some items the spec ID, for some the formal title, for some, some other formulation, will be the common name. For a much narrower sets of items, it may be possible to provide more detailed guidance, and if so, typically, a project will discuss and recommend; or an individual will curate. As Scs notes, the specifics of labels, descriptions and aliases are of less importance in Wikidata; they exist mainly to enable the item to be discovered through search. Where it is important to record a definitive "label", title (P1476), and/or a specific ID, such as RfC ID (P892), is preferred. Having good labels, descriptions and aliases is important for discovery; but spending too much time worrying about which of several candidates gets the "label" slot and which gets the aka is probably not time well spent. --Tagishsimon (talk) 16:05, 14 March 2020 (UTC)
If you'd like full colour 3D models to be available on Wikimedia projects please subscribe to this phabricator task to show that it is a feature people want.
You're right on the first on; it's a subclass, not an instance. Arguably wrong on the second. From https://en.wikipedia.org/wiki/Internet_Standard "An Internet Standard is documented by[1] a Request for Comments (RFC) or a set of RFCs", and, "Engineering contributions to the IETF start as an Internet Draft, may be promoted to a Request for Comments, and may eventually become an Internet Standard." RFCs can, I think, legitimately be seen as that set of IAs that are still at the RFC stage. This may not be your preferred ontology; but that is beside the point. Wikidata is capable of rich descriptions and, explicitly, in Help:Statements#Multiple values: "If an item by nature can have properties with multiple values (like children of a person or official languages of a country), it is perfectly acceptable to add them." --Tagishsimon (talk) 21:24, 14 March 2020 (UTC)
The "Internet Official Protocol Standards" document, now as RFC 5000[RFC5000], has always been listed in the Internet Standard series as
STD 1. However, the document has not been kept up to date in recent
years, and it has fallen out of use in favor of the online list
produced by the RFC Editor [STDS-TRK].
This memo provides information for the Internet community. This memo
does not specify an Internet standard of any kind. Distribution of
this memo is unlimited.
In fact, each
RFC has a status, relative to its relation with the Internet
standardization process: Informational, Experimental, or Standards
Track (Proposed Standard, Draft Standard, Internet Standard), or
Historic. This status is reproduced on the first page of the RFC
itself, and is also documented in the periodic "Internet Official
Protocols Standards" RFC (STD 1).
Further, even if every single RFC was on IETF's Standards Track (which not every RFC is), then every RFC would still not be a internet standard (Q290378) (propper noun) - it may be a "internet standard" (regular noun) - and I don't object to such a classification (under said circumstances, which are not the case) - but that is not what is being discussed here. Iwan.Aucamp (talk) 21:53, 14 March 2020 (UTC)
In so far as magazine is a subclass of periodical, you can make that change without affecting the price of bread. I hope you develop the confidence to make changes that don't need discussion without bringing them here. As to the descriptions. Sigh. They're pithy strings which rarely encompass the whole scope of their item, for the obvious reason of their brevity. By all means tinker with them if it makes you happy, so long as by doing so you do not change their fundamental meaning. Magazines are a subclass of periodical; it's fine to describe a news magazine as a blah blah periodical blah blah. Users will pick up the Magazineiness from the label. --Tagishsimon (talk) 12:00, 15 March 2020 (UTC)
All "professional magazines" can be described as "periodical published by the governing body of a profession". Not all "periodical published by the governing body of a profession" are "professional magazines"; see, for instance, society journal (Q73364223). Probably better to work from the label to the description, rather than from the description to the label. --Tagishsimon (talk) 13:49, 15 March 2020 (UTC)
I welcome any effort attempting to create some order in this area. I've mainly been looking at Q2514037 and Property:P2813 and found the space to be rather messy/confusing as it stands. Don't have much input other than to say my personal preference would be to see us move towards the model currently promoted by Project Books, with a single P31 for all magazines and the many, many subclasses and variants moved to Property:P7937 and other relevant properties. Moebeus (talk) 13:18, 15 March 2020 (UTC)
Error in the Library of Congress database
Does anyone have a contact or know the procedure for alerting the Library of Congress of an error in their catalog? I've noticed that their listing linked from George Van Santvoord (Q61782159) includes both that author's works as well as those by George Van Santvoord (Q61782109). The authors have the same name but one was a Yale scholar of English literature and the other specialized in US law. The error has spread into VIAF and WorldCat. --EncycloPetey (talk) 04:49, 15 March 2020 (UTC)
I can see a half dozen entries for determination_method for why something is in the public_domain, what is the determination_method we use to show something is still under copyright? I do not see a list of methods. --RAN (talk) 18:07, 15 March 2020 (UTC)
Wouldn't that just be the law in the applicable jurisdiction? And if an item is missing for a country's copyright law it can be created. Ainali (talk) 22:16, 15 March 2020 (UTC)
line of duty death (Q87013982) I want to use it to tag police and firemen, just as we have killed in action for military people. Should it be added as a second line for "manner of death"? Eventually I want to see how many entries at Wikidata have entries in the officer_down website, to see if it is worth adding as a new data field. Please make additions to this q-entry. --RAN (talk) 00:29, 4 March 2020 (UTC)
In English it is not just murdered but died doing their job as a first responder, a fireman killed when the roof collapses, a ambulance driver hit by a car. --RAN (talk) 13:54, 5 March 2020 (UTC)
This is getting off-topic but that's wrong. Homicide is one person causing the death of the other, be it murder, self-defense, or unintentional. If a cop is chasing a suspect and that suspect shoots the cop dead, the suspect has committed homicide and the officer is considered killed in the line of duty. If a cop is getting out of a car and killed by a passing motorist, that is homicide, considered killed in the line of duty. Many English-language sources use this connotation. [6], [7], [8]. -Animalparty (talk) 21:39, 5 March 2020 (UTC)
If an on-duty police officer has a heart attack from running too fast after a suspect, he was "killed in the line of duty" but barring some really arcane legal theory, there was no homicide. Similarly, if a police officer or firefighter dies trying to save someone from a burning building, etc. - Jmabel (talk) 23:00, 5 March 2020 (UTC)
@Afaz: I agree with you to merge but you have to do something about the Wikipedia article which only appears to be about firefighters ... what to do? --RAN (talk) 18:04, 15 March 2020 (UTC)
I'd say to create out a second, more specialized item for that (I presume English-language) Wikipedia article. - Jmabel (talk) 00:01, 17 March 2020 (UTC)
Do something about Covid-19
Hi! Like many others I am currently exploring how the stuff that I know best could be used to help with a crisis like COVID-19 (Q84263196) and also similar situations in the future. One thing that came up in discussion with others would be to be able to visualize and push emergency notifications or important alerts from nearby relevant organizations, such as hospitals, counties, school districts, etc.
Making sure the data for certain domains - say all hospitals in a country have all required data, such as coordinates, websites, phone number - is complete.
Setting up a Wikiproject to discuss ideas and explore what we could do, where we could easily reach out to each other who are willing to help with such situations?
Option 2: improving health provider, hospital and emergency services structured data, most particularly official website, social media handles, coordinates. This can be expected to improve google infoboxes, amongst other things. Not sure offhand how we relate localities served to provider items, where that's relevant. --Tagishsimon (talk) 23:25, 13 March 2020 (UTC)
Sounds great and super actionable. We should be able to start with hospitals, and make sure they are well covered. If a quick check to the Web is reliable, the US has about 5000 hospitals, so the whole world might have up to an order of magnitude more. That sounds like a doable list. --Denny (talk) 00:05, 14 March 2020 (UTC)
In the past, I had a look at them once a year or so. This is slightly better than what I had seen at Wikipedia (some lists there include places closed for years). If you find a hospital that no longer exists, please add former hospital (Q64578911) to it. Given the lack of other updates, I'd rather see the official website added than a possibly incorrect or outdated phone number.
Key properties we have are number of hospital beds (P6801) and emergency services (P6855). These somehow mitigate that the definition of "hospital" varies now from one country to another. Coverage for Japan is probably quite good.
BTW, not sure how to add "if you have symptoms, don't walk into the next hospital and infect everybody else". Maybe building a list of relevant hotlines might help. In general, South Korea seems to have figured out how to go about it. --- Jura08:00, 14 March 2020 (UTC)
OpenStreetMap is licensed in another way as Wikidata. So I think it is not allowed to reconcile the data from OpenStreetMap with Wikidata. --Hogü-456 (talk) 18:03, 15 March 2020 (UTC)
Surely you also need to know the point in time as the administrative entity could not have existed at the same time as the entity? --SilentSpike (talk) 20:37, 15 March 2020 (UTC)
Agree with Ymblanter. Many many coordinates are wrong. Some are grossly wrong, that may be picked up by Paleim's tools. Some are slightly wrong, due to import from a rather low-precision coarsely-rounded source (sometimes via cebwiki), which then pushes them into the wrong administrative area. Auto-estmimation of P131s has to be heavily chacked, and is no substitute for extraction for authoritative databases or archival sources. Jheald (talk) 22:26, 15 March 2020 (UTC)
Scotland had very many wrong coordinates, certainly in the hundreds of, for (at the time) ~46k geocoded items. For many items, it was possible to acquire P131 data from a reliable source, meaning that we could visualise coordinate errors - see an example at https://twitter.com/generalising/status/1167880251336052736/photo/1 ... that would get lost if P131 had been ascribed automagically based on coord. Equally, I added many P131s based on coords, for item sets for which I've not yet found an RS. Such P131s have no ref. Some will contain garbage.
I wouldn't recommend indiscriminate addition of P131 based on P625, but think it's a fair thing to do for a dataset you intend to curate; e.g. for Scotland, it's possible to cross-ref P131s with en.wiki categories for the linked article, as one means of wringing out errors. As for technique, I used WDQS's 'search around point' geospatial search; you can see some evidence of that in https://twitter.com/Tagishsimon/status/1173019508233166848 such as the big East Lothian circle. --Tagishsimon (talk) 23:12, 15 March 2020 (UTC)
Hi Jheald and others. I agree that most coords are inaccurate, and automatically mass adding things like P131 based on coords is a bad idea. I am seeking if there is a way to do it for selected sets of articles - and that too maybe semi-automatically instead of fully automatic (we need to click confirm, etc). To give an example, I've recently went ahead and updated all coords in en:List of mountains of Sri Lanka. I now know all of those coords are accurate. Hence it would be nice if there was a way to fill (at least) P131 based on the coords of that set of articles (I'm sure there would be more properties that could be populated - country, timezone, etc). This is a small set, but suppose we have a bigger set, a tool that could do this would surely help. Rehman04:51, 16 March 2020 (UTC)
Notability of news articles
Today we create items for all journal articles out there, and try to link them together with each other (if they are used as a reference in another journal) and to their individual authors.
Currently, many journalists are notable enough to have Wikipedia articles about themselves, or at least a Wikidata item. Would it be inconceivable to create items for news articles as well, if either the author is notable, or it could be used as a reference in a Wikipedia article in some way (i.e. storing references as individual items on Wikidata).
Well, we wish to have items for all of a book author's works. We wish to have items for all of a journal author's works. We wish to have items for all music artist's works. Why not notable journlaists as well? Especially if those work are being used as references on Wikipedia. (t) Josve05a (c) 13:52, 15 March 2020 (UTC)
In my opinion if it is being used as a reference from Wikipedia then there is a structural need to have it in Wikidata so that it can be cited using Cite Q. Beyond this I would say you need to specifically identify what criteria from Wikidata:Notability you are satisfying and make some argument as to why it is satisfied. I would for example ask what serious publicly available sources can be used to describe it - Wikipedia would be one IMO, but if Wikipedia was not citing it then I would say there would have to be another. Iwan.Aucamp (talk) 15:48, 15 March 2020 (UTC)
Notability at Wikidata is to actually exist and have a link to show it exists. Our guideline states that it "[must] be described using serious and publicly available references". Saying "doesn't seem to be notable" has no meaning. We have a huge tranche of scientific articles that we imported. The argument for not including is always "now someone will import a billion news articles", well, no one wants to do that work, or they would have done it already. --RAN (talk) 18:00, 15 March 2020 (UTC)
One of these scholarly articles might contain a cure for the cancer you may get in ten years. With news that probabiity is much less. --SCIdude (talk) 15:29, 16 March 2020 (UTC)
This however made me wonder about the appropriate structure, what organizational items should there be, what should be an subclass of (P279) and what should be instance of (P31) etc.
For other cases, like IEEE it is a bit of a mess really (not that ITU-T is not a mess). Similar to ITU-T there is IEEE standard (Q55755785) and IEEE standard version (Q55936932) but there is almost no consistency in how they are used with the following statements occurring:
I'm working on a detailed proposal for a 3GPP publication ontology here if someone wants to have a look.
This raises a couple of questions:
Where is the right place to capture a recommended ontology for 3GPP publications and possibly other technical recommendations, reports, specifications and standards such as ISO and IEEE? Should it go under a WikiProject? If so which one? If not where?
If we were treating standards as documents with editions, then maybe a "technical standard" Q317623 is not really a form of written work equivalent to "novel" or "poem" or "technical report", since we want the editions to be the technical standards, not the "written work" item. Then these standards documents would be treated like other published technical reports, i.e., with a main item which is an instance of written work (Q47461344), or a subclass appropriate for technical reports, and the individual standards like C99 would be editions of that written work and also marked as "technical standards". The "technical standard" item would no longer be a subclass of "written work" or "document". Alternatively, maybe using editions would be a inappropriate, and each new version of the standard would be better treated as a written work in its own right. Ghouston (talk) 07:36, 16 March 2020 (UTC)
Cookbook recipe properties for main ingredients and optional ingredients
Porridge in the microwave (Q87066630) is one of the Wikidata items I'm working on. The Wikidata item was created for the Swedish Wikibooks Cookbook("Kokboken") recipe item "Kokboken/Recept/Gröt i mikrovågsugnen". I want to have "main ingredients" and "optional ingredients" properties. Because in this recipe there aren't any blueberries based on the original recipe, though personally I wanted to add blueberries, so I added them as optional ingredients. Is it ok if I create 2 properties for my needs or is it better if I go to a proposal page and suggest community input? What is the best way to go about this?(if this experience doesn't lead anywhere and we just keep on discussing and discussing I think I'll take the risk and create 2 properties for my needs, just saying as I'm experienced with discussions at some instances not leading to anything productive). If you can connect others who are interested in English Wikibooks Cookbook and Wikidata, you are welcome. It might be proper for me to do the same. Datariumrex (talk) 08:57, 16 March 2020 (UTC)
Unless you're a property creator you cannot create new properties (and if you were you would not bypass the approval process like that). As for your question, I'm personally unfamiliar with modelling recipes, but see that you're using has part(s) (P527). For optional ingredients you could qualify it with has characteristic (P1552)optional (Q59864995) (this item needs some details added, but seem to represent the right quality). That said, I personally think it would appropriate to propose a new "ingredient" property because a recipe itself is not made of ingredient parts (the relation is different than P527 in my opinion). --SilentSpike (talk) 10:03, 16 March 2020 (UTC)
Wikidata weekly summary #407
Here's your quick overview of what has been happening around Wikidata over the last week.
Note: Due to the ongoing COVID-19 situation, most meetups, including all those funded by Wikimedia Foundation grants, have been cancelled, or moved online, for the foreseeable future.
All identifiers are now sorted mostly alphabetically according to the RfC regarding the sorting of identifiers, which remains open if you have improvement proposals. Feel free to comment here!
Maximilian Klein applied for a project grant to merge and improve WHGI and Denelezh, tools that heavily rely on Wikidata to provide statistics about gender gap and biographical content in Wikimedia projects.
QuickStatements change (4 March). QuickStatements is now executing "run in background" batches with the same priority as direct batches run from the browser. Background batches may now run many times faster than they previously did (discussion), when the WDQS updater can handle this.
Where should the image of an award received go? At Chester Bennett Bowen (Q5093459) the medal of honor is displayed for image=, I don't mind having an image for an award, but we need image=null to search for people that need portraits. Any suggestions where it can go, or just delete the image? --RAN (talk) 19:16, 16 March 2020 (UTC)
Ways to mark possibly fictitious genealogy entries
Hi. Can anyone offer some advice on the best way to handle entries from genealogy databases that are believed to be fictitious? Sigrid of Halland (Q75437282) is recorded in The Peerage and WikiTree as the mother of Ulf the Earl (Q1710670) and previous editors have replicated the connection on Wikidata. Oleryhlolsson has removed the connection with the claim that the relationship is not supported by more reliable sources.[10] This is possibly a case of a speculative entry being created in one dataset on the internet and then migrating around the internet to other datasets.
We can't treat the connection as fact but what is the best way to handle the situation? If we delete the item, the disputed datasets will remain on the internet and a future editor will eventually recreate the same entry. If we blank the connection (as is the current position) another editor will re-establish the relationship in error through replicating the linked databases (this could also be done by bot edits copying data from the external databases with minimal oversight). If we record the relationship in our database then we are perpetuating and spreading the disputed record.
My view is that we need to retain the item but have some way to record the disputed nature of the claim. Are there any suggestions of alternate ways to handle this? From Hill To Shore (talk) 22:57, 16 March 2020 (UTC)
Anyone know why Hazard-Bot has stopped archiving requests on that page since the 26th? (I'd ask Hazard-SJ myself, but it is not clear that that user edits on-wiki anymore.) Mahir256 (talk) 05:00, 3 March 2020 (UTC)
I'm not sure anyone would know but the owner. Maybe we should look for a replacement? Have you tried reaching out to him? -- Ajraddatz (talk) 15:41, 7 March 2020 (UTC)
I now have a script that does the archiving, but I need to start it manually once or twice a day or so. I would rather not take on this task permanently, however. —MisterSynergy (talk) 06:56, 11 March 2020 (UTC)
Hmm, reviewing my bot's audit trail, it might be the MaxLag that is causing the scripts to fail.
Logs
118 thread(s) found on [[wikidata:User talk:Legoktm]]
Looking for: {{User:Hazard-Bot/Archiver}} in [[wikidata:User talk:Legoktm]]
Pausing due to database lag: Waiting for all: 5.15 seconds lagged.
Sleeping for 5.0 seconds, 2020-03-17 15:24:33
Pausing due to database lag: Waiting for all: 5.15 seconds lagged.
Sleeping for 5.0 seconds, 2020-03-17 15:24:38
Pausing due to database lag: Waiting for all: 5.15 seconds lagged.
Sleeping for 5.0 seconds, 2020-03-17 15:24:43
Pausing due to database lag: Waiting for all: 5.15 seconds lagged.
Sleeping for 5.0 seconds, 2020-03-17 15:24:48
Pausing due to database lag: Waiting for all: 5.15 seconds lagged.
Sleeping for 5.1 seconds, 2020-03-17 15:24:53
Pausing due to database lag: Waiting for all: 5.15 seconds lagged.
Sleeping for 6.2 seconds, 2020-03-17 15:24:58
Pausing due to database lag: Waiting for all: 6.2 seconds lagged.
Sleeping for 8.7 seconds, 2020-03-17 15:25:05
Pausing due to database lag: Waiting for all: 6.2 seconds lagged.
Sleeping for 9.9 seconds, 2020-03-17 15:25:14
Pausing due to database lag: Waiting for all: 6.2 seconds lagged.
Sleeping for 11.2 seconds, 2020-03-17 15:25:24
Pausing due to database lag: Waiting for all: 6.2 seconds lagged.
Sleeping for 12.4 seconds, 2020-03-17 15:25:35
Pausing due to database lag: Waiting for all: 6.2 seconds lagged.
Sleeping for 13.6 seconds, 2020-03-17 15:25:47
Found 1 wikidata:wikidata processes running, including this one.
Pausing due to database lag: Waiting for all: 6.3833333333333 seconds lagged.
Sleeping for 15.3 seconds, 2020-03-17 15:26:01
Pausing due to database lag: Waiting for all: 6.3833333333333 seconds lagged.
Sleeping for 16.6 seconds, 2020-03-17 15:26:17
Pausing due to database lag: Waiting for all: 6.3833333333333 seconds lagged.
Sleeping for 17.9 seconds, 2020-03-17 15:26:33
Pausing due to database lag: Waiting for all: 6.3833333333333 seconds lagged.
Sleeping for 19.1 seconds, 2020-03-17 15:26:51
ERROR: Error occurred while processing page [[wikidata:User talk:Legoktm]]
ERROR: MaxlagTimeoutError: Maximum retries attempted due to maxlag without success.
Consider the item Sanyathara (ID: Q16832086). She is an actress. Should this item be placed under actor, actor(actress) or actress? Adithyak1997 (talk) 17:17, 15 March 2020 (UTC)
I would place it under actor and I think that this is a good question and there is a solution needed. female form of label (P2521) is usually used in cases like that at the item about the occupation (P106) or position held to show the female label of the job. The current situation is not reader friendly. How the data users who download the dump or use data from a query work with this is not clear. I think it were good if there could be another label displayed as the main label at the Statements of the Template:P106 property if the gender of a person is female. Maybe also if the gender is divers. There I dont know how the property for that case is called. For queries it is from my point of view as a user with not big knowledge about SPARQL better to have one item for all genders. I dont know how I can write a query to get a list with all persons from a country with actor or actress as their occupation if there is a item for every gender. If there is a solution for that it were also good for me and then the other thing is not needed. --Hogü-456 (talk) 18:23, 15 March 2020 (UTC)
If you open this Wikidata item actor (Q33999) you will find "actress" under "Also known as". I think a query can have custom labels where you can input "actress" into the label and that way the query will display actresses the way you prefer. If you want a general change of the label, I suggest a consensus with maybe the Wikipedia article community or other related communities in relation to the English language Wikimedia projects. Datariumrex (talk) 09:16, 16 March 2020 (UTC)
There is not any such "property". I think you are confused. There is an ITEM actor (Q33999) which has an also known value of actress. There is no ITEM for "actress" apart from that one. There is no ITEM "actor(actress)". The actor (Q33999) ITEM is typically used with PROPERTY occupation (P106). To give you any more help, should it be needed, please provide at least URLs for the things you are trying to explain to us. thx --Tagishsimon (talk) 13:53, 16 March 2020 (UTC)
The best method, according to previous discussion, is to link both VIAF ids, and ignore the constraint flag. Apparently this helps people at VIAF notice the problem and perhaps they'll fix it. If you want to try to take it up with them directly, email bibchange@oclc.org. Ghouston (talk) 23:24, 16 March 2020 (UTC)
Our old contact at VIAF is no longer making corrections, does anyone know someone there that we can show how our system works, so they can start up making corrections again? --RAN (talk) 01:34, 18 March 2020 (UTC)
Case insensitive queries
How can I make a wuery case insensitive? E.g. the capitatlisation found in DOIs:
@Evolution and evolvability: There is no way to avoid using a (slow) FILTER if you are trying to do a case-insensitive string search in SPARQL. SPARQL literals are always case sensitive. If you are just searching for specific strings, it might be easier to use Wikidata's search than SPARQL (all string values are indexed in a case-insensitive fashion by the search engine here) - just use the search box on any Wikidata page. ArthurPSmith (talk) 18:12, 17 March 2020 (UTC)
While seeing geographical data on Wikidata connected to OSM a question arose that I am not able to answer:
What data is best kept only on OMS and what OSM data do we want a Wikidata item for? Where is the border/limit between these two? How much overlap should there be between Wikidata and OSM? – The preceding unsigned comment was added byDaanvr (talk • contribs).
I think instead of trying to find the exact universal line in the sand, it's better to just consider each individual case against the inclusion criteria of each project. As you're asking this on Wikidata, that just means following WD:N. --SilentSpike (talk) 15:55, 17 March 2020 (UTC)
@Daanvr:You're asking a very broad questions. Maybe you can be a bit more specific with some examples?
In the Netherlands I think most municipalities and towns are pretty well connected, but we could even go one level lower and connect neighborhoods. At first that might not look so interesting, but if you combine that with the data from https://www.cbs.nl/nl-nl/reeksen/kerncijfers-wijken-en-buurten-2004-2019 you can make all sorts of nice visualizations on the fly. The geodata would be on OSM, the statistics in CBS and Wikidata acts as the glue.
I'm not an expert on what kinds of data are in OSM, or on what is (or ought to be) linked in Wikidata, but for my purposes I would say:
Islands, lakes, rivers, mountains: all of them (whether we currenty have entities or not)
Political / administrative divisions and subdivisions: all of them, down to individual cities, neighborhoods, and other settlements (whether we currenty have entities or not)
Streets and railroad lines: no
Buildings, bridges, other structures: no (unless they're particularly notable ones we already have entities for)
Land cover: no
Elevation and hydrography data: no
Actual outline data for the entities in #1 and #2: yes, but only at the coarsest scale
I'm not sure if it's germane here, but in many cities, neighborhoods are a bit vague, and a definition given at OSM may or may not be objective/sourced/verifiable/etc. - Jmabel (talk) 00:08, 18 March 2020 (UTC)
On Q4663975, I cannot save, I think that's because I earlier removed interwikis from wikicode on w:pl:WP:Łasice. The message shows "Could not save due to an error. The save has failed." --Keyacom (💬 | 🖊)11:00, 12 March 2020 (UTC)
Problem resolved: On Wikidata, this error appears when just one of the listed pages is a redirect page. --Keyacom (💬 | 🖊)11:07, 12 March 2020 (UTC)
Well, I'm trying to add[11] and it won't let me. Just now I can't even remove it from Q87470714. Both tell me my changes can't be saved. Usually magic elves come along and Wikidata poofs into existence on the pages I create on other projects, so I don't come over here too often; this has been straight forward in the past though. I'll try again after the workday and maybe try bugging those other project pages then. Thanks! -- Kendrick7 (talk) 12:29, 13 March 2020 (UTC)
Both are correct:
One page can only be connected to a single item at a time (integrity check).
It doesn't say you cannot:
Warning: The action you are about to take will remove a sitelink from this item. Sitelinks should only be removed if the page in question has been deleted, or if that link is being moved into another item. If you are trying to do neither of these, please do not submit this edit again.
Huh. That link says: "Sitelinks can be moved manually from one item to another by first removing the link(s) from the obsolete item(s) (i.e. the item(s) you are merging from), and then adding it to the recipient item" which sounds an awful lot like what I've been trying to do.... -- Kendrick7 (talk) 22:32, 13 March 2020 (UTC)
Either the property is useless because there are dozens of IDs for the same thing, or something about the University's interface is producing multiple false values. I mention it here in case similar issues with this property are appearing on other data items. --EncycloPetey (talk) 05:01, 17 March 2020 (UTC)
We may have to just block the bot. It's made the same mistakes again see and the last few messages I've posted to Magnus Manske about his bot have been met with replies that the bot is working fine and he will fix nothing. --EncycloPetey (talk) 16:02, 17 March 2020 (UTC)
Looking into it more, the bot may be functioning correctly. It is taking information entered into the "mix and match" tool and copying it to Wikidata. The problem lies in how to prevent the wrong data from appearing in the tool. I haven't encountered the tool before but I will investigate. From Hill To Shore (talk) 16:59, 17 March 2020 (UTC)
@EncycloPetey: I've manually updated all of the entries I could find in the mix'n'match tool, so those ones won't be restored. The bot will restore any that I missed though; don't remove them until you have updated their entry at mix'n'match or the bot will just revert you. The mix'n'match URL is in the bot's edit summary. If you are not sure what to do, I'll make the updates when I have time. I'll keep the page in my watchlist. From Hill To Shore (talk) 22:05, 17 March 2020 (UTC)
The approval discussion said the property would be limited to persons. That should not be changed without discussion, but since the values are not limited to persons, it may call into question the original approval. --EncycloPetey (talk) 23:58, 17 March 2020 (UTC)
The bot continues to add multiple values and I've now seen bad values being added to other data items and for other properties. On data items where here I have checked, there have been more bad values added than good values. --EncycloPetey (talk) 14:56, 18 March 2020 (UTC)
Is there an automated way to reorder a list in Wikidata
Hi! I was just going through the low-numbered items out of curiosity, when I found that Q6 does not exist. I looked at its public logs, and it apparently never did. Why was this number skipped? Was there a technical reason, a glitch, something lost when the servers changed at some point, etc.? DemonDays64 | Talk to me13:49, 20 March 2020 (UTC) (please ping on reply)
Some Q numbers were reserved. Those that were created can be seen in the earliest contributions of Special:Contributions/127.0.0.1, but others were not used, such as Q6. Higher numbers that were never created, such as Q26536624, were unsuccessful attempts to create an item, for example if name and description matched an existing item. I don't know about the properties, if any numbers were reserved there. Peter James (talk) 17:47, 20 March 2020 (UTC)
It wasn't needed, Wikidata:Humour explains (although not all in that list were reserved - when Q12321 and Q12345 were created it was possible to look at the newest items and create a new item that would probably have the next number). Peter James (talk) 19:33, 20 March 2020 (UTC)
I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. SilentSpike (talk) 18:32, 24 March 2020 (UTC)
Author= for anonymous news articles
For anonymous news articles Author=anonymous or Author=unknown_value or Author=no_value or Author=unknown or just not add an author property? --RAN (talk) 23:11, 10 March 2020 (UTC)
Definitely "unknown" (if it's actually unknown--if it's anonymous but we do know the author, include it). Anonymous authors are authors, so it's incorrect to say that there is no value, it's just a value that we don't know. —Justin (koavf)❤T☮C☺M☯23:19, 10 March 2020 (UTC)
Please don't use "unknown". Writing is an art, even if done anonymously. For works by unknown artists, we use "anonymous". The problem with "unknown" on Wikidata is that it translates to "no value", so for example it could have been a chimpanzee. We know it was written by a human. Jane023 (talk) 08:14, 11 March 2020 (UTC)
@Jane023: No, it does not translate to « no value », which is a thing on Wikidata. Maybe some people interpret it that way, this is a mistake and they should be corrected and not maintained into a false belief. One big problem, I think, is to translate its own cultural judgement into Wikidata … translating « unknown » to « anonymous » is definitely a possibility at the data consumption step, for example in a query code or an infobox code. This does not mean that there is a need to use a special value « anonymous ». The fact that an author is usually human is represented in Wikidata … on the « author » property page, as a constraint (the domain constraint one). So « no », logically, this does not imply in Wikidata that the author can be a monkey. It implies in the Wikidata language that it’s probably a human. author TomT0m / talk page10:18, 11 March 2020 (UTC)
Thus my "Please", for all the reasons you gave. The word "anonymous" is often used for "unknown" as well as "known but desires anonymity". The convention in art historical terms is to use "anonymous". Jane023 (talk) 10:25, 11 March 2020 (UTC)
@Jane023: Wikidata does not have to use the convention, it just has to represent the fact that it has been published anonymously. We may need a qualifier or something like that, to be defined. But using a « special value item » is probably not the way to go. Imagine a query of all the text of different authors, for example, « anonymous » ( and also « some value » / « no value » but this is irreducible) would need special treatments for « anonymous » not being the most prolific author of all times … This is probably not the way to go. We need another representation and we can do better, maybe a qualifier or a property « publication mode » (values « anonymous », « pseudonymous ») or something like that, to denote voluntarily anonymous publication and this other kind of specificity.
This would have the advantage that if an anonymous author is later revealed and published, we can add the value to the statement without losing the fact that it was published anonymously. author TomT0m / talk page11:01, 11 March 2020 (UTC)
To be sure, my positition is use unknown valueHelp with some additional informations, probably in forms of a property that does not exist yet. Not « anonymous » as a value. author TomT0m / talk page11:26, 11 March 2020 (UTC)
The existing item anonymous (Q4233718) violates our ontology and as such shouldn't be used. Using it means that peoples will be told that two items are by the same author when that's not true. Author=unknown_value is the correct way to note the relationship. In case you want to add more information object has role (P3831) can be used to add them. ChristianKl ❪✉❫ 19:19, 11 March 2020 (UTC)
Using an actual "anonymous" item instead of the special <unknown value> (with good qualifiers) is not really ontologically sound and would lead to data users having to create an unnecessary special case for this "anonymous" item or else that would become a really prolific nobody in data queries. —seav (talk) 09:57, 14 March 2020 (UTC)
"Author=unknown_value" is the best way to go. "Author=no_value" would mean an article without author (maybe if it's auto-generated? maybe a chimpanzee, but some may argue that animals do have author rights so not sure it fits). Author=anonymous (I'm assuming here "anonymous" is the item anonymous (Q4233718)) is not wrong per see but very tricky, it could lead to several problem, like believing a same person wrote text for millennia :/ (obviously, there is an exception for edge cases, like when the person is known but not their name, Master of the Saint Bartholomew Altarpiece (Q632385) for instance). Cheers, VIGNERON (talk) 17:59, 14 March 2020 (UTC)
Oh no, we're having this discussion again! :-) I'm actually quite happy this is put forward again because it has been bothering me for a while. Wikidata is a wiki and a wiki works by consensus. Our current consensus for art is documented at Wikidata:WikiProject Visual arts/Item structure and is to use anonymous (Q4233718). Another aspect of a wiki is to try things, change things and improve things. As mentioned above, the current way of using anonymous (Q4233718) makes it harder to work with our data. It's a like an extra special value that is not easy to pinpoint. I think the source of the problem is what exactly some value means. The interface makes it look like "unknown value", but it's much broader than that. It's for everything where we know we have some value (as opposed to no value), but we don't know what that value is in our ontology or we know we don't have an item for that in our ontology and everything in between. Do we agree on this part?
I would love to switch to some value (yes, I prefer using that over unknown value) for art and particularly paintings. Yes, it will cost us some time to do conversion, but that has never kept us from doing much larger projects. In the current situation we use creator (P170) -> anonymous (Q4233718) and it might have some qualifiers. In the new situation we would use creator (P170) -> some value with qualifier object has role (P3831) -> anonymous (Q4233718) and whatever qualifiers we already have (and maybe object stated in reference as (P5997) in the reference section).
I hope we can reach consensus here. I'll advertise this discussion in several places so people have the possibility to comment. Multichill (talk) 19:26, 17 March 2020 (UTC)
This impacts the rules for copyright in Wikidata as well as Commons. We have anonymous in the legal sense as 'first publication without the name of the author published' (70 years after publication rule for pseudonymous and anonymous works), and we have works were we are not sure of who the author is, as well we have works like old paintings without the name of the creator, but 'problably painted by' or 'school of', or 'master of'. For mass imports to convert all this kind of 'anonymous' to all kind of unknown values, this should be handled by special bots as we can't ask insitutions to convert their logic where those are identitities in their system to our logic. --Hannolans (talk) 20:08, 17 March 2020 (UTC)
I am really very sorry to see that discussion take a wrong turn. If you look at the historiography of art history and attributions, you see attributions bounce around between non-names, anonymous, and various creators. The word "anonymous" is a standard, no matter what form of art it is. There is a very small group of works that are signed "anonymous" when the creator is in fact common knowledge, but to be clear, that is not the meaning that is currently the subject of the conversation. My opinion remains that using Marie Antoinette (Q47365) is the correct item to use in creator (P170) statements. Jane023 (talk) 21:01, 17 March 2020 (UTC)
Hello. Imho, the differents approachs are not necessary a problem, because it depends on the area of contribution. I must say on this question I even don't agree with myself. There are different and contradictory approaches:
ontologic approach
We don't know the creator, there is a way in Wikidata for this, so claim with "unknow" value. Discussion closed.
domain-based approach
Anonymous sounds absurd for ancient objects but for artworks it's very common.
approach according to art history and knowledge bases
In art history and in databases, especially in collection website, there is very often anonymous. It is even generally the case for modern visual artworks collection when creator is unknown. If Wikidata has vocation to base claims on knowledge database and references sourced, and to link knowledge data, so we maybe should preserve this approach. Jane023 is right to aware us on this point.
approach according to the historical notion of creator in visual art
It's a new version of the Quarrel of the Ancients and the Moderns! Not signing an atwork is the practice before Renaissance and after could be a choice, especially in painting. In the first case, the creator could be unknown and in the second the creator is anonymous. Both are true. That's why imho we have different approachs, depending on the area in which we contribute. The idea to associate a person (or several) to an artwork is now systematic for modern artworks and "anonymous" become the standard way to express it when a name is missing.
So, in Wikidata in artworks domain, we see this two differences in contributions on the main lines:
Creator:Unknow value until Renaissance
Creator:Anonymous after Renaissance
I think we should keeping things like this and for myself that's the ways I contribute, "Unknow value" for antiques, except rare exceptions, and anonymous for modern visual artworks.
And in others domain, like litterature, things could be different.
You seem to have missed the part about not removing anonymous, but moving it to a qualifier. Or at least you didn't address this point in your response. Multichill (talk) 09:12, 18 March 2020 (UTC)
I saw it. It's factually the unknow approach, called some value, with a complex and subtle add for anonymous. Imho it will be hard to understand, edit and link. For example if someone want to extract easily data on an artwork set with the creator. Wikidata is made to be reused and unfortuntely for outside I doubt that such statements makes the journey. Best regards --Shonagon (talk) 09:44, 18 March 2020 (UTC)
OK for the record I would approve using Marie Antoinette for anonymous! Just kidding. As I see it, the problem is with how we expect queries to behave after implementation. We either expect Mr. Anonymous aka Marie Antoinette to have millions of works, or we expect to have millions of works by unique, nameless individuals. Neither is optimal, but since the historic literature (i.e. book indexes including entries for "anonymous") chooses anonymous, this is the easiest and least ambiguous way to do it. You can always put lots of qualifiers on any item of course, and I have considered making items for "Circle of Rembrandt" where I make it instance of not-name, but give it members (all pupils and colleagues of Leiden and/or Amsterdam). Jane023 (talk) 10:00, 18 March 2020 (UTC)
I would definitely support switching from anonymous to somevalue. That is really what we aim to express here, and Wikidata is more concerned about meaning that about words. I am actuually not sure about the use of anonymous (Q4233718) in qualifiers actually. Note that "anonymous" has at least two, essentially opposite meanings that may require two different items:
a work whose creator's name has not beed passed down to us, but was known at the time. Say a sculpture in a gothic cathedral.
I made a query fetching case numbers and death toll by country by date. But data is not up-to-date. Would you help updating the situations of your country? Best regards. – Kwj2772 (talk) 18:12, 14 March 2020 (UTC)
By the way, the local epidemic are a mess. To indicate a location, we have at least 3 properties used : country (P17) / location (P276) and jurisdictionSearch as a qualifier of instance of (P31). This does not make easy, for example, to write a graph of deaths by country.
For country (P17), for example, we also have some items (like the one for the whole Europe) with several countries values … author TomT0m / talk page16:36, 19 March 2020 (UTC)
@Eurohunter: I can't see this property in your recent contributions. Can you please give some more information about: what you tried to do; what you expected to happen; and what happened instead? Cheers, Bovlb (talk) 19:59, 18 March 2020 (UTC)
@Eurohunter: Let me see if I understand what you are saying. You want to give a stage name a start time (as in the time that the underlying person started using that stage name), and further qualify that claim with the age the person was (stated to be) at the time. Is that right? Bovlb (talk) 21:15, 18 March 2020 (UTC)
@Eurohunter: OK. I think it would be more natural to represent that on the person, but all it seems redundant when we know the date of birth of a person. To be honest, I am not inclined to reify a stage name at all, unless there is something interesting to say about it, apart from the underlying person. Bovlb (talk) 14:10, 19 March 2020 (UTC)
Mark an Item for check
Hello,
I am interested in Ambassadors and other public positions and I try to get an overview about them. The Label of the following item is not the full one and I want to change it. ambassador of Germany to Switzerland (Q22230846) is about the ambassodor of Germany in Switzerland. It also serves for Liechtenstein and this is included in the full title but not in the label of the item. In cases like that I am not able to change all labels in the different languages. Is it possible to mark items like that one, that other people can help me by translating the labels. I know that there are some filters and lists out of these filters. I think that after the number of items who hit a criteria is at some filters high it is then not easy possible to look them through. How should things like that be handled. -- Hogü-456 (talk) 21:14, 18 March 2020 (UTC)
Can't answer the question about labels. The change you're proposing is worrying. It only makes sense if all German ambassadors to Switzerland were at the same time German ambassadors to Lichtenstein, without exception. Are you sure this has always been the case, for all items on which we use the value? If not, then adding a new "German ambassador to S & L" would seem to be the way to go, rather than changing the meaning of the existing item. Changing meanings of items should always ring alarm bells; occasionally the change will better reflect the thing being modelled. More often it does unimagined damage. --Tagishsimon (talk) 23:28, 18 March 2020 (UTC)
I know that this is the case. There is since the begin of the diplomatic mission of the Federal Republik of Germany in Switzerland 1951 the same ambassador also serving for Liechtenstein. --Hogü-456 (talk) 16:42, 19 March 2020 (UTC)
Yes, but the German Wikipedia lists German ambassadors to Switzerland since the 1800s. A simple name change could also imply the 1800s ambassadors also covered Liechtenstein. From Hill To Shore (talk) 17:39, 19 March 2020 (UTC)
As you tell me I need to pay more attention in cases like that what happens throug my change and it is an example that it is not easy use the right item. There were other states before the Federal Repubublic of Germany and so the example uses at the moment a not correct item after Federal Republic of Germany started in 1949. For the periods before 1951 and the positions there I create another items. These are not the same positions.Is it right that there should be an own item for every position or when is a new item needed. For example if a minister gets further tasks and his title is changed through that. Is then a new item needed. It is not in every case clear what the official title is and usually the normally used title is not the full official title. Defining a correct label in different languages is not easy and if the number of positions describing mostly the same thing from usually used title increases then there should be a solution to find all these items. Is there a efficient way to search for aliasses. --Hogü-456 (talk) 20:17, 19 March 2020 (UTC)
It's not the first time I saying it but it's impossible to obtain lanuage code from list of headers (labels) so If you see certain language in one item you are unable to add the same language to other item because you don't know its code. The only way to get them is to "provoking" edit/make "empty" edit of lasbels so their codes appear in item history. Eurohunter (talk) 21:02, 18 March 2020 (UTC)
@Eurohunter: Do you see a tab between "Read" and "View history" that says "Labels list" (depending on your interface language, of course)? When you click on it, do you see a table with language codes at the left?
An alternative, but somewhat more fiddly way to get label language codes from an item is to use the JSON view. E.g. for Basshunter (Q383541), change the URL to https://www.wikidata.org/wiki/Special:EntityData/Q383541.json (does someone have a gadget to add a link for this?) and examine the "labels" section. (You may want to install a "JSON Viewer" extension on your browser.)
@Eurohunter: Hmm. Are you using the released version or the beta version? (I see the beta version has only language names.) It should look like this image. Bovlb (talk) 16:42, 20 March 2020 (UTC)
Hi. I love hiking trails and work in OSM mapping them. There exist a lot of data about the different stages, their length, connection to other trails, whether you can bring a wheelchair, whether there is accomodation in each end of the stage, e.g. Currently none of this data is modeled in WD it seems. See e.g. Kungsleden (Q59780).
This particular trail also has 5 sections [12] which are subdivided into sections e.g. [13].
Would it be acceptable to create a Qitem for every section and stage to model this data? WDYT?--So9q (talk) 13:05, 20 March 2020 (UTC)
Intentional sitelinks to redirects don't seem to work
Hi again! I just made Legion (Q88081101) and was trying to give it a sitelink to the redirect page wikipedia:Legion (novella) (which redirects to a page that has an item linked to it), but an error happens. Above the place where I typed in the sitelink, it says:
Could not save due to an error.
The save has failed.
This happens repeatedly, and with attempts to link different items to redirects. The very confusing thing is that you can mark a sitelink as an "intentional sitelink to redirect".
I think I remember that it's intentional to prevent redirects from being added, but could be wrong about that. I think you can only link to a redirect if you link to a non-redirect page and then change it into a redirect. In the mean time, I've linked the book item to the series item. -- Ajraddatz (talk) 15:13, 20 March 2020 (UTC)
@EncycloPetey, Magnus Manske, From Hill To Shore: I fully protected Q1339 for a week. Here's hoping that the issues with the addition of faulty identifiers can be addressed before that week ends. Any other admin may undo this full protection, but I insist that they first consider whether the aggregation of that many errors on a single item is appropriate to allow. Mahir256 (talk) 21:05, 18 March 2020 (UTC)
@EncycloPetey: Can you please link to the relevant conversation with Magnus Manske ?
It doesn't seem like fully-protecting an item is the right way to respond to a problem like this. Do we have any reason to suppose that the problem is limited to one item?
If this bot is just proxying edits for a Mix'n'match user who is logged in through OAuth, then why are the edits not credited to the user? Bovlb (talk) 21:08, 18 March 2020 (UTC)
(after edit conflict) From all the ones I fixed so far, they appear to have been auto-generated in mix'n'match in 2016. I don't think protecting the page will be productive here. The only way I have to find what needs fixing is through the bot importing the old and incorrect data. As soon as protection expires the bot edits will continue. However, if someone more knowledgeable on mix'n'match is able to produce a data dump of all entries pointing to Johann Sebastian Bach (Q1339) then we can work through resolving the errors before the protection expires. From Hill To Shore (talk) 21:23, 18 March 2020 (UTC)
I've now found how to see every entry linked to a Q at mix'n'match and am breaking the links from bad items where I find them. For some reason the tool has automatically matched a large number of entries simply because they have a first name of Johann (or similar). God knows the quality of data being imported across the project if this is the quality of matching. From Hill To Shore (talk) 21:54, 18 March 2020 (UTC)
@Bovlb: I don't know how to link to the sort of talk page threads used on his talk page, but the most recent conversation about this kind of problem is titled: "Bad LoC data link". And no, this isn't limited to just Johann Sebastian Bach (Q1339). Other recent instances of similar bad data being added are [21], [22], but for these other instances, only a single bad data value was added. For Johann Sebastian Bach (Q1339) it is a flood of bad data, and hence we've now had two threads in Project chat trying to resolve the problem. --EncycloPetey (talk) 21:19, 18 March 2020 (UTC)
Indeed. Neither is the User identified in the edit comment, nor is the link active to the Mix'n'Match tool. So, I have to copy-paste the link to another window to access the edit request page, and then manually remove the action item. --EncycloPetey (talk) 21:40, 18 March 2020 (UTC)
I started a thread about requiring proxying bots to attribute edits to a specific user. My view is that either botops should be responsible for their bot's edits, or they should indicate who is responsible. The third path we're on here, where the botop won't take responsibility, but there is no-one else to blame, does not seem defensible. Bovlb (talk) 21:59, 18 March 2020 (UTC)
Good grief. Protecting an individual entity isn't just a bad/ought-to-be-unnecessary idea, it sounds like it's not even sufficient, if there are unknown numbers of other entities also being affected.
Seems to me we just have to temporarily stop the bot! I'm not blaming the bot or its owner, but until we can properly vet the input, the bot is doing more harm than good. —Scs (talk) 21:59, 18 March 2020 (UTC)
I think I have managed to clear out all of the bad entries for Johann Sebastian Bach at mix'n'match (assuming the search returned all of the linked entries), so the page protection can probably be lifted (the bot will probably still try to load in the remaining valid entries). However, that doesn't address the wider problem. From Hill To Shore (talk) 22:15, 18 March 2020 (UTC)
I found more in the most recent contributions. I have blocked the bot for now until Magnus can sort out what is going on. Bovlb (talk) 01:24, 19 March 2020 (UTC)
I was not aware of systematic errors (rather than occasional, manual false matches in MnM). I have altered the bot code to not add values to an existing property, based on a manually curated property whitelist. I will try to sort out the mismatches in MnM later. Please unblock Reinheitsgebot again. --Magnus Manske (talk) 13:43, 21 March 2020 (UTC)
I suggest a new property may be useful in relation to borders, one that is based on https://wiki.openstreetmap.org/wiki/Key:access and allows modelling of restrictions against freedom of movement (Q1344824). The property could perhaps be broad enough to allow modelling of access restrictions to cross borders, enter buildings, participate in events, etc. Qualifiers could include:
prerequisite (P4967) or new property to state which visa, permit or other documentation may be required.
applies to people (P6001) to state that the access restriction applies to a particular group of people (by profession, citizenship, etc).
Yes, I wish I could query the current closed borders and map their (red) lines... Timatic might help on this topic. @Twofortnights:, is there any dataset/database that could be useful to modelize (for instance US normally require Europeans an Esta, North Korean people cannot enter South Korea etc) Bouzinac (talk) 17:52, 21 March 2020 (UTC)
Hmm. I think it would be more logical to add the same value (e.g. "Macedonians") multiple times, with a single population and point in time qualifier in each. It makes no sense to have the qualifiers as "parallel arrays" as this is lost in the RDF. Bovlb (talk) 21:11, 21 March 2020 (UTC)
Yes, you need multiple of the same value with differing qualifiers. Then set the preferred rank on the most recent claims. --SilentSpike (talk) 22:37, 21 March 2020 (UTC)
Also, on a second look, I don't think P172 should be used as a main claim like this. It's intended to link the ethnic group of the subject (item the statement is on). Unsure how best to capture population ethnicity breakdown like is trying to be done here, but the main statement should almost certainly be P1082. [Wrote this in the time which above reply was posted, may be your best bet]. --SilentSpike (talk) 22:43, 21 March 2020 (UTC)
Hey, yes. It was caused by reworking the term store. Doing an edit for the item fixed it now. Let me know if the problem persists or it continues to happen in other places. Amir Sarabadani (WMDE) (talk) 14:39, 18 March 2020 (UTC)
From Wikidata:Project chat/Archive/2020/03#Date of baptism/religion. In that discussion, I suggested it was not a good idea to have the constraint asking for religion when adding a date of baptism. Is it going to be removed? I'm not the only one who expressed doubts about it, although the discussion is not resolved yet. I think this is a case of "when in doubt, leave it out" Levana Taylor (talk) 14:27, 20 March 2020 (UTC)
You are baptized into a specific religion, if you switch religions, then use the start and end date. If you know the baptism date, then you know the religion. --RAN (talk) 16:31, 20 March 2020 (UTC)
Not necessarily. I have seen sources that say a person was baptised on a particular date but don't mention a religion. Usually these are records from very long ago and you can infer the religion due to one being dominant in that particular state. For example, if someone from England was recorded by a source as baptised in 1553, the odds are very good that they were Christian (though whether Protestant or Catholic is less certain). While the religion can be inferred, you can't really put a reference against it. From Hill To Shore (talk) 18:23, 20 March 2020 (UTC)
This may be a European-American cultural difference! I have talked to some people in Germany who say "I'm a Lutheran, I suppose, anyhow that's what I was baptised and what's on my papers, not that I ever go to church." Whereas to me as an American it seems like religion is a matter of belief and you're misrepresenting someone by stating they have a certain religion if they don't actually practise it. So I would rather not run the risk of making a false claim about someone by putting in a "religion" statement, unless you actually know something about their adult beliefs. Again, I say, err on the safe side and omit the religion statement ... Levana Taylor (talk) 22:24, 20 March 2020 (UTC)
If you have evidence of baptism and the related religion then we should insert the appropriate religion. It is a correct statement that they were part of that religion when baptised. If you have evidence of different adult beliefs then set qualifiers for religion with a start time of baptism and end time of when your source says (use "latest date" if your source of alternative belief doesn't say the specific date of change). From Hill To Shore (talk) 22:37, 20 March 2020 (UTC)
I disagree with that. Almost always you are not going to know an end date or any positive information about their actual beliefs, leaving to be inferred that they kept the same religion all their life -- I just don't want to run the risk of misrepresenting someone like that. "Don't know" should always be clearly marked as "don't know." So how about this? Have a property "religion of childhood baptism." That says exactly no more than what it says. Levana Taylor (talk) 22:45, 20 March 2020 (UTC)
"Don't know" is definitely the wrong stance here. If we have a reliable source that states something at a point in time, then we do know. It would be misrepresenting them to say otherwise. Denying a sourced religious belief (even in childhood) is just as damaging as implying an unsourced religious belief in adulthood. A new property could work if it gets broader support. One issue you may have with your philosophy though is when do you draw the line? Hypothetically, we get the new property and say someone was baptised a catholic before their first birthday. We then get another source that says they were still going to church at the age of 10. In your view would it be okay to use the Religion parameter then, or is it also invalid as they are still a child? Unless you can clarify your position on the age when you are willing to accept someone's belief as sufficient evidence for usage on Wikidata, a new parameter is only going to be a patch on the problem. From Hill To Shore (talk) 23:00, 20 March 2020 (UTC)
Baptism shortly after birth may tell you something about someone's parents' religious affiliation or beliefs, but I don't see how it can tell you anything about their own. Do we also (for example) consider Mormon baptisms of the dead to retroactively make people Mormons, even if the religion didn't yet exist in their lifetime, or if they were (for example) Jews killed during the Holocaust? Obviously, an adult convert who gets baptised is normally an entirely different matter, but even there: what about mass baptisms by the priests who accompanied the Conquistadors? Did that really turn a bunch of American Natives into Catholics?
In a similar matter: as an atheist of Jewish ancestry, I am certainly a Jew by Halachic law, and I'll even help out my religious friends when they need an extra adult male Jew for a minyan, but that doesn't change my religious beliefs or affiliation. - Jmabel (talk) 23:09, 20 March 2020 (UTC)
I have no personal view on religious beliefs as I am approaching this from the impartial position that we should reflect the source material as best we can. If we have a fact from a reliable source, we should not be deleting it because it makes us uncomfortable. Instead of deleting, as a bare minimum we should be deprecating the statement with a qualifier that we need more sources to support it (possibly cannot be confirmed by other sources (Q25895909), possibly invalid entry requiring further references (Q35779580) or not been able to confirm this claim (Q21655367)). One thing I've noticed is that we have yet to link to the property in question; religion or worldview (P140). Now it may not be the case with every language, but several of the descriptions that I can interpret say this property includes those who are "associated" with the religion. The very act of being baptised establishes the association. If you want to convert the property to a meaning of "confirmed belief" or "confirmed affiliation," then that would support your argument. The change in property would need a broader discussion though. From Hill To Shore (talk) 23:30, 20 March 2020 (UTC)
To "reflect the source material as best we can" would be to state just that the source material exists without adding any additional assumptions or inferences. As was pointed out in the previous iteration of this discussion, sometimes parents are compelled to acquire documentation from their locally dominant religion even if they don't believe in it. Now, there's an argument to be had regarding religion as a collective social phenomenon, which we can talk about someone's participation in while avoiding inquiring into their internal thoughts, but that's a subtle argument that Wikidata editors should not have to think about every single time they enter a baptism date! No, there should just be a property (independent or as a qualifier) indicating what religion the document is for, end of matter. That is indisputably factual in anyone's eyes. Levana Taylor (talk) 07:31, 21 March 2020 (UTC)
A new parameter won't fix the problem unless you also gain consensus to reduce the scope of the existing parameter. So long as "association" is the threshold for adding religion or worldview (P140), it is, to use your words, indisputably factual that the baptism has established an association. Unless you are offering to patrol every single addition of religion or worldview (P140) in the database, a clarification/reduction in scope of the existing parameter is the only way to achieve your goal. From Hill To Shore (talk) 09:13, 21 March 2020 (UTC)
Further to the discussion above, to achieve the intended outcome, you will need to separate the threshold of "association" from religion or worldview (P140). Tinkering with the scope of an existing parameter is a bad idea as it changes the interpretation of every item it is currently linked to. Instead, I would advise gaining consensus to split the single parameter into separate parameters and then deprecating the original. You can have your "Religion of baptism" as one parameter but we would also need a "Practised religion" for where there is evidence of continued involvement. We could also have "Professed religious belief" if desired. We would certainly need a property for "Associated religion" but with a constraint that it can't be used on human items; that will allow items about buildings or artifacts to still be linked to the related religion. From Hill To Shore (talk) 09:33, 21 March 2020 (UTC)
Define religion in this context. Is it just christian, muslim, hindu, buddhist, pastafarian etc, or should it somehow differentiate between some fractions of those cults, like shiite vs. sunni, or catholic vs. protestant, or even further down the schisms, down to the en:The People's Front of Judea kind of infighting? Grüße vom Sänger ♫ (talk)10:20, 21 March 2020 (UTC)
I fail to see the list of allowed religions, and on what level of differentiation this should be put to the items. Where is it hidden? Grüße vom Sänger ♫ (talk)17:19, 21 March 2020 (UTC)
Um, let me refer back to the actual issue in question which is in danger of getting lost: The point was the constraint recently added to date of baptism (P1636) "An entity with date of baptism in early childhood should also have a statement religion." Now, since we have seen that the issue of stating a person's religion is complicated and contentious, why have this constraint, forcing people to confront the vexed question every time they enter a baptism date simply as proxy for birthdate? No no no. Levana Taylor (talk) 11:30, 21 March 2020 (UTC)
When we say that a person's occupation is "laywer" without specifying anything about when that's true, we are saying that there was at least some time in the life of the person where they were a laywer. We aren't saying that they were a lawyer with 10 years. There's no false statement if we have statement that someone's religion is X if it was at some time and their lives and not other times of their lives. ChristianKl ❪✉❫ 14:50, 21 March 2020 (UTC)
I think Levana Taylor has this exactly right. The religion should be a property of the baptism, not of the person. The person's religion is a separate matter. ChristianKl, as has been made clear by several examples she an I have brought forward, it is quite possible for someone to have been baptised into a religion that neither they nor their parents ever subscribed to. To underline with an obvious example: even if you were "free evangelical" in 19th-century Norway, you still had to have your children baptised in the State church. - Jmabel (talk) 17:31, 21 March 2020 (UTC)
Thanks to Jmabel for the examples. Those provide concrete instances in which, if the religion on a person's baptism record was simply added as a "religion" statement with no restriction, that would constitute introducing bad data into Wikidata. Levana Taylor (talk) 12:16, 22 March 2020 (UTC)
Given that the new constraint under discussion is considerably controversial (in the two iterations of this discussion I count three people definitely opposed), why hasn't it been removed yet? The proper order of actions with a disputed new addition is not to add it until it's been agreed on. Levana Taylor (talk) 12:16, 22 March 2020 (UTC)
Hi. QS is a valuable tool used by a lot of people daily to import large quantities of statements. Unfortunately it receives very little love from developers since the rewrite some time ago.
A quick glance at the list of issues and the lack of communication there tells me that Magnus has moved on to other ideas.
I would like to help with a fork but I'm not sure I can pull it of alone so I write this instead to ask if others are up for joining together.
Have you tried contributing improvements? That's usually an easier/friendlier first step than jumping straight to a fork. Also, not sure if you're aware, but Magnus has been working on a new Rust version, which does seem to have some much more recent development. --Oravrattas (talk) 09:54, 22 March 2020 (UTC)
Oh! I was not aware. No I have not tried to contribute patches, good idea, I will take a look at the rust version.--So9q (talk) 12:27, 22 March 2020 (UTC)
I have looked a bit at other first order metaclasses and it is hard to say if what is happening this is correct. I have also looked for guidelines or explanation of how to use metaclasses and all I can find is Wikidata:WikiProject_Ontology/Modelling.
Specific cases where I would say metaclasses are being used more correctly (don't take this as an endorsement of all statements):
If I had to sum up what my main problem is with cell (Q7868), cell type (Q189118) and anatomical structure (Q4936952) is that we have items which end up both subclasses and instances of the same item - which I would say should not happen - or at least in this case does not make a lot of sense to me.
I am not entirely sure about all of this and I don't want to go around breaking things. I would appreciate other opinions on the matter. If we could declare that some specific use of metaclasses are correct and maybe give some examples of incorrect use and why it is incorrect it would be great.
I agree that we currently do't have a consistent way to model the ontology of anatomy and it would be great to have a consistent model. I'm not sure whether solving the problem one-item-at-a-time is the best way to go about it. I think we need more fundamental agreement.
We currently have no policy page that defines what we mean with second-order class (Q24017414). To have a clear and consistent way the concept is used throughout Wikidata, it would be best to have an RfC that defines a clear usage for it.
In absence of a more global policy we could have a policy for anatomy, but I think it would also make sense to write this into a help page at project anatomy instead of solving the problem one-item-at-a-time.
Wikidata Lexeme Forms allows quickly generating a new lexeme with all its forms in selected languages; you can also use the tool to add forms to an existing lexeme, or bulk upload many lexemes and forms at once.
Hello, my name is Siddhartha Dukkipati and I am a PG student from IIIT Hyderabad, India, working on the development of Telugu Wikipedia.
I would like to import some of the data which I have into the wiki data and then retrieve it on a later purpose. – The preceding unsigned comment was added byDukkips (talk • contribs) at 11:09, 18 March 2020 (UTC).
It is a csv file wherein I have data of school records based on the district and mandal so on. I have up to 76 thousand records. – The preceding unsigned comment was added byDukkips (talk • contribs).
What is the data about? If it's about students at the schools, I doubt that would be useful beyond a straight-out count of students. If it's about the schools themselves, or if it is about faculty at the university level (but presumably not at the primary or secondary level) that would be useful. Or at least that is my take. Does someone else want to weigh in? - Jmabel (talk) 15:53, 20 March 2020 (UTC)
This data is about the school like the number of students studying that school, the number of teachers and classes which they offer and area of campus and which syllabus do they teach likewise I have these details with me. Now I would like to know, the structure if wikidata and how to import the data which I have with me and retrieve that information.
@Dukkips:, first of all, you should sign your messages here. Also, what you are talking about is complex. You need an expert to guide you. For the software solution, QuickStatement or OpenRefine would do the job, but those tool are not for new comers and people with little experience… That being said, I could help you, in part. My email is: antoine@beaubien.qc.ca. Regards, Antoine2711 (talk) 23:51, 25 March 2020 (UTC)
Indian cinema personalities
Hi,
I work in a film archive as a cataloger. During these years, I have to come across lot of names of Indian film personalities in my records. For convenience, I have been preparing a spreadsheet with all the names with their variations. Trust me, in India there are a lot of variations of names.
As a result, I am now thinking to develop somewhat a Name Authority project for Indian cinema personalities in Wiki data so that a unified common name index is available online for users.
Please guide me with the steps to achieve the same.
Thanks!
– The preceding unsigned comment was added byMurchana (talk • contribs) at 17:39, March 18, 2020 (UTC).
@Murchana: when you'll be ready to push your data, you should consider a tool to push in a batch your spreadsheet. I can help you if you are interested. Regards, Antoine2711 (talk) 00:35, 26 March 2020 (UTC)
The german term is "In der Schlacht getötet" vs. "Im Einsatz getötet", thus {{Q|Q18663901} is restricted to the war, while killed in action (Q210392) is for all occupations, as someone dies while doing their duty (Fire brigade, soldier, doctor... Grüße vom Sänger ♫ (talk)08:58, 22 March 2020 (UTC)
At the moment, though, most of the descriptions I see in most languages on killed in action (Q210392) seem to confine it to military situations as well, just not necessarily battle. I think we should either expand that to firefighters, police, etc. or have a distinct parallel item for those non-military occupations, if we do not already have one. - Jmabel (talk) 19:26, 22 March 2020 (UTC)
We have killed in the line of duty for first responders, killed in action is a military classification. Yes, KIA includes accidental deaths and deaths from disease while on duty. --RAN (talk) 04:12, 25 March 2020 (UTC)
How might I represent a song intended for a particular occasion? How would you represent that a song is for Christmas, or for a birthday, or for another holiday? What property would you use? Daask (talk) 16:47, 22 March 2020 (UTC)
To be a step clearer, though: the sheet music is not the song, nor is it the tradition. You need an item in between for the song, and whatever is the usual modeling between sheet music and song. - Jmabel (talk) 19:32, 23 March 2020 (UTC)
@Tagishsimon: I am asking because different Wikipedias point to one or the other. English Wikipedia, for instance, has a single article, Fingerpaint, to which "Fingerpainting" redirects. The English, German, and Swedish Wikipedias are linked to by Wikidata's "Fingerpaint", but the French and Chinese Wikipedias are linked to by Wikidata's "Fingerpainting". How can this be resolved so that the English Wikipedia article points to all of them? Would adding them to the redirect do it? —DocWatson42 (talk) 22:42, 25 March 2020 (UTC)
Stating the tower heights of a particular wind turbine model
Hi. V136-3.45 MW (Q71812982) is a wind turbine model which has multiple tower heights. The different tower heights are used based on the site conditions. I've used series ordinal (P1545) to state the different tower heights, but I'm not too sure if that is the best way. Adding a range also seems one option, but it seems like it is not supported on Wikidata? Any thoughts? Rehman06:15, 25 March 2020 (UTC)
Currently we use bare strings for pseudonyms/pen-names etc. almost everywhere. pseudonym (P742) has a datatype of string, and the standard way of denoting their use is via a subject named as (P1810) qualifier (also as a bare string). However, sometimes the pseudonym itself also has a Wikidata item (e.g. Richard Bachman (Q3495759), Mary Westmacott (Q12326342)), largely due to one or more Wikipedias having separate items for them. Such a Wikidata item ends up being pretty close to being an orphan, though, as other items that would usually be expected to link to it (e.g. the books authored under that name) have to use the string form, e.g. "Richard Bachman", rather than the item. There have been a few discussions around this at WikiProject Books, but no strong consensus, and as this is a wider issue that applies to other areas as well, it was suggested to bring it up for wider discussion here. Do other fields/projects handle this sort of case differently? Is there a nicer way to create links in cases like this? --Oravrattas (talk) 05:10, 18 March 2020 (UTC)
Since this applies to pen names of authors as well as stage names of actors and musicians, there will be MANY, including Mark Twain, George Eliot, George Sand, Poor Richard, Richard Bachman, Prince, Frankie Valli, Lon Chaney, Puff Daddy, Ludacris, ... --EncycloPetey (talk) 15:02, 18 March 2020 (UTC)
I had a quick scan of items using solo project (Q7558495) and most of the ones I looked at are again orphans, not linked from anywhere else. Of the ones that also obviously have a separate item for the person using that "stage name", all the albums etc I checked that were released as the 'solo project' seem to have a performer (P175) of the underlying person instead, and nothing to denote the name under which they were actually released. This seems wrong to me. When someone creates a separate persona to produce music or a book or whatever, it seems to me that Wikidata should generally prefer to credit that persona directly, with the underlying 'human' behind that then linked from the alter ego item. --Oravrattas (talk) 07:40, 26 March 2020 (UTC)
I agree with this, there might be an artistic (or even legal) reason for using different names and it seems wrong to second-guess that. We also have to remember using "stated as" as a qualifier for every single "performer" statement, which can be bit of a hassle and is basically not done at scale (and how would bots know?). But I have a feeling it's a slightly controversial position. Good ol' Richard D. James has notably been the subject of some back-and-forth on this. Moebeus (talk) 20:52, 26 March 2020 (UTC)
At some point I'm thinking we may need a property for "more specific version of" (which might handle calorie (Q87260855) and related cases better), but that's a discussion for another day. —Scs (talk) 13:19, 20 March 2020 (UTC)
We need proprerty like "is part of other article", "partly included in other article", "is covered in other article", "other article which thread is mainly covered". Eurohunter (talk) 10:34, 21 March 2020 (UTC)
Presumably that wouldn't be a property as such. There's a good reason that the links to Wikipedia articles aren't normal properties. Items (and their properties) are primarily about notable things in the world, not about WMF project content. - Jmabel (talk) 17:18, 21 March 2020 (UTC)
Sometimes, it exists only because of the rule that every Wikipedia article has to have a Wikidata entity. (And that's what the wording of the property main subject (P921) seems to suggest.)
Other times, it's because of various kinds of ambiguity. The canonical example of a Wikipedia article covering multiple topics (Q21484471) is JPEG (Q2195), which might mean the JPEG file format which is Q26329975, or the Joint Photographic Experts Group (the committee which both invented the file format and gave it its name) which is Q1702547. Another example, the one that got me thinking about this at all, is calorie (Q87260855), a notorious bit of English-language ambiguity which strictly means small calorie (Q130964) but which in common usage is often used for the "food" calorie (Q26708069) which is actually a kilocalorie, i.e. 1,000 times bigger. I don't think this case applies to Heatseekers charts (Q2303631), unless perhaps if people are in the habit of saying "Top Heatseekers" when they really mean "Heatseekers Albums" or "Heatseekers Songs". (Which, come to think of it, might be the case. I dunno.)
I think we could maybe use a better way of handling some of these situations, but I'm not sure. Similar to the "calorie" situation, I'm also thinking of traditional but ambiguous units like "ounce", which today have multiple, incompatible, more-precise definitions like ounce (Q48013) and troy ounce (Q1323615) and fluid ounce (Q69878540).
But, bottom line, I'm not sure we have to worry too much about representing these issues "just right" -- and especially when the main reason for the ambiguous entity is that it matches an equally-ambiguous Wikipedia article somewhere. —Scs (talk) 01:55, 22 March 2020 (UTC)
Is there a help page somewhere for this kind of situation? If not, it may be prudent to set one up as a resource to reference when encountering such articles/items. This also enabled the guidelines to be collaboratively iterated on to hopefully reach a point where there's consensus on the best way to model things. --SilentSpike (talk) 12:32, 24 March 2020 (UTC)
I wasn't pointing to Help:Modelling/Wikipedia and Wikimedia concepts#Compound_Wikipedia_pages so much for the has part(s) (P527) as for the fact that different interwiki links may need to go to different items. It's legitimate to create an item for whatever has a Wikipedia article in any language -- our linking approach pretty much requires that -- but then we have to model the appropriate relation of that item to other items more or less independently of the various Wikipedias. If the Wikipedia article is a really baggy monster, that may be difficult (or impossible) to do perfectly. - Jmabel (talk) 23:53, 26 March 2020 (UTC)
For an architectural drawing, this was added at Q64853752 with the description "Cathédrale de Marseille. Modification des fonts baptismaux : p[ou]r suppression de chaîne de suspension du couvercle". It seems to me that this have become a free form description, some of us were worried about. @Archives nationales DJI: --- Jura17:35, 26 March 2020 (UTC)
We are lagging behind with COVID-19 data - HELP WANTED
As long as random diseases are spammed to biographic items, there are obviously people with too much time and far too less sense for privacy and decency around here. Perhaps some of them could do some useful stuff instead. Grüße vom Sänger ♫ (talk)22:37, 22 March 2020 (UTC)
From the policy "Values for living individuals should generally not be supplied unless they can be considered widespread public knowledge or are openly supplied by the individual themselves". If the people are publically announcing that they have it, then it's within the policy to record it. ChristianKl ❪✉❫ 16:13, 25 March 2020 (UTC)
Yes LOL, not much have happened since I wrote above, so our data is now 8-5 days old for most of the countries, see query.--So9q (talk) 19:44, 24 March 2020 (UTC)
Honest question: what are people doing with the COVID-19 data on Wikidata? Certainly there are more reliable, more credible, and more up-to-date sources. Hopefully no health care professionals are making critical decisions based on Wikidata completeness (or lack thereof): they rely on CDC, WHO, ECDC, and local health offices. It's great when data is reasonably complete, but Wikidata is and will always be a haphazard, hodge-podge, idiosyncratic database run by volunteers, with no guarantees on data quality or completeness. -Animalparty (talk) 20:56, 24 March 2020 (UTC)
Wikidata can allow the smaller Wikipedia's to provide COVID-19 data in lanugages that currently don't have good official sources. Given that unfortunately COVID-19 does spread in more developing countries now, that seems to be a valid use-case to help those populations to inform themselves. ChristianKl ❪✉❫ 08:16, 26 March 2020 (UTC)
A majority of countries are now 7 days or more out of date. The main item about the pandemic is updated though.--So9q (talk) 20:55, 27 March 2020 (UTC)
Important for tool maintainers: last steps of wb_terms table migration
Hello all,
This is an important announcement for all the tool builders and maintainers who access Wikidata’s data by querying Labs database replicas directly.
In April 2019, the Wikidata development team started a heavy redesign of an important table of the database, wb_terms. Over years, this table has become too big, causing various issues. That’s why we decided to split it into several smaller tables. For more context, you can have a look at previous announcements: April 2019, May 2019.
We encountered a lot of unexpected issues on the road which required us to postpone the timeline, but the final steps of the migration are now happening. We published a blog post to explain more in detail what we did and why it was important to optimize the database structure.
Here’s an overview of what happened, what are the next steps and how you can update your code:
The new structure is now in place on Labs. You can read more details on this documentation page
Wikidata is entirely reading from and writing to the new tables since March 18th, which means the existing wb_terms table is not updated anymore
This means that any tools that are still reading from wb_terms and updating Wikidata based on its content have to be updated as soon as possible
In order to fully deprecate the table and make it clear that it is outdated, the wb_terms will be renamed to wb_terms_no_longer_updated on Monday, March 30th. Tools still querying this table at that date will certainly encounter errors.
To get an overview on how to update your queries, you can have a look at these examples
If you have questions or encounter any issues, please write on this Phabricator ticket or write me an email.
@Lea Lacroix (WMDE): with your action to rename "wb_terms" to "wb_terms_no_longer_updated" you're actively breaking tools. That's not very nice at all. Please just truncate the table (empty all rows) and maybe create an empty "wb_terms_no_longer_updated". What phab task is this tracked in? You completely missed the announced timeline and you can't expect tool maintainers to jump up with just a week notice. Multichill (talk) 17:37, 24 March 2020 (UTC)
The wb_terms table is not updated anymore. This means that tools that keep reading data from it, and potentially editing Wikidata based on this outdated information, can cause some damage to Wikidata on the long run. This is why we need a clear change. Yes, this may break tools if people don't update the code. Our intention here is not to break tools but to strongly encourage people to stop reading from the outdated database.
In that regard, truncating the table will not achieve that. If people are fine with their tool reading wrong data, they can still change the table name in their code to be the new table name.
We are aware that this project took way longer than planned, and that the initial announcement didn't reflect the current timeline. However, this also let a lot of time for people to understand the new structure and adapt to it. The next steps that we can finally take are only the continuation of the project and the new structure didn't change since 2019.
We apologize for the 1 week notice. We chose this date in order to give some time to people to update their code while not letting this outdated table be used for too long. Lea Lacroix (WMDE) (talk) 10:58, 25 March 2020 (UTC)
By removing a table you make some database queries go down in flames and effectively completely disabling a tool.
From my point of view it looks like you don't really care about the tool maintainers, you didn't even bother to link the relevant phabricator task and you're planning to disrupt our tools and make it our problem.
We have been discussing with the team to come up with an additional solution to avoid a sudden break of tools.
Since March 18th, the wb_terms table is not updated anymore. We failed to announce this step of the process and we apologize for that. This step is unfortunately not reversible. This means that wb_terms contains data that is outdated and will be even more outdated every day. The up to date data is in place in the new structure.
In order to ease the transition, we are going to add a wb_terms_no_longer_updated table which contains all the wb_terms data from the point in time the updates stopped. This should happen in the next hours. On April 6th (previously announced March 30th), we will empty the wb_terms table. This is similar to the solution proposed by Multichill but preserving a version of the current table, in case people still want to use it.
This way, people maintaining tools that directly query the database will have three options:
Update their code to use the new structure
Update their code by renaming “wb_terms” in “wb_terms_no_longer_updated” (temporary solution, as it is an outdated and frozen version of the data)
Change nothing, starting on April 6th the tools will query the empty wb_terms, which will return no data, but that might be less drastic for the tool than having no table to query at all
What do you think about this solution? Would it be an OK compromise to facilitate the transition to the new structure?
I understand that the change is a technical necessity, and I applaud the effort that went into it. But, why can there not be a view on the new structure that replicates wb_terms? Maybe even call it wb_terms, after moving the table to wb_terms_no_longer_updated? One of the reasons I did not change my code over until now is that before March 18, querying the new structure would have missed older labels etc., and I could not support both at the same time. A view, with a time limit of a few weeks, would give me and others time to migrate now that this is an actual option. --Magnus Manske (talk) 17:07, 26 March 2020 (UTC)
@Lea Lacroix (WMDE): Thanks for your reply Lea. That was the compromise I was looking for.
Yes, we don't want to leave an outdated table accessible for too long. But we should leave it long enough so people have time to update their code. We think one month is reasonable. Lea Lacroix (WMDE) (talk) 08:54, 27 March 2020 (UTC)
Mozilla does not list that as a "recommended extension". Do you know anything about its developer? Otherwise, that seems like quite a security risk for a small convenience. - Jmabel (talk) 20:13, 26 March 2020 (UTC)
┌────────────────────────────────────────────────────────────────────────────────────────────────────┘
The Mozilla listing for the extension links to this page, which says;
To be clear, just because an extension is not Recommended, that doesn't mean it's unsafe. It simply means it's not vetted by Mozilla and you should install at your own risk. Here are additional assessing the safety of an extension. [sic]
Do we have a consensus 1. not to use some sources or 2. to remove some sources that is considered "unreliable" or "deprecated" by some projects (e.g. Daily Mail, NNDB)? Previous we have some discussion not to do that, but @Nikkimaria: does not think so. This user also removed some Find a Grave sources where there's consensus to keep (in this case there's more reliable sources present, So I have no opinion against removing them, but they may be readded by Reinheitsgebot).--GZWDer (talk) 09:50, 28 March 2020 (UTC)
I think it varies if they are generally unreliable, or just considered to be so by some project, for some purpose or by user Nikkimaria.
We had countless RFCs on the deletions of Nikkimaria, asked them not to delete them, and they still continue. Not sure if another RFC would really help. --- Jura10:02, 28 March 2020 (UTC)
The whole point of deprecating claims is to keep the data but mark it as wrong. If anything this can help to show that sources are unreliable, whereas removing them just leaves the data open for re-addition. --SilentSpike (talk) 10:59, 28 March 2020 (UTC)
I think I have misunderstood, I thought this was about removing deprecated statements supported by unreliable sources, but it's just about removing "deprecated" sources. Haven't quite formed my opinion on this, I can see the argument for removing a source if a better is available, but also the logic in listing as many sources as exist. --SilentSpike (talk) 13:21, 28 March 2020 (UTC)
Looking at the type of statements that you do reference with it, it doesn't strike me as problematic. Nikkimaria would probably replace them with Britannica. Personally, I find Britannica problematic as it's a tertiary source, but I understand that this view isn't necessarily shared by them. Other references would be interesting. --- Jura14:14, 28 March 2020 (UTC)
"mul" perhaps. Or maybe not, if you don't know somebody's native language, maybe the property shouldn't be added. Ghouston (talk) 22:33, 27 March 2020 (UTC)
You want to change it from the meaningful "Wohnhaus in geschlossener Bebauung Ritterstraße 11" to "Wohnhaus in geschlossener Be..."? That makes no sense. No, I am not doing that. - Jmabel (talk) 23:39, 27 March 2020 (UTC)
Ich habe keinen Zugriff auf das zweite Alias. Es heißt wohl Wohnhaus in geschlossener Bebauung Ritterstraße 11 - kann man es entfernen bitte? Es ist jetzt unten in dm Feld für die Beschreibung in der ursprünglichen Datenbank eingefügt . Ich habe keinen Zugriff auf dieses Alias, kannst du es ändern/löschen? Danke, Conny (talk) 23:50, 27 March 2020 (UTC).
@Aschroet: You added this alias. I'm not taking sides in this. I thought this was about just a technical edit to do something obvious. Would you and User:Conny please discuss work out whether this alias should be there or not? I'm very hesitant to just remove what someone thinks is a valid alias. - Jmabel (talk) 03:55, 28 March 2020 (UTC)
It is a known bug, right click the edit button and it takes you to "Set Item" where you started adding aliases and separating them with "|". It is the only known workaround. When aliases overflow the text field it causes the error for some people. --RAN (talk) 05:29, 28 March 2020 (UTC)
Is it possible to make it so linking to a deleted item will show it's name? It's hard to follow the discussions on Wikidata:Requests for deletions when you can't see which item represents which entity --Trade (talk) 11:43, 25 March 2020 (UTC)
Yes, but i want the name of the deleted items to appear when linking to them items outside of the deletion log. --Trade (talk) 23:52, 29 March 2020 (UTC)
property for "represented by" in an electoral district?
Can represented by (P1875) be used for a political electoral unit (Q192611)? Is there a better "represented by" property for electoral districts? For purposes of discussion, here is an example of an electoral district with a statement for the name of the particular politician representing that district. Not sure P1875 is appropriate. Thoughts? Thanks. -- M2545 (talk) 15:31, 28 March 2020 (UTC)
Much better to do this with statements on the person and not worry about making it reciprocal, I think. The lists can get absurdly long and complicated otherwise. (Edit: I see that was the gist of the discussion above, as well. I think I missed that one.) Andrew Gray (talk) 14:50, 29 March 2020 (UTC)
Need help for merging
I can't merge Q28150740 and Q5295136, can anyone help? I've deleted the Wikipedia links from the first entry and I've added them to the second entry yet the entries should be merged, I think. Thanks. — Ark25 (talk) 18:34, 29 March 2020 (UTC)
They aren't the same thing. Laurell (Q28150740) is a family name item and Laurell (Q5295136) is a disambiguation page item. In particular, en:Laurell is a disambiguation page of given names and family names using Laurell, so can't be moved to Laurell (Q28150740). If the other language pages are specifically for Laurell as a family name then you can unlink them from the disambiguation item and relink them to the family name item. A merge is not appropriate. From Hill To Shore (talk) 19:15, 29 March 2020 (UTC)
@Ark25: Family name items shouldn't be deleted. They are a vital part of the structure for biographical entries. Both the disambiguation item and the family name item should be retained as separate items. From Hill To Shore (talk) 21:37, 29 March 2020 (UTC)
You need to use a different template at enwiki to indicate that it's a disambiguation page. --- Jura21:41, 29 March 2020 (UTC)
Without trying, I guess it is the English and Serbian descriptions that are conflicting. The solution is: choose one description for each of these languages, remove the other ones and repeat merge. (Looks like a question for a Serbian speaker.) --Matěj Suchánek (talk) 14:49, 31 March 2020 (UTC)
I think that this discussion is resolved and can be archived. If you disagree, don't hesitate to replace this template with your comment. Scs (talk) 13:42, 4 April 2020 (UTC)
can't add link to "en"
Hi there, on Alfa Romeo (Q622489) i would like to add the link to the article on the English Wikipedia which is "Alfa Romeo in Formula One" however, for some reason i can't add that link. What is happening here? Kind regards, Saschaporsche (talk) 18:28, 25 March 2020 (UTC)
This is because Wikidata requires items be dissected to their finest element, and broad concept encyclopedia articles like "history of X" or "X in Y" don't necessarily have 1-to-1 correlations with other language wiki articles covering the same topic. The English article is linked to Alfa Romeo in Formula One (Q65960697), which seems to be the most broadest scope (a subclass of Alfa Romeo in motorsport (Q173886)). Alfa Romeo Racing (Q622489) refers to the Alfa Romero team(s), which compete in the events, but are not the events. -Animalparty (talk) 18:56, 25 March 2020 (UTC)
Q622489 is the current team, there are other items for previous Alfa Romeo teams in the 1950s (Q61467970) and 1970s-1980s (Q84562554), and as an engine manufacturer (Q31890903). Q65960697 combines all of these, but some links have been moved to Q622489 when the items were merged, possibly incorrectly. Peter James (talk) 00:02, 26 March 2020 (UTC) I looked at the content again, and they are duplicates - it was just the title that suggested they were not. Peter James (talk) 14:06, 30 March 2020 (UTC)
If you take a look at Alfa Romeo (Q622489) you can see that there is NO link to any article in English language. I still think that adding "Alfa Romeo in Formula One" to Q622489 is the correct solution. kind regards Saschaporsche (talk) 12:25, 26 March 2020 (UTC)
How to find the Wikidata entry for a Wikipedia article?
I'm at en:Diviš and I want to add it to Wikidata (cs:Diviš - Diviš (Q1231604)) and it doesn't work (I was ussing the button "Add links"). It says "Could not link pages: failed to merge corresponding items on Wikidata. Attempted modification of the Item failed.". So I guess the English Wikipedia article already has an entry at Wikidata and it can't be merged with this entry Diviš (Q1231604) (which contains items for cs, de and ru languages).
If you are viewing on a desktop computer, the Wikipedia article should show a link to the Wikidata item on the left hand panel. This should be present in all language versions of Wikipedia if the article has been linked (if you can't find the Wikidata link in the foreign language, just hover over each link until you see the one pointing to the Wikidata URL).
In this case, the English article is sitting correctly at Diviš (Q21502071) as it relates solely to family names. If the other articles are solely about family names they should be relinked to the family item. If they include other subjects in addition to family names they should remain linked to the more general disambiguation item. From Hill To Shore (talk) 07:48, 30 March 2020 (UTC)
Article “De rerun natura” (Q861986) of “enwiki” should be linked with the article “O přírodě” (Q15640251) of “cswiki”
The article “De rerum natura” (Q861986) of the “enwiki” doesn’t link to the Czech language version - article “O přírodě” (Q15640251) of the “cswiki”. I am obviously not allowed to save the link (I tried it to be sure and received an error).
Wikidata objects for instances of Wikidata:Events? Out of scope or relevant?
I'm soon hosting an event. Regarding Wikidata:Notability it may qualify for "2." but I doubt it fulfills "The entity must be notable, in the sense that it can be described using serious and publicly available references.". Am I in the wrong for bringing this up or is this a topic of any value? Datariumrex (talk) 09:44, 30 March 2020 (UTC)
Merging entries about name disambiguations
Q16445474 – Bartuska (Q36935631) — Bartuška (Q82014430) — They are all about the family name "Bartuška" (and it's variant "Bartuska"). Yet some of the Wikipedia articles were created as plain disambiguation pages (cs, de, lt) instead of being created as name disambiguation pages. That's why Wikidata has more than one entry for the same thing. These entries should be merged into a single entry and the other two should be deleted. And some people will argue they should not be merged - like for example here.
At this moment, it is actually harder to interwiki-link such articles between them, because of Wikidata. It was easier to link them in those times when Wikidata did not exist yet.
For sure, there are (tens of) thousands of situations like this - for example "Bareikis" and "Diviš". So I think someone should be in charge of solving such issues. — Ark25 (talk) 03:46, 30 March 2020 (UTC)
Already solved, we distinguish items about disambiguation pages, given/family names and its variants (see Wikidata:WikiProject Names), although wikis often conflate these into one or more pages. No need to merge these.--Jklamo (talk) 11:12, 30 March 2020 (UTC)
Some languages use disambiguation templates on surname pages, then when the Wikidata items are created, disambiguation statements to them, usually by bots. If these are not disambiguation pages, the people who operate these bots need to know which wikis do this, and to stop adding disambiguation statements. Peter James (talk) 18:36, 30 March 2020 (UTC)
how bad are loops in the subclass-of hierarchy?
I think I've heard this mentioned somewhere, so it's probably no surprise, but I just had one of my little programs crash on what turned out to be a loop in the subclass of (P279) hierarchy:
Is this considered a problem, and is there any wish to / hope of fixing it? (In the meantime I'll fix my program; I'm not holding my breath. :-) ) —Scs (talk) 11:25, 30 March 2020 (UTC)
A loop in the subclass of (P279) hierarchy is not *necessarily* an error, but such loops are almost invariably a symptom of a modelling problem. It can be difficult, however, to determine just what the fix should be. In this particular case, which of the above subclass of (P279) links are incorrect, if any? Or are they all correct, and the four concepts in question are actually the same (and should probably be merged). The english descriptions of the concepts are not very helpful here. Peter F. Patel-Schneider (talk) 12:47, 30 March 2020 (UTC)
matter (Q35758) says it has more than one definition; substance (Q27166344) appears to be an attempt to define an ambiguous term (two definitions of "substance" are the items with "substance" in the title, but there are more). Because of the ambiguity, either "substance" or "matter" may be a subclass of the other. The difference between material substance and matter is also unclear - matter can include antimatter. Peter James (talk) 13:17, 30 March 2020 (UTC)
If I understand the idea behind it, physical substance (Q28732711) is supposed to represent anything made out of fundamental particles, including the particles themselves (since massless particles travel at the speed of light and don't really form "substances".) So I've broken the loop by removing the relevant subclass statement from that item. Ghouston (talk) 22:54, 30 March 2020 (UTC)
Wikidata weekly summary #409
Here's your quick overview of what has been happening around Wikidata over the last week.
Upcoming: The Celtic Knot Wikimedia Language Conference will take place fully remotely in July 2020. Call for submissions with remote formats is open until April 30th.
Upcoming: the Wikidata Wochenende (previously in Ulm) will take place fully remote on June 12-14
Could not save due to an error. The save has failed.
For wikidata: https://www.wikidata.org/wiki/Q88659818, I am getting the error: "Could not save due to an error. The save has failed", when trying to publish the Wikipedia item as en:David H. McConnell.
@Greghenderson2006: Sorry about the bad error message. However, there already was a Wikidata item for this person - David H. McConnell (Q5234543) - which had that enwiki link. I've merged the two items now. You can find an existing Wikidata item for any Wikipedia page by looking at the left-hand navigation bar and clicking on the "Wikidata Item" link (under "Tools"). ArthurPSmith (talk) 17:49, 30 March 2020 (UTC) (fixed ping - ArthurPSmith (talk) 17:49, 30 March 2020 (UTC))
@Greghenderson2006: Done. Usually we merge into the item with the lower ID (which would have been the 3rd in this case) but since it had nothing linking to it internally I went ahead as you asked. See Help:Merge for how to be able to do this yourself... ArthurPSmith (talk) 20:16, 30 March 2020 (UTC)
I posted about this before, just wanted to update in case any others have been dealing with this as I've figured out the cause.
Since 2015 the BBFC have classed archive.org as an adult content website (see their re-review in 2018 here). This means in the UK it is blocked by ISP adult content filtering if your ISP is a mobile operator. For some reason my ISP had not started enforcing this until recently I guess (although I can still access clearly adult websites elsewhere). Anyway, if anyone else has the issue you should be able to disable the filtering somewhere. --SilentSpike (talk) 17:46, 30 March 2020 (UTC)
Data on Coronavirus > Wikidata > Wikitable in English
I think a key part of this would be to translate fr:Modèle:Tableau qualificateurs; it seems currently not to be available in English. I could probably translate the documentation from French into English, but I don't have access to the code that actually does the work. I don't know what language that's written in, but I could probably find the French-language labels and translate those into English and make it all work with a little help. (I started writing Fortran in 1963. I've written code in many different languages since then and debugged code in languages I didn't really know.)
Updating commons:File:CoViD-19 US.svg automatically will take more than just this, but this is a big part of what would be needed to automate that. And, of course, this capability has many other applications in facilitating crowdsourcing all kinds of data collection and analysis, e.g., a v:International Conflict Observatory, which I believe could potentially facilitate a dramatic reduction in armed conflict worldwide. Thanks, DavidMCEddy (talk) 04:47, 31 March 2020 (UTC)
COVID-19 confusion
The COVID-19 page here is pointing at a mixture of topics: some are descriptions ofthe disease but others are pointers to pages about the current pandemic. I am no expert on naming conventions but In my wiki-travels I met one enwiki wikipedian who has been doing nothing else for years. Wikidata should engage some experts like this here imio. Ottawahitech (talk) 14:58, 31 March 2020 (UTC)
Best way to go about getting all "classes" that have atleast one {?thing wdt:P31/wdt:P279 ?class} relationship
Hi,
I'm building an interface for performing simple SPARQL queries on Wikidata. I need an autocomplete feature for this, for which I've used the Wikidata query service thus far. However, I only want to list wikidata entries that have at least one entry that is an instance- or subclass of that first entry.
Should I download a dump and set up my own index? If this is possible to do using SPARQL or some other service, that would be greatly preferrable. – The preceding unsigned comment was added byMatsjsk (talk • contribs) at 09:06, 25 March 2020 (UTC).
Done by User:IN (redirected page) DemonDays64 | Talk to me 17:59, 8 March 2020 (UTC)
Not done@demonDays64:IN is change name from "霜奶仙"(Alcremie),so this page isn't delete.--IN (talk) 15:28, 11 March 2020 (UTC)
Notified participants of WikiProject Medicine. -Theklan (talk) 10:20, 12 March 2020 (UTC)
WikiProject Ontology has more than 50 participants and couldn't be pinged. Please post on the WikiProject's talk page instead.
/ location (P276) 
Info Just a valid ping for Magnus, as you skipped the blank in his username. Grüße vom Sänger ♫ (talk) 11:15, 22 March 2020 (UTC)
Comment Per @Dhx1:, has part(s) (P527) is not allowed on this purpose, please feel free to convert the remain has part(s) (P527)s to be main subject (P921)s. --Liuxinyu970226 (talk) 00:39, 24 March 2020 (UTC)