Shortcuts: WD:PC, WD:CHAT

Wikidata:Project chat

From Wikidata
Jump to: navigation, search
Wikidata project chat
Place used to discuss any and all aspects of Wikidata: the project itself, policy and proposals, individual data items, technical issues, etc.
Please take a look at the frequently asked questions to see if your question has already been answered.
Please use {{Q}} or {{P}}, the first time you mention an item, or property, respectively.
Also see status updates to keep up-to-date on important things around Wikidata.
Requests for deletions can be made here.
Merging instructions can be found here.

IRC channel: #wikidata connect
On this page, old discussions are archived. An overview of all archives can be found at this page's archive index. The current archive is located at 2017/04.





for permissions


for deletions


for deletion

for comment

and imports

a query



Properties in general. Property Category tagging to Projects. Teaching Ps & Qs.[edit]

Been looking "top down" at properties that have been created to date.

My interest lies mostly for the following projects: Organizations, Economics and Companies. I see that properties (at least via the related discussion page, may be specifically tagged to Wikidata Projects.

There is great confusion among Wikipedia users migrating to Wikidata over what should be an item (Q) and what should be a property (P). Questions around property inheritance from parent items is also a typical stumbling block for new project members.

I understand project members, typically go out and tag properties (via a Category: tag) that appear relevant to their project. This is to guide new users to preferred / frequently used properties, and avoid new property proposals that merely duplicate the standing ones. This category tagging (to a Project) process seems very "hit or miss", especially since properties have expanded greatly.

Category:Properties_by_number_of_uses, a potential aid for Projects focusing on the most important properties first seems to be in error. The numbers should go up at the "bottom" and never decline. I assume this is auto generated? Any chance for a fix?

Also, any suggestions on how to teach new project members "what should be an item (most) vs. what should be a property (few)", and how to do that, along with how to focus new project members on filling in important property instances vs. creating more "far flung" properties very much appreciated. All projects must wrestle with these issues. Rjlabs (talk) 15:14, 18 April 2017 (UTC)

Can you point me to discussions where contributors are confused as to what should be a property and what should be an item? I have heard a lot of things to be confused about, but have never encountered this one. --Denny (talk) 20:56, 18 April 2017 (UTC)

Pictogram voting comment.svg Comment Here is an example. Accounting standard is a fairly abstract concept. For a reported number you need to know what standard was used (generally country dependent, and specific to a year. Accounting standards constantly evolve and are typically maintained, codified and implemented by year. Frequently there are XML Schema files (.xsd) by year that very specifically describe reporting under a specific standard. These .xsds are written in the eXtensible Business Reporting Language (xBRL)

Looking at what we have now:

Accounting standard (Q1779838)

Generally Accepted Accounting Principles - US (Q650978)
Generally Accepted Accounting Principles - US - 2015 (Q29168386)
Generally Accepted Accounting Principles - US - 2016 (Q29168379)
Generally Accepted Accounting Principles - US - 2017 (Q29168365)
Generally Accepted Accounting Principles (Q16981751) (Canada)
Generally Accepted Accounting Practice (Q16981746) (UK)

Note: accounting ontology is not unified across years or across countries, even in xBRL. For example, labels used in the UK 2015 .xsd might be exactly the same or similar to those used in the US 2015 .xsd but the actual meanings could be different. Likewise the exact same measurement might have two different labels. A global ontology for accounting standards, with concordances between different countries/years/etc. will likely take years to fully emerge.

In addition to the above Q's is Wikidata:Property_proposal/accounting standard. An accounting number with a general label is nearly useless unless you know the specific standard under which it was purportedly prepared. Context information is essential. For example:

Take a quick look at how xBRL handles company financial data because they have been at it for many years now and are good at it. See specific example in XML at [[1]]. In the simplified example you can see they are reporting four numbers: Other Operating Income, Other Administrative Expenses, Other Operating Expense, Other Operating Income. Each of those “numbers” has a 'context id reference which shows the period (start date and end date, here a year), and the unit id (here EUR, which is specifically identified as iso4217:EUR. Best of all xBRL has a very good way to describe a huge variety of units in a flexible, and very specific way. More at:

In designing its company object/property structure there are several articles that WikiData can "go to school on" to facilitate high quality linking:

  • An Ontological Approach to XBRL Financial Statement.pdf
  • Dave Raggett - xBRL & Sparql.pdf
  • Fundamental Analysis Powered by Semantic Web.pdf
  • Publishing XBRL as Linked Open Data.pdf
  • Representing financial reports on the semantic web a faithful trans…_ORIGINAL.pdf
  • Semantic Integration Approach to Efficient Business Data Supply Chain - Integration Approach to Interoperable XBRL.pdf
  • Semantic Paths in Business Filings Analysis (thesis).pdf
  • Translating the FINREP taxonomy using a domain-specific corpus (2103).pdf
  • Translating XBRL Into Description Logic - An Approach using Ptotoege, Sesame & OWL.pdf
  • Triplificating and linking XBRL financial data.pdf
  • Using Semantic Web Technologies to Facilitate XBRL-based Financial data Comparability.pdf
  • XBRL - Consequences to Financial Reporting, Data Analysis, Decision Support, and others.pdf
  • XBRL and open data for global financial ecosystems - A linked data approach.pdf
  • XBRL taxonomies and OWL ontologies for investment funds.pdf

There is a significant volume of financial and economic data of interest to WikiPedia currently. But, this represents only the very top level of what is actually available in readily accessed public repositories. At present WikiPedia seems only interested in less than a dozen numbers per company, and only to report for the most current year. That's a great start but over time WikiData may ultimately want more. (For example a 10 year historical reporting of the key numbers to show trends, etc.) In any event WikiData is today confronted with the cross country compatibility issues.

Getting the Q's and P's right at WikiData, especially to elicit all required contextual data around financial statement number presentation will greatly facilitate (or inhibit) linking company data. "Accounting Standard" as a P and as a Q is just one example of potential P/Q options and confusion. Rjlabs (talk) 14:27, 21 April 2017 (UTC)

Thanks, that was insightful. Thank you for taking your time to answer. I see that you already found the relevant Wikiprojects, Economics and Companies. I would claim that it is the task of these Wikiprojects to figure out how to best represent the data within their domain, and then check with the wider community if that's OK - or just go ahead and do it. Missing or badly defined properties would also be discussed there. --Denny (talk) 18:43, 21 April 2017 (UTC)
  • @Rjlabs: Thanks for another insightful posting on company data --Vladimir Alexiev (talk) 09:43, 22 April 2017 (UTC)
    • I suggest you copy it to the relevant project, since it will be lost quickly here.
    • Do you have links for the papers you listed above? Could you add those links above?
    • Your original question is quite simple. After a prop "accounting standard" is made, you'd state <financial report> "accounting standard" Generally Accepted Accounting Principles - US - 2015 (Q29168386) (assuming the report uses that particular standard).
    • But that's assuming <financial report> is a structured item (holding the numbers). If that's just a link to a PDF, then it's enough to say <company> URL <financial report> / of "financial report" (or whatever more specific thing they filed). There's no necessity to state which accounting standard the PDF conforms to, since a user can read that in the PDF, and you're not processing the numbers by machine
    • IMHO WD won't be ready to take detailed financial reports for a couple of years yet. One question is capacity, the bigger problem is having an interested community of people like yourself who know the domain, what needs to be integrated, and what apps can be made on top of it
    • As you say, financial terms can mean different things in different countries or even editions of a standard. So then, before those numbers are input to WD, you need to harmonize meanings across national standards and variations, so you can make properties that have a fixed well-defined meaning. This sounds like a huge undertaking to me.

How do I enter a quantity in QuickStatments without lower and upper bounds?[edit]

I try to add Q4115189 P3864 5.5 P518 Q27177113 P585 +2015-01-17T00:00:00Z/9. Unfortunately it doesn't add any of the qualifiers. The datasource is and it doesn't specify anything about the accuracy. ChristianKl (talk) 16:12, 19 April 2017 (UTC)

@ChristianKl: Add "±0" after the "5.5". I believe the new QuickStatements should work. Mahir256 (talk) 16:57, 19 April 2017 (UTC)
Thanks, the new version seems to handle it better. I have a follow up question: What's the easiest way to go from a list of country names to adding information to the corresponding item via QuickStatements? ChristianKl (talk) 18:28, 19 April 2017 (UTC)
If the list of country names corresponds exactly to the names of their Wikipedia articles, you can run something like "(country name) Senwiki (country name)" using the old version to get their respective QIDs and simply replacing the column containing the country list with those QIDs before adding your information with the new version. Mahir256 (talk) 19:02, 19 April 2017 (UTC)
Well numbers like 5.5 or 6.5 seem to work well. No luck with 5.6, or 0.641. I get something like 5.5999999999999996447286321199499070644378662109375 or 0.6410000000000000142108547152020037174224853515625 123 (talk) 20:18, 19 April 2017 (UTC)
The new QuickStatements are not the only tool to suffer from this bug. HarvestTemplates also had to block numbers with decimal points although it's using a different API. I think it happens when data are being converted to JSON in order to send them to Wikidata. Matěj Suchánek (talk) 06:33, 20 April 2017 (UTC)
The wikibase API actually has a call to convert a string to the right form in a way that works like the wikidata API - people should be using that rather than their own (whatever language their code is in) string parsing to real numbers. ArthurPSmith (talk) 15:53, 20 April 2017 (UTC)
I notice this edit to one of the sandboxes. It obviously does not produce the correct result. It indicates 5.5 is an exact value, when the source clearly indicates it is an estimate. Jc3s5h (talk) 15:28, 21 April 2017 (UTC)
Well, I guess the user was just experimenting how to produce a value without lower and upper bound. In the context of a sandbox I wouldn't say then, that the edit didn't produce the correct result. (It produced what it produced and the user now knows.) 123 (talk) 15:43, 21 April 2017 (UTC)

Wales buildings[edit]

Heads-up: I (via User:Reinheitsgebot) am creating ~27K buildings with heritage status (Cadw Building ID (P1459)) in Wales. Data via User:Jason.nlw, from original source. Added some semi-automatic inferred statements where possible (GPS from easting/northing etc.) --Magnus Manske (talk) 22:02, 19 April 2017 (UTC)

@Magnus Manske: you triggered 27055 constraint violations for something that is marked as mandatory. Can you please add the missing country (P17) -> United Kingdom (Q145) and undo this edit? Multichill (talk) 19:45, 21 April 2017 (UTC)

How to say "Telescope was used to discover object"?[edit]

James Lick telescope (Q6138045) was used to discover Amalthea (Q3257). What's a good way of recording that? discoverer or inventor (P61) (that I'm currently using) seems to be more about the person that discovered it, while site of astronomical discovery (P65) and location of discovery (P189) seem to work the wrong way around ("X was discovered by Y"). Any suggestions, or does this need a new property? Thanks. Mike Peel (talk) 23:29, 19 April 2017 (UTC)

item operated (P121) ? and I was thinking as a qualifier to P61.  — billinghurst sDrewth 15:44, 20 April 2017 (UTC)
Thanks @billinghurst, but I think that's also the other way around ("X was discovered by Y" rather than "Y was used to discover X"). Maybe that's the better way of doing it, but it seems odd we can't do the link both ways around. Thanks. Mike Peel (talk) 21:17, 20 April 2017 (UTC)
I was seeing it more as A used B to discover C. B is passive, so I don't think that it discovered anything. Cook -> Endeavour -> eastern Australia (to Europeans as the indig. already knew it was there)  — billinghurst sDrewth 23:52, 20 April 2017 (UTC)
@billinghurst: I was hoping to pull this into en:Template:Infobox telescope to say that "this telescope was used to discover these things". Reverse queries are complicated and expensive... Thanks. Mike Peel (talk) 00:04, 21 April 2017 (UTC)
significant event (P793) ?  — billinghurst sDrewth 00:07, 21 April 2017 (UTC)
There is also the simple, generic properties "uses", "used by" +++ others (subject of-, object of- type stuff) around.
@billinghurst: significant event (P793) -> discovery (Q753297) -> (which property?) -> Amalthea (Q3257) might work. Except I'm not sure what property can be used in the middle - "astronomical object" would be the obvious thing, but that's not suitable for a property. Other generic properties probably have the same issue... Thanks. Mike Peel (talk) 00:48, 22 April 2017 (UTC)
Ah, of (P642) might work for that middle property! It's convoluted, but it might do. Thanks. Mike Peel (talk) 00:57, 22 April 2017 (UTC)
W00t, that seems to work. :-) en:James Lick telescope now has a "discovered" line in the infobox. :-) Mike Peel (talk) 01:04, 22 April 2017 (UTC)

Missing tools/toolchain[edit]

I started to contribute in wikidata because I hoped I could help to provide consistent and up to date data for changing entries in country and economy infoboxes. But I unfortunately have to note that wikidata is still missing the tools to do so in the form of batch jobs. What is missing are tools which allow for properties with "quantities" - including possibly decimal points - with and without units:

  • Delete or deprecate deprecated values. (For many datasets not only newer values are added, but the older values are revised. At least one should be able to deprecate those values - but I guess it would be better to even simply delete them and replace them by the revised values.)
  • Add the rank "preferred" to the newest value.
  • Add needed qualifiers.
  • Add the full set of information about the source (reference) needed to create a reference in wikipedia.

This is not to complain, but just to indicate where I think some progress would be really helpful. 123 (talk) 21:11, 20 April 2017 (UTC)

I can see at least 3 distinguishable activities in your proposal:
  1. Adding "economic figures" with point in time (P585) qualifier and source
  2. Choosing which statement (among several existing) need to be displayed by wikidata-consumer
  3. Purging statements that contains old (or no longer relevant) data
My understanding is that (1) is fully covered by quickstatement. (2) has to be decided by consumer (e.g. infobox code might select statement with most recent p585 and show up/down arrow depeneding on the value of second most recent). In order to impelemnt (3) we need to reach consensus regarding thresholds --Ghuron (talk) 11:23, 21 April 2017 (UTC)
To be able to perform so many different tasks we need bot and not tools. What is currently missing is a large number of bot operators able to provide some services for data import and data update. I thing the first step should be the development of a format for data which have to be imported in order to be able to share the tasks between the bots. Currently each bot operator develops its data format according to its use. Snipre (talk) 11:38, 21 April 2017 (UTC)
Re 2) For preference do we want the latest set of figures to be preferred? or to push preference based on the latest update? Maybe you do, maybe you don't. What you are discussing is more about having a query that retrieves the figures based on the latest date set. I am not certain if you have a set of data that you would want to delete it. Not something that I think that we wish to promote. If you have two sets of data for March 2016, wouldn't you deprecate a set at that point. Showing a comparison between the same set of data though from initially cast, to finally cast can have value; eg. (un)employment figures often have variance.  — billinghurst sDrewth 12:10, 21 April 2017 (UTC)
  1. Is not covered by quickstatement: While for instance the NIOSH Fork supports quantities with numbers, neither the original version of quickstatements nor do. But the NIOSH Fork doesn't allow to include additional information to sources (title, author, date accessed etc.) because of issue #52. All versions suffer from issue #65.
  2. According to Help:Ranking#Preferred_rank "preferred rank" has exactly the function to indicate which is the most reasonable default value to choose. (For instance: data just published by the IMF on GPD for 2016 would be the most reasonable candidate for inclusion in an infobox.) Consumers could override that if they want. Looking at the reality of the wikidata modules now: they choose either the value with the rank preferred or (if there is none, what very often is the case) the entry first added or the list of all entries (neither makes much sense).
  3. Whether deprecating or deleting is the better solution should probably have its own discussion. If at the moment somebody looks up the HDI for Venezuela she will find 0.762 for 2014. (The wikidata module will return 0.628 from 1980!) If she follows the source she will find the value 0.767 for 2015 - looks like a significant increase. If she looks more closely, she will find that the source gives 0.769 (not 0.762) for 2014. (So there was a slight decrease from 2014 to 2015.) I think that's a problem and one at least should mark the data as deprecated. I don't have a huge problem with retaining them, as long as they are marked as deprecated (and the newest revision is added). I just don't see much use in retaining such data - given that there are provided newer ones, based on the newest revision. If something wants to study the ways the data are changed through revisions she really should go to the primary data source. (As far as I could see if people do provide revised data manually they just replace the older data.)
Actually, if either would allow (1) to use units, (2) to add/change ranks and as a bonus fix issue #65 or the NIOSH fork would fix issue #52, allow to add/change ranks and delete statements (and as a bonus fix issue #65) all the different tasks I'm missing could be performed! As a first huge stepp it would be enough if just would allow to use units (whithout loosing any of the features it already has): I could then go ahead and for instance add for all countries GPD values etc. for 2016 and replace the deprecated values for 2014 with revised ones, which agree with the source given. 123 (talk) 15:21, 21 April 2017 (UTC)
Re "preferred" here as you are identifying the figures with other identifying and selectable data, eg. GPD as at YYYY, forcing a preference is unnecessary. I generally see that preferences apply when there is no means to qualify a preference, eg. two images and you want to pull one; two dates of birth provided though the sources are of different verifiability; two VIAF identifiers, we just need one.  — billinghurst sDrewth 22:34, 21 April 2017 (UTC)
Indicating a preference doesn't "force" anything it just indicates a preference. If you want to call it forcing than the absence of a preferred value "forces" on you either a unordered list or some random value (the first one added) and I think that's worse than "forcing" on you the most up to date figure. In infoboxes for countries and economies that's what people normally want. As far as I know the wikidata modules on various wikipedias, they don't allow you to choose values based on qualifiers (like point in time (P585)) rendering properties with various values without a preferred one unusable for most cases. 123 (talk) 00:28, 22 April 2017 (UTC)
Looking for most recent point in time (P585) qualifier is not exactly rocket science, and, as you can see, for Venezuela (Q717) among Human Development Index (P1081) statements it will select 2014 value (0.762) which is decrease comparing to 2013 value (0.764). It is much more sustainable solution comparing to playing with ranks. But I agree units (I thought they were there) and more accurate calculation of floats (#65) is desperately needed for quickstatement. BTW I suspect that long trail of random digits can appear only if you specify ±accuracy. --Ghuron (talk) 06:00, 22 April 2017 (UTC)
It looks like conversion to float occurs even if ±accuracy is not specified. @Magnus Manske: why wouldn't you just pass value as-is in this case (without *1)? --Ghuron (talk) 06:25, 22 April 2017 (UTC)
@Ghuron: Where can I see a wikidata module returning the 2014 value in the Venezuela HPI example? (And possibly try it out for other examples?) 123 (talk) 12:55, 22 April 2017 (UTC)
@123: I was assuming that you can see it in source code Face-smile.svg In ru-wiki we have no integration for Human Development Index (P1081) yet, but mechanism works for many other properties. For instance ru:Маунтфилд infobox is constructed based on Mountfield (Q21180783) and all financial figures formatted with the piece of code, mentioned above. If you notice any glitches there, please let me know --Ghuron (talk) 15:50, 22 April 2017 (UTC)
@Ghuron: Well, this is exactly one of my points: There is no wikidata module allowing me as an editor to choose, that for any property I want to use the newest year or (even better) a given year is shown. I don't think the approach that the developers of the wikidata module decide which properties to integrate in such kind of mechanism is efficient. Add to this, that point in time (P585) is just one property which might be relevant to determine the preferred value. Have intelligent humans to give an indication what - as default - probably provides the most useful value, and allow editors to freely choose another one, seems to be much more promissing. And as an asside: In Mountfield (Q21180783) the values used in ru:Маунтфилд are actually (and IMHO rightly so) marked as preferred ... 123 (talk) 18:23, 22 April 2017 (UTC)
There is nothing wrong with your approach - if you believe that "intelligent" fetching through millions of statements and determining their ranks is the way to go - why should I stop you? Face-smile.svg But please do not "delete" old information (quickstatement supports that), technically it is possible to build infographics based on how HDI was changing over time period --Ghuron (talk) 11:43, 23 April 2017 (UTC)

We have 2 items for marketplaces[edit]

market (place) (Q132510) marketplace (Q330284)

I propose to use any. d1g (talk) 18:30, 20 April 2017 (UTC)

I tried to merge it but there are 4 conflicts. I had to remove the enwiki one because it was a redirect. MechQuester (talk) 01:57, 21 April 2017 (UTC)
Given their descriptions I think someone thought that e-Bay is a marketplace (Q330284) but no market (place) (Q132510). ChristianKl (talk) 12:49, 21 April 2017 (UTC)
It should be discussed at WD:IC, or at least at nl, pl, pt or zh-forums. I've created conflict page. --Infovarius (talk) 11:59, 26 April 2017 (UTC)

Changing the data type of properties[edit]

Is it possible to change the data type of properties after creation? I think ISIL ID (P791) should be classified as an "External Identifier", not as a "String".

When I became property creator, I learned that it is not possible. However, things may have changed in the last two years. Jonathan Groß (talk) 09:02, 21 April 2017 (UTC)

When the external identifiers were introduced, many string properties were converted to external ids, so this it is technically possible. However, on WD:Identifier migration/0, P791 is listed under Properties with serious objections to conversion. Matěj Suchánek (talk) 11:52, 21 April 2017 (UTC)
Indeed. However, no valid objection is raised there. As I said over a year ago "The first three letters of 'ISIL ID' stand for 'International Standard Identifier'". It's high time this was converted. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 22:21, 26 April 2017 (UTC)

LC Demographic Group Terms (LCDGT) needs to be able to be added as an identifier.[edit]

I would like to be able to add identifiers for Library of Congress Demographic Group Terms, but that is not currently an option. How can we add this? For example for the Wikidata entity Q4172847 (Filipino people), there is an LCDGT established: I would like to add the identifier dg2015060630.  – The preceding unsigned comment was added by (talk • contribs).

Tool to look for the existence of one identifier in Wikidata[edit]

Hi, I am looking for tool able to search the existence of one identifier in Wikidata. My problem is the following: I found several cases where contributors created manually new item but the concept was already existing in WD in another item. Searching an existing item using the label is not efficient enough as we can have different names for an unique concept and not all languages have one label.

We should have an user friendly interface where we can put the property and the value and the search can be done in a similar way as the current search tool. For example I want to know if an item has the property P235 with the value JPUKWEQWGBDDQB-DTGCRPNFSA-N. How can I do this search ? I know we have SPARQL query but we can't propose that tool to contributors who contributes occasionally to WD. Snipre (talk) 08:35, 22 April 2017 (UTC)

I don't think it would be a good workflow to tell new people to run a search beforehand with a new tool given that this would make the work of entering data more complex.
To me this issue feels like it's about implementing constraints at the moment data get's entered. If a person put's in a value for an identifier that violates the single value constraint, they should be alert that they are doing so and have to confirm their actions. It would also be possible to offer the user at this point a choice to merge the two items. ChristianKl (talk) 09:17, 22 April 2017 (UTC)
@Snipre: ran into that problem several times. Take for example RKDartists ID (P650) -> 32439. How do I find Vincent van Gogh (Q5582) based on "32439"? I can do a sparql query, but that's not very user friendly. I thought I found a tool where you could put in the property and identifier and it would give (redirect?) you the Wikidata item. The only thing I could find was beacon. Not extremely user friendly, but it is very fast and easy for the power user.
The answer based on beacon is Trifolin (Q7841515). Multichill (talk) 10:09, 22 April 2017 (UTC)
@Lea Lacroix (WMDE): This is perhaps something to add in the development plan: create an "Advanced search" tool in order to find item not only based on the labels/aliases but using an interface where the user can enter one property and a value. In order to avoid to use this tool to extract data, we can reduce the choice of the properties to the ones defined as identifier. Something like this page where the "All Fields" menus can be replaced by the property list like the one used to choose property when adding a new statement. Snipre (talk) 12:42, 22 April 2017 (UTC)
@ChristianKl: I don't propose to force contributors to search before creating a new item, but I want a tool in order to be able to recommend it when people create duplicates. Currently I doing a lot of maintenance and one of most important tasks is the merge of duplicates. And currently I don't think we will be able to avoid creation of duplicated items because the current Search tool only focused on labels/aliases which are one of the less reliable identifier parameters. Reinforcing data input based on constraint violations was rejeted by the development team in its development plan, see Wikidata:Development_plan#UI_redesign. Snipre (talk) 12:59, 22 April 2017 (UTC)
@Snipre: Phab:T99899 /
You can't just search for JPUKWEQWGBDDQB-DTGCRPNFSA-N because the search engine doesn't index strings in statements (external-id uses string). Can't find the task for that one. Multichill (talk) 14:45, 22 April 2017 (UTC)
@Snipre: In practice it isn't true that Wikidata doesn't enforce constraints. Preventing Redirects from getting easily added is a constraint that enforced by the software.
As far as I otherwise understand the goal in the times the development decisions live up to it, it's about allowing people to enter constraint violating data. It's would be compatible with that goal to offer a user to merge items in those cases where they provide an identifier that's already in use. ChristianKl (talk) 17:49, 22 April 2017 (UTC)
I think Magnus' Resolver tool is what you are looking for. --Denny (talk) 14:57, 22 April 2017 (UTC)
@Denny: that seems a little flaky - it's not clear what it's doing when there's no match for instance. Does it work for URL properties like official website (P856)? Doesn't seem to... I think it would be really helpful to add a simple statement search functionality to Special:Search that could find a string in any statement on a wikidata item, or filtering by specific properties. ArthurPSmith (talk) 19:01, 24 April 2017 (UTC)
  • @Snipre:I don't think that it's useful to tell a user who created a doublicated item: "You made a mistake, you should spend more time searching before creating an item" in cases where the straight search doesn't find the item. It increases the effort for the user to contribute and even if it might reduce the number of doublicate items, I don't think it's worth it.
An alternative solution could be to have a bot that writes messages to the discussion page of an item. The bot could ping the item creator and the person who added the external ID that's doublicated and ask them to check whether it's the same person and the items can be merged. ChristianKl (talk) 10:38, 23 April 2017 (UTC)

FormatterURL needs to be more flexible[edit]

FormatterURL is a URL template with a single slot $1 where the external ID is interpolated. There are many examples where this simple capability is not enough.

Eg for EU VAT number (P3608) I put FormatterURL =$1&number=$2 but that doesn't quite work: it results eg in

Another example: Wikidata:Property_proposal/WarSampo_ID where I gave up on proposing "WarSampo Place" since the site URLs use "types" and making IDs like "places/municipalities/m_place_509", "places/karelian_places/k_place_195", "pnr/P_10239521" seems like the wrong thing.

Can something be done about this?  – The preceding unsigned comment was added by Vladimir Alexiev (talk • contribs).

@Vladimir Alexiev: many examples? Looks like an edge case to me. Probably less than 10% of the properties affected, maybe even less. Maybe someone can get some hard data? The examples you give are sites that seem to have made a bit of a mess. Why should we do all sorts of work-arounds? I think we already have a (low priority) bug for this in phabricator.
In the meantime you can do a pull request on to use for the properties with funky url's. Multichill (talk) 09:59, 22 April 2017 (UTC)
@Multichill: You're right, "many" is probably an exaggeration. Don't think we can criticise sites for their URL design, eg the VIES choice to split country code and number in separate fields is wholly reasonable. How does the tool of @ArthurPSmith: help me in this case? --Vladimir Alexiev (talk) 10:08, 22 April 2017 (UTC)
There are several "solutions", currently used here:
  • create a different property for each "sector" (country in this case) - not a solution in this case, of course
  • use URL datatype
  • use "FI&number=15243611" as property value
each of these have their props and cons. Ideally I think it would be like statement with general qualifier (in this example: value - 15243611, qualifier - FI). --Edgars2007 (talk) 14:10, 22 April 2017 (UTC)
Just to note that the externalid-url tool only works with the existing wikidata formatter URL's and can take only one parameter on a per-item basis (of course you can put as many parameters as you want in on a per-property basis but that doesn't help when you need different URL's for different items). So yes it wouldn't help in this case, unless the $2 value can somehow be derived from the $1 (eg first characters, type of string, etc.) or you create an artificial combined ID that can be split somehow. We have been asking for better formatter URL's for some time now - there's T150939 and T148170 at least in ArthurPSmith (talk) 15:17, 22 April 2017 (UTC)
I think that glueing together property (FI15243611) and then using wikidata-externalid-url to get URL is another option. Actually I wanted to propose some time ago. Of course, it's not very user-friendly, but it's probably the easiest to implement. --Edgars2007 (talk) 15:38, 22 April 2017 (UTC)

Is it possible to produce a list of Wikivoyage articles sorted by the amount of articles that exists for each destination in each wikivoyage edition?[edit]

Is there any way to generate such a list? ויקיג'אנקי (talk) 20:18, 22 April 2017 (UTC)

I think the list was even available but I am not sure.--Ymblanter (talk) 22:10, 22 April 2017 (UTC)
Maybe you can ask at Wikidata:Request a query. I am not able to construct the query, but I think it is possible.--Jklamo (talk) 12:35, 23 April 2017 (UTC)
Thanks ויקיג'אנקי (talk) 18:19, 23 April 2017 (UTC)

How should we handle best player classifications?[edit]

There are currently open proposals for a bunch of properties for the best player at X under . To me the proposed properties feel like they are a bit to specific and not generalized enough. The proposal discussions itself aren't conclusive at the moment.

Does anyone have additional input? ChristianKl (talk) 08:15, 23 April 2017 (UTC)

Taxonomy and authors[edit]

Hoi, in taxonomy an author is not by his or her name but by a unique string of text. There is a property for this. In principle to describe an taxon (species, genus) properly, the author is a mandatory part. The name of the person is typically not really well known but any publication does include the abbreviation for the name. Would it be possible to only show the abbreviation when we add the author (for a taxon) property and can we please only show this abbreviation in Wikidata? Thanks, GerardM (talk) 05:17, 24 April 2017 (UTC)

If a biologist has their abbreviation, then it makes him notable for Wikidata, doesn't it? A Wikidata item can hold more information then a plain string. Matěj Suchánek (talk) 10:01, 24 April 2017 (UTC)
I know. I do not want to find him though his label but for the value that is the author information. Thanks, GerardM (talk) 11:16, 24 April 2017 (UTC)
Assuming that by "taxonomy" is meant "taxonomy of algae, fungi, and plants", there do exist "standard forms" which are recommended to be used when author citation is desired. Author citation is found in scientific works but not elsewhere (it is not "mandatory"). As things are, it is possible, when entering data, to put in the relevant "standard form" in the data field of "taxon author", and just about always the system will find the item of this taxonomist. - Brya (talk) 11:21, 24 April 2017 (UTC)
When you want to indicate a particular taxon, without author and publication (including date) there is no way that your identification suffices. The taxonomy approach at Wikidata is cutting corners, that is not really relevant here. What is relevant is how we can easily identify the author(s) associated with a specific taxon. To do that it is easiest to search by author and not by the label for the name of the person as you know. Thanks, GerardM (talk) 14:43, 24 April 2017 (UTC)
Sorry, but I do not understand what you want to achive. Could you give an example please. --Succu (talk) 15:24, 24 April 2017 (UTC)
I am sorry to hear you disagree with the International Code of Nomenclature for algae, fungi, and plants. The whole purpose of a Code of nomenclature is to make sure that a name with a particular spelling can only be used for one particular taxon (each taxon name is unique). This taxon may vary considerably in circumscription and it usually is more important to know which taxonomic viewpoint is used, rather than who published the name. - Brya (talk) 16:45, 24 April 2017 (UTC)
When a description for a taxon exists, it varies widely in what it is, what is included etc. Often subspecies, varieties change their existence and up as species or the other way around. So what makes that a taxon is not cast in stone. Consequently the name of the author and the publication are part of any description of a taxon. You could know it is part and parcel of the nomenclature. With Wikidata we intent to have an immutable identifier for a concept. Consequently as you lot have picked fixed positions that are problematic in and of itself, author and publication data are not nice to have, I understand.
As to what I want to achieve is that when I know the abbreviation for the author of a taxon, I can find him by his abbreviation and not by the Wikidata label. Thanks, GerardM (talk) 05:03, 25 April 2017 (UTC)
You are mixing nomenclatural with taxonomic issues. A circumscription of a taxon reflects the opinion of one or multiple authors and is called taxon concept. An example is the taxon labeled as Dionysia Fenzl (1843). The circumscription of this genus started with a single species Dionysia odora (Q15339763) (according to Eduard Fenzl (Q113291) in 1843 - Dionysia sec. Fenzl (1843)). In 2007 Magnus Lidén (Q5989947) revised the taxon in Dionysia sec. Lidén (2007) in which he accepted 49 species. Lidén is the author of the currently accepted taxon concept of Dionysia and Fenzl is the author who published the name Dionysia for the first time (in accordance with the rules of International Code of Nomenclature for algae, fungi, and plants (Q693148)). For the first kind of authorship we have author (P50) for the second one taxon author (P405) and ex taxon author (P697). To find an author by his abbreviation you can use the searchbox (try "Britton" or "L.") or use a SPARQL query for botanist author abbreviation (P428) or author citation (zoology) (P835), but be aware that only names regulated by International Code of Nomenclature for algae, fungi, and plants (Q693148) are more or less standardized. Seaching for a zoological author e.g. "Li" could be funny. I remeber a taxon name which was published by three different persons named Li. --Succu (talk) 15:27, 25 April 2017 (UTC)
Really.. What is it that we have in an item? Is it unique or is it not? If it is not, we need items for every permutation. As we currently have one specific outlook on taxonomy it should be obvious that an author, a publication and a publication date uniquely describe what it is we refer to. I do accept that given current assumptions we do without.
When you consider author designations, Britton often goes with Rose.. But never mind. Thanks, GerardM (talk) 17:51, 25 April 2017 (UTC)
They were coworkers and published - as you know - The Cactaceae (Q1388502) and some other works together. So what? How is this related to your question? We don't talk about math, so your term permutation (Q161519) is misplaced here. Our items represent taxon names (assumed to be unique within a Code), and apparently not taxon concepts (labeled with a subjective taxon name). We haven't „one specific outlook on taxonomy“ as you stated it. E.g. we have implemented the different views of The Angiosperm Phylogeny Group (Q539818) (APG I-IV). You refused do give an example for your concerns requested above. So I can only guess what this thread is about: „Would it be possible to only show the abbreviation when we add the author (for a taxon) property and can we please only show this abbreviation in Wikidata?“ So again what do you expect, GerardM?! --Succu (talk) 21:01, 25 April 2017 (UTC)

Another technical question[edit]

Is it possible to convert page pile # 8670, which contains only Wikidata items, to the names of the parallel page names on the English Wikivoyage? ויקיג'אנקי (talk) 05:46, 24 April 2017 (UTC)

Something like this Petscan psid=895527? Shinnin (talk) 10:02, 24 April 2017 (UTC)
@Shinnin - how would I do a similar query but for a different Wikivoyage editions (French Wikivoyage, Italian Wikivoyage, etc) ? ויקיג'אנקי (talk) 13:27, 24 April 2017 (UTC)
In the above mentioned Petscan query, go to Categories -> Language, and change the value 'en' to another language. For example, if you want to get articles from French Wikivoyage, change the language from 'en' to 'fr'. Like this. Shinnin (talk) 14:31, 24 April 2017 (UTC)

Kingdom property[edit]

A ruler is ruler when he rules some land. Do we have a property to mention the kingdom of a ruler? Please guide me. Capankajsmilyo (talk) 06:33, 24 April 2017 (UTC)

Take a look at George V (Q269412). It uses position held (P39) with qualifiers. ChristianKl (talk) 07:37, 24 April 2017 (UTC)
Thanks Capankajsmilyo (talk) 09:27, 24 April 2017 (UTC)

Reverse fetching?[edit]

Does Wikidata do reverse fetching? I found a lot of cases where father of a person was given but he/she was not listed as child of the said father. Is there a bot or something, or it all has to be done manually? Capankajsmilyo (talk) 09:26, 24 April 2017 (UTC)

Yes, this is mostly done by bots but they usually require human oversight to prevent reproducing mistakes. Matěj Suchánek (talk) 09:59, 24 April 2017 (UTC)
Are the bots disabled? I rarely find such operations being carried out. Capankajsmilyo (talk) 12:48, 24 April 2017 (UTC)
I would say they don't run on items about people as they are sensitive against invalid data. I don't remember now whether there were some agreements in the past. Other inverses are completed very often though. Matěj Suchánek (talk) 13:06, 24 April 2017 (UTC)
Ok, so I have been working today on reducing missing inverse statements. I made the following tasks:
  1. Add inverse spouse (P26) where the persons had opposite gender and at least one common child (P40).
  2. Add inverse sibling (P3373) where the persons had at least one same parent.
I'm currently preparing child (P40) import based on father (P22) and mother (P25) secured by spouse (P26). Matěj Suchánek (talk) 20:10, 24 April 2017 (UTC)
That's awesome. Thanks. Capankajsmilyo (talk) 20:46, 24 April 2017 (UTC)

first line (P1922) & last line (P3132)[edit]

These properties will use in film, music works, television series and anime labels? --Nakare✝ (talk) 09:27, 24 April 2017 (UTC)

@นคเรศ: If following the trail of subclass of (P279) from each of film (Q11424), TV series episode (Q21191270), and song (Q7366) leads back to work (Q386724) (as required by one of first line (P1922)'s constraints, then it is fine to include either of those properties in all of those things. (I would highly suggest, however, that you ask about what constitutes the 'first line' of any of those things before deciding to add those properties.) Mahir256 (talk) 05:42, 25 April 2017 (UTC)

Wikidata weekly summary #257[edit]


Is there a property to indicate the "anniversary" of a subject? I am looking for it for military units, for example most Canadian regiments have a day in the year that they celebrate as their anniversaries. But it could be used by other types of subjects as well. The Infobox military unit on English Wikipedia has a parameter for "anniversaries". Thanks, Amqui (talk) 13:31, 24 April 2017 (UTC)

Can I suggest annual commemoration Q18574943 (or awareness day Q422695 if appropriate). MassiveEartha (talk) 14:03, 24 April 2017 (UTC)
Those are not properties. Amqui (talk) 15:33, 24 April 2017 (UTC)
my mistake, I misread the question. MassiveEartha (talk) 15:43, 24 April 2017 (UTC)
(edit conflict) Like significant event (P793): annual commemoration (Q18574943); qualifier day in year for periodic occurrence (P837)? Matěj Suchánek (talk) 15:45, 24 April 2017 (UTC)
I thought significant event (P793) was only for past events, not recurring events, but reading the description and the constraints, I guess nothing refrain that. All examples given are for past "one-time" events, but, if that's the case, I will add one example that shows a recurrent annual event as an example in significant event (P793). Also, if that's the case, I added public holiday (P832) as a subproperty of significant event (P793). Thanks, Amqui (talk) 16:38, 24 April 2017 (UTC)
I meant just as a suggestion, otherwise a new property for "recurring event" should be created. Matěj Suchánek (talk) 17:45, 24 April 2017 (UTC)

Gendered Given Names[edit]

What do I do if I have a female Audley Q26986303? Why are given names gendered? MassiveEartha (talk) 14:05, 24 April 2017 (UTC)

Create a new item for the female version of the name. ChristianKl (talk) 14:09, 24 April 2017 (UTC)
Hello MassiveEartha
in fact, there are 2 possibilities :
  1. the name is often given to men and women , and you can change the first name item from male given name (Q12308941) to unisex given name (Q3409032) (and correct the descriptions)
  2. the name is really a male name, but has been exceptionnally used for a woman. You can then use the male name, and add a qualifier on the given name (P735) property : instance of (P31) = male given name borne by a female (Q18220911)
Do not create a second item for the female form of the name, except if the origin/language of the name is really not the same, like Jean (male in fr) / Jean (female in en). --Hsarrazin (talk) 19:22, 24 April 2017 (UTC)
I added precisions to your message @Hsarrazin: --Harmonia Amanda (talk) 19:24, 24 April 2017 (UTC)

Thanks for the advice - I've already created a female give name item - will try to delete/merge to align with one of the options above. MassiveEartha (talk) 19:55, 24 April 2017 (UTC)

@MassiveEartha: You may merge the "female" version to the "male" one if it is not needed anymore. -- LaddΩ chat ;) 21:44, 24 April 2017 (UTC)


The new TLG author ID (P3576) was implemented with only a single support. The TLG database is paywalled; none of the data is accessible without an expensive subscription. As a result, the links return an error message for anyone except the few people who have a subscription. Including the database here creates a divide that is contrary to the goals of MW. --EncycloPetey (talk) 16:19, 24 April 2017 (UTC)

this was raised during the property proposal discussion (see the link to the proposal on the discussion page for the property). If we are concerned about this in general perhaps a policy proposal with an RFC on this subject would be a good idea? ArthurPSmith (talk) 18:52, 24 April 2017 (UTC)
@EncycloPetey, ArthurPSmith: There is an alternative to linking TLG. The numbers are used outside of the TLG, e.g. by the Perseus Project which uses the TLG IDs in their canonical URNs. The catalogue entry for Aristotle, for example, is
I suggest we change the formatter URL to the Perseus catalogue URN, and keep the original TLG resolver – but as a deprecated value, so it doesn't get transcluded.What do you think? Jonathan Groß (talk) 20:01, 24 April 2017 (UTC)
changing the formatter to a third party resolver is fine with me. ArthurPSmith (talk) 20:03, 24 April 2017 (UTC)
That sounds like a better solution to me as well. --EncycloPetey (talk) 20:48, 24 April 2017 (UTC)
I agree with changing the resolver to the third party resolver. ChristianKl (talk) 20:50, 24 April 2017 (UTC)
@ArthurPSmith: RE: "If we are concerned about this in general perhaps a policy proposal with an RFC on this subject would be a good idea?" I haven't seen comments posted concerning this point. I think this is an area of general concern, but if no one agrees, then it may not be worth the effort of running an RFC. Full comments and reasons are not needed at this time, just a simple indication of desire to see the issue discussed. --EncycloPetey (talk) 21:23, 24 April 2017 (UTC)
There was some discussion going on at Wikidata:Requests for deletions/Archive/2017/Properties/1#P2690 and Wikidata:Property proposal/Church of Sweden place ID. --Pasleim (talk) 21:43, 24 April 2017 (UTC)
I do think there are differences in opinion between different users about whether the fact that an ID is paywalled provides sufficient grounds for not including the ID. Having a PDF to form a consensus would have value. ChristianKl (talk) 09:28, 25 April 2017 (UTC)


I thought it would be good to merge Exmoor National Park (Q29511798) and Exmoor (Q593627) about Exmoor National Park and Exmoor. According to the different wikipedia language versions the area of both is the same, and the English version does not have a separate page about the National Park (and the languages with an article about the national park don't have a separate article about Exmoor). However I get an error and it was suggested to ask here about this appararent problem. Maybe there are good reasons not to merge. Thanks, Hobbema (talk) 21:01, 24 April 2017 (UTC)

@Hobbema: Your error most likely is caused by Exmoor (Q593627) being located in protected area (P3018) Exmoor National Park (Q29511798). If you delete that statement, the merge should go through. (You may want to clean up the statements in the resulting item afterwards.) Also bear in mind the templates I am using in this comment and which I have added to yours. Mahir256 (talk) 05:31, 25 April 2017 (UTC)
@Hobbema:I don't think merging the two things is a good idea - a National park is not the same as a landscape area, not all of the landscape area is part of the National Park, and for example the property inception (P571) for the NP is ridiculous for the landscape. The fact that currently no Wikipedia separates the two topics doesn't mean those different items should be munged together here. Ahoerstemeier (talk) 09:24, 25 April 2017 (UTC)
I am against the merge for the same reason. Not everything that applies to a park applies to a landscape. Thierry Caro (talk) 09:33, 25 April 2017 (UTC)
yes, please do not merge. rather split english article into two separate articles. Slowking4 (talk) 00:21, 26 April 2017 (UTC)

Crossing keeper or Railway keeper[edit]

I think that every language misses this profession:

commons:File:Yogyakarta Indonesia Crossing-keeper-house-at-Tugu-Railway-Station-01.jpg d1g (talk) 21:22, 24 April 2017 (UTC)

@D1gggg: Is this signalman (Q1455706) what you're looking for? --EncycloPetey (talk) 21:26, 24 April 2017 (UTC)
I think that pictured person is in charge of Level crossing than signalling control
Level crossings are still semi-automated or need human intervention (at least in some places).
I don't know a real job title. d1g (talk) 21:51, 24 April 2017 (UTC)
"Crossing keeper" is the standard job title in British English at least. Sometimes the roll might have been combined with that of signaller, especially when the level crossing is adjacent to a signal box. For automatic crossings the job role doesn't exist. For modern remotely operated crossings it is almost always done by a signaller and operating the crossing will be fully integrated with that role such that it will be seen as an integral part of the duties of the signaller. Thryduulf (talk) 07:54, 27 April 2017 (UTC)


As I am designing the main page for Marathi Wikipedia I wanted to add Commons motd and potd to it. The code for potd is {{#property:P18|from=Q14334596}} and I need such for motd please provide me with the code as I am new to wiki data thanking you --Tiven2240 (talk) 03:19, 25 April 2017 (UTC)

@Tiven2240: I understand that "potd" stands for Picture of the Day but what does "motd" stand for? By the way, you can also try {{#statements:P18|from=Q14334596}}. Matěj Suchánek (talk) 09:48, 25 April 2017 (UTC)
@Matěj Suchánek: It is Media of the day. @Tiven2240: we have Commons:Media of the day (Q16939274). Is it what you are looking for?--Micru (talk) 09:57, 25 April 2017 (UTC)
@Matěj Suchánek:@Micru: yes it is Commons media of the day motd can I have that code by metadata on local Wikipedia as I am using the code of potd from metadata on local Wikipedia?--Tiven2240 (talk) 10:41, 25 April 2017 (UTC)
As you can see, the item links to no media, so the above code won't work. Anyway, it looks like a very complex task since we have many media properties. Matěj Suchánek (talk) 18:56, 25 April 2017 (UTC)
we need an item for c:Template:Motd, just like we have Template:Pic of the day (Q14334596) for c:Template:Potd. there is a conflict with w:Template:Motd as well. -- Slowking4 (talk) 00:16, 26 April 2017 (UTC)

@Slowking4: @Matěj Suchánek: so isn't there any solution to the problem --Tiven2240 (talk) 05:35, 26 April 2017 (UTC)

Why an Wikipedia article doesn't match Wikidata entry?[edit]

I'm unable to match (Portuguese Wikipedia entry) with (Wikidata entry). Is it possible to do it? Thanks, Carlos Luis M C da Cruz (talk) 14:21, 25 April 2017 (UTC)

Carlos Luis M C da Cruz: Done. --Succu (talk) 14:27, 25 April 2017 (UTC)
Many thanks! Carlos Luis M C da Cruz (talk) 18:02, 26 April 2017 (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. Matěj Suchánek (talk) 07:18, 28 April 2017 (UTC)

List of tasks that are currently regularly done by Bots[edit]

Currently, it feels hard to me to know which tasks are regularly done by bots and which bots do those tasks. Do we have any list besides the approval process? ChristianKl (talk) 15:09, 25 April 2017 (UTC)

Just created for my BOT. --ValterVB (talk) 16:19, 25 April 2017 (UTC)
Not that I know. Actually there's unlimited nubmer of tasks bots can do, given that we have tools like PetScan, QuickStatements or Pywikibot. So perhaps we should first put together the "core" tasks (whatever that means) and then search. Matěj Suchánek (talk) 18:52, 25 April 2017 (UTC)
What do you mean with „regularly“? Triggered by a scheduler? --Succu (talk) 19:35, 25 April 2017 (UTC)
Pictogram voting comment.svg Comment @Matěj Suchánek: I see that petscan and qs are done by humans rather than by bots, though they are tools-assisted. I was presuming that ChristianKl was meaning by accounts with bot permissions as it was about permissions.  — billinghurst sDrewth 01:35, 26 April 2017 (UTC)
hundreds of categories have the implicit information to add missing statements to Wikidata. They are defined by "is a list of" and have a qualifier expressing the statement (eg "educated at" "University of Amsterdam"). When this is picked up by a bot the first run will add tens of thousands if not hundred thousand and more statements. Thanks, GerardM (talk) 05:08, 26 April 2017 (UTC)
@ChristianKl: there's a list of bots by edit count here that maybe helps. You can see their recent activities by clicking the 'c' link on each bot. Many bot operators post what the bot is doing on the main page also. There's only 50 bots with over 1 million edits, so you could get a rough idea what the regular tasks are by reviewing that group. Not a comprehensive list of bot activity though. ArthurPSmith (talk) 13:45, 26 April 2017 (UTC)
In this particular case I want to understand which bots do tasks related to properties. I know that some statements are automatically copied from the proposal discussion template and I have seen some bot deletes dublicate information from the template but I'm for example not sure about whether there a bot that copies information from constraint statements into constraint templates on the talk page. If there would be a clear list of what tasks are done by bot in this case that would decrease the total human workload that's involved.
I can imagine a similar concern when it comes to creating many inverse properties. If the bot is doing it anyway there's human labour to be saved if the human knows what's done by bot. Maybe it can be listed somewhere on the property discussion page of a property like sibling (P3373) if there's a bot that creates the inverse statements? ChristianKl (talk) 06:48, 27 April 2017 (UTC)

biologically motion (Q29373587) and locomotion (Q22269718)[edit]


I think that biological process (Q2996394) is correct: covers both individuals and populations.

Therefore item Q22269718 represented both individuals and groups.

The most extreme case is when swarming animals (e.g. bees or ants) behave individually or as group.

If we are about to change classification, we should start with Q2996394 and remove link to Q2625603 IMO. d1g (talk) 21:15, 25 April 2017 (UTC)

About biological process (Q2996394), description: "collection of molecular events ...". So, description of biological process (Q2996394) is: 1) "process...within the individual" or 2) "process...within the individual + between individuals"? --Fractaler (talk) 09:58, 26 April 2017 (UTC)
Probably some sitelinks should be moved from animal locomotion (Q925958) (treated as animal locomotion) to locomotion (Q22269718) or both should be merged and a new item for animal locomotion should be created. --Succu (talk) 13:38, 26 April 2017 (UTC)
Wikidata isn't ready yet to describe "between individuals" part in detail. We need participants property at very lest to describe processes.
My suggestion is to use 1. one item for individuals and 2. other for groups.
Herding information can be very complex and not well-studied as processes within one organism.
Movement information belongs to individuals for the most part, other part is rather complex or specific (kangaroo inside pouch, a parasite within animals) d1g (talk) 04:59, 28 April 2017 (UTC)
Yes, we have individual and group (of individuals). But what is individual? Superorganism (Q916139) is group (of individuals)? Consist of individuals, =individual? Of course, "herding step" is not same as "cattle (Q830)'s step". {{|QQ2918584}}? About parasite+animals: it is a passive moving. --Fractaler (talk) 10:41, 28 April 2017 (UTC)
I'm not sure what biologically motion (Q29373587) is meant to cover, since "biologically motion" is not even a grammatical phrase in English. Is the item intended to cover any kind of movement associated with a biological origin (such as blinking, the unfolding of flowers), or is the item concerned only with movements that change the location of an individual, or only movements initiated by an individual to change its location, or only movements initiated by an individual to change its location in response to specific stimuli? All of these have different biomechanical and physiological terminology applied to them. --EncycloPetey (talk) 21:18, 26 April 2017 (UTC)
physical model (Q12149837): linear motion (Q2035846). For organism (Q7239) we have: active motion (Q29374366)/passive motion (Q29374520) --Fractaler (talk) 14:16, 27 April 2017 (UTC)

We also have a travel (Q61509) "travel" "movement of people between relatively distant geographical locations" as subclasses of Q79782 and Q1914636. d1g (talk) 04:35, 28 April 2017 (UTC)

Tele-tourism/tele-travel, three-dimensional virtual tourism (Q7797256), virtual tour (Q2915546), etc. are "travel"? --Fractaler (talk) 08:12, 28 April 2017 (UTC)

Means to search for whole/lead word only? not in a phrase[edit]

Does anyone know of a means to utilise local search to set the whole/lead/beginning word? I am trying to do a whole word search and only want the whole word, not elsewhere in the description? I am trying to replicate the "typeahead" function but as a search result. I need this as I am doing a lot of given name and family name additions, and the expanded search result can be quite problematic when searching for a simple name. I cannot always rely on the typeahead search as lists can be long for simple terms, and the description is not always suitable for ready finding. Due to our items and the fact that embedded items are not searchable the ability to search "SEARCHNAME name" or "SEARCHNAME Q101352" etc. all fail.

mw:Help:CirrusSearch gives lots of good tips, but in this sort of query it isn't helpful, and less useful again when at Wikidata. Thanks if anyone has any ideas here.  — billinghurst sDrewth 01:26, 26 April 2017 (UTC)


At GEO600 (Q697906), it would be very useful to be able to define a range of wavelength (P2808) that it works at, which in this case is 43-10,000 km. Is there a good way to do this? At the moment I've just been setting the minimum and maximum values separately for the wavelength property, but that's not good as it looks like two distinct wavelengths rather than a range of wavelengths. Thanks. Mike Peel (talk) 15:09, 26 April 2017 (UTC)

@Mike Peel: The data format underlying Quantity values allows for a range (lower bound, upper bound) but the UI doesn't display it well in a case like this - it would show as 5022 +- 4979 km I believe. I special range datatype might be a very good idea... ArthurPSmith (talk) 15:35, 26 April 2017 (UTC)
@ArthurPSmith: Good point! Lower/upper bounds aren't quite the same thing as a range, though, since (at least in scientific measurements) the bounds are normally due to an uncertainty in the measurement, while the range here is more absolute. Maybe a new datatype is the way to go, but surely this issue has come up before somewhere? Thanks. Mike Peel (talk) 23:29, 26 April 2017 (UTC)
it has come up here and there but I'm not aware of a major discussion or proposal or anything. Example of recent previous discussion along these lines (though perhaps more uncertainty-related than range-related) is here. ArthurPSmith (talk) 13:51, 27 April 2017 (UTC)
@Lydia Pintscher (WMDE): Any suggestions of how to tackle this? Thanks. Mike Peel (talk) 00:33, 28 April 2017 (UTC)
Right now the best we have is to create 2 properties or use a qualifier like "instance of: minimum value". Neithere of them is particularly pretty but the "proper" solution isn't trivial and I don't think we'll be able to invest time into it anytime soon unfortunately. --Lydia Pintscher (WMDE) (talk) 16:00, 28 April 2017 (UTC)


Can someone explain me how to edit coordinates on Wikidata? Sweetcorn (talk) 16:14, 26 April 2017 (UTC)

When you're on item's page go to the "statements" section, find there a property called "coordinates", and then click "edit" next to it. There you can edit coordinates. Stryn (talk) 17:28, 26 April 2017 (UTC)
It's quite funny, I didn't know it was that easy simply for the reason that there is no "Edit" tab next to "Coordinates" property of my statement section. I don't know if this action is just not allowed in my Wikidata account or while I am not logged in or I cannot access this feature due to the fact that I am on a mobile device. You solved the problem for me on that item so thank you. Sweetcorn (talk)

What happened to P31?[edit]

Why was Property:P31 changed to "example of" instead of "instance of"? (at least in other languages than English). --Superchilum(talk to me!) 08:17, 27 April 2017 (UTC)

I have reverted these changes. Fortunately the Property:P31 is at least already semi-protected. But I am not sure about other languages. Simply the Property:P31 vs Property:P279 concept is not easy to understand.--Jklamo (talk) 08:37, 27 April 2017 (UTC)

The strategy discussion. The Cycle 2 will start on May 5[edit]

The first cycle of the Wikimedia movement strategy process recently concluded. During that period, we were discussing the main directions for the Wikimedia movement over the next 15 years. There are more than 1500 summary statements collected from the various communities, but only 3 from the Wikidata discussion. The strategy facilitators and many volunteers have summarized the discussions of the previous month. A quantitative analysis of the statements will be posted on Meta for translation this week, alongside the report from the Berlin conference.

The second cycle will begin soon. It's set to begin on May 5 and run until May 31. During that period, you will be invited to dive into the main topics that emerged in the first cycle, discuss what they mean, which ones are the most important and why, and what their practical implications are. This work will be informed and complemented by research involving new voices that haven’t traditionally been included in strategy discussions, like readers, partners, and experts. Together, we will begin to make sense of all this information and organize it into a meaningful guiding document, which we will all collectively refine during the third and last cycle in June−July.

We want to help your community to be more engaged with the discussions in the next cycle. Now, we are looking for volunteers who could

  • tell us where to announce the start of the Cycle 2, and how to do that, so we could be sure the majority of your community is informed and has a chance to feel committed, and
  • facilitate the Cycle 2 discussions here, on Wikidata.

We are looking forward to your feedback!

Base (WMF) and SGrabarczuk (WMF) (talk) 08:47, 27 April 2017 (UTC)

Thanks for this post. While there was not much input from wikidata on this so far, I would recommend that all wikidata-interested folks look at the summary linked above - wikidata appears very frequently (usually positively) in the strategy suggestions from our sister projects. I think it behooves us to think about what that means for wikidata in the coming years. ArthurPSmith (talk) 14:22, 27 April 2017 (UTC)
Sure, think and share the thoughts, and the best place to do that is here, on Wikidata talk:Strategy 2017 :) SGrabarczuk (WMF) (talk) 19:53, 27 April 2017 (UTC)
Seriously, SGrabarczuk (WMF): „What role do we want to play in the world in 2030?“ Sounds to me like hubris (Q735766). Something like the question what a role fullfills a modern mobile phone (Q17517) produced around the last no label (Q1678709) fifteen years later. I think currently Wikidata has some impact on linked open data (Q18692990), but we are far away from modelling a ontology (Q324254) around that data. Supporting the last is more important to me than fabulating about fast evolving technologies. --Succu (talk) 20:38, 27 April 2017 (UTC)
@Succu: this is the movement strategy process, so we're focusing on the broadest directions. Of course we don't know what exactly happens in 15 years, but we can discuss e.g. how to increase the base of potential Wikidata users (including re-users), whether Wikidata should stay under Wikimedia umbrella, what role should Wikidata fulfill and what shouldn't within our movement, anything broad and abstract enough, anything that could be applicable to / cover 15 years. Please have a look at the summary of 1500 summaries linked above. Or, should I actually write, we could have discussed that, because now, there's going to be the Cycle 2, when we focus on 5 out of all those ideas. So again, please have a look at the links. SGrabarczuk (WMF) (talk) 20:51, 27 April 2017 (UTC)
No, SGrabarczuk (WMF), it's a WMF request, not something the „movement“ initialised as your answer states. Why should I'm interested in the „summary of 1500 summaries linked above“. Repeating strategy headlines is far away from something I would call a strategic planning (Q932522), but ignorance (Q815577) about the current state of Wikidata. --Succu (talk) 21:26, 27 April 2017 (UTC)
A year ago, in Esino Lario, Wikimania attendees were heavily interested in a broad strategy. Only a part of them were WMF staff. WMF is organizing it, but I doubt if this is really by WMF request (by the way, who is WMF in this case?). SGrabarczuk (WMF) (talk) 21:34, 27 April 2017 (UTC)
Not sure why people are objecting to this - WMF may be facilitating, but they are asking the community for direction here, this is important. ArthurPSmith (talk) 14:58, 28 April 2017 (UTC)
Probably I'm suffering on some kind of marketing allergy. --Succu (talk) 21:57, 28 April 2017 (UTC)

How to add the url and title of a reference to a web page via QuickStatements?[edit]

I want to add external ids derived from a mapping with the QuickStatements tool, and to make that more traceable, I'd like to add a reference to the statements, with URL and title of the web page about the mapping.

I meant to follow the help page when I tried:

Q368982    P227    "170060039"     S854    "" S1476   "Via P2428 lookup derived from ZBW's RAS-GND authors mapping"

The statement is executed and a reference created (example), but the title (P1476) part is lost. Which mistake did I make? -- Jneubert (talk) 11:29, 27 April 2017 (UTC)

Wikidata:Project chat/Archive/2017/04#QuickStatements - references: For a monolingual string, prefix it with the language and a colon, e.g. en:"Some text. Matěj Suchánek (talk) 14:57, 27 April 2017 (UTC)
Thanks for the explanation! However, the statement
Q368982        P227    "170060039"     S854    "" S1476   en:"Via P2428 lookup derived from ZBW's RAS-GND authors mapping"
still does not create a title. Is there something else I could have missed? Different endpoint? (I used this) Different user rights? -- Jneubert (talk) 15:22, 27 April 2017 (UTC)
If I remember correctly, the old endpoint, which you are using, had problems with adding multiple properties as a single reference. Instead, you can try the new version (under development) or a fork. Matěj Suchánek (talk) 15:30, 27 April 2017 (UTC)
It worked with the new version (not with the fork) - thanks! However, there are some - probably outdated, but drastic - warnings on the tool's talk page ("it might nuke Wikidata ...") Would you consider it safe to use it with a rather constant form of statements, with a batch size of, e.g., 100? -- Jneubert (talk) 16:45, 27 April 2017 (UTC)
Personally, I only use the new one if I need to use QS. The only problem I sometimes have is getting stuck on certain items (I found out it's caused by PHP fatals, so it cannot be considered stable yet). So yes, smaller batches with oversight should be okay. Matěj Suchánek (talk) 07:18, 28 April 2017 (UTC)
Worked great for ~450 added external ids - thanks, again, for all your help! -- Jneubert (talk) 10:45, 28 April 2017 (UTC)

Script and API module for constraint checks[edit]

Hello there,

In the past few months, the development team has mentored a student, Olga, to help us developing a user script that displays the constraints violations on the item pages.

To use the script, add the following line to your user/common.js:

mw.loader.load( '//' );

You will see an icon Icon used on the user script for constraint reports on Wikidata next to violations. When you click it you will see the full report.

Screenshot of the user script checking constraints on Wikidata

This script is based on a new API module for constraint checks that one can use to check constraints on items and statements. At the moment, the constraint checks are only derived from the property discussion page constraint templates, not directly from statements. They are then stored in a database table. We are running a script to update this table every now and then or when you ask for it. Also note that some constraint checks are disabled (for example the format check). In the future we will support adding and updating constraints on property statements and we will implement support for some constraints that are currently still missing.

If you try it, feel free to give us feedback! You can also add comments or subtasks on Phabricator (see the ticket for the API module and the user script).

If there is no major disagreement, we would like to turn this script into a gadget in the next days.

Thanks go to Olga and all the developers that helped her providing this new feature :)

See also:

Cheers, Lea Lacroix (WMDE) (talk) 12:45, 27 April 2017 (UTC)

Seems promising but it doesn't work for me. On Saint-François-de-Sales Church (Q962691) for example, I see no icon for EgliseInfo ID (P1644), while there is two values... — Ayack (talk) 21:11, 27 April 2017 (UTC)
Some bugs:
  • Qualifier constraint violation icons are placed on every qualifier of any statement that has any violating qualifier, rather than just the particular violating qualifier itself. This is very confusing.
  • "Target required claim" messages don't say what the missing statement is, "Qualifiers" messages don't mention which qualifiers are accepted or prohibited, "Range" messages don't say the minimum/maximum, and "Value type" messages don't say what value type is required.
  • Some properties such as P570 have the text "⧼wbqc-violation-message-range-parameters-needed⧽" shown as the message.
  • Upon clicking "edit" for a statement and then clicking "cancel", qualifier-level violations disappear but statement-level violations remain.
  • The "Value only" constraint does not work.
--Yair rand (talk) 21:23, 27 April 2017 (UTC)
@Ayack: Still no icon for EgliseInfo ID (P1644)? I see one now.
@Yair rand:. Thank you very much for your reports. Do you have some examples for your points #1 and #5? About the content of the messages, we're currently working on it to improve them. Lea Lacroix (WMDE) (talk) 06:28, 28 April 2017 (UTC)
@Lea Lacroix (WMDE): The script doesn't work when the "Drag'n'drop" gadget is selected. When I deactivate it I see the icon. — Ayack (talk) 08:59, 28 April 2017 (UTC)


This is a strange item with 37 backlinks but all sitelinks deleted. Does anybody know what is going on? The item is about taxonomy. Thanks.--Ymblanter (talk) 18:03, 27 April 2017 (UTC)

I guess it is a taxonomic concept that mainly exists in botany (section (Q10861426)), but that is rarely (or only historically) used in zoology.--Pharos (talk) 20:18, 27 April 2017 (UTC)

What is king?[edit]

Is it an occupation, or position held, or both?  – The preceding unsigned comment was added by User:Capankajsmilyo (talk • contribs)..

Henry II of England (Q102140) position held (P39) Monarch of England (Q18810062). But we could also say Henry II of England (Q102140) occupation (P106) ruler (Q1097498) Runner1928 (talk) 21:31, 27 April 2017 (UTC)
a position held. A coup d'etat would remove you from the position.  — billinghurst sDrewth 09:04, 28 April 2017 (UTC)

Wikidata Query Service + Mediawiki API = Love[edit]


I am developing functionality that will allow WDQS query to get data from Mediawiki API (See phab:T148245).

Currently, the design is as follows:

  • The API should have a pre-defined template
  • The template also specifies which results are available from the API

The need for template is currently because we need to convert data from API's treelike format to tabular format that SPARQL needs, and the template allows to specify how the conversion is done. See d:Wikidata:WDQS and Mediawiki API for detailed description of how it works.

The prototype implementation is running on (only Categories API described in the page above is configured now, but more will be soon). I'd like to hear feedback about this:

  • does template model make sense at all? Is it enough?
  • what APIs would we want to expose?
  • any other features that would be useful?

Other comments and ideas on the matter are of course always welcome. Please comment on the talk page or here.

--Smalyshev (WMF) (talk) 21:37, 27 April 2017 (UTC)

Wow. Very cool.
Some potentially useful APIs: prop= contributors / linkshere / pageviews / links / pageassessments / wbentityusage, prop=info&inprop=displaytitle, and something to find when a page was last edited if that exists.
It would be helpful for the results to be more closely associated to the linked Wikidata items. The example given at the bottom of the page relies on Wikipedia titles being the same as the labels, which isn't always the case, so the results can be inaccurate. Maybe have a bd:serviceParam mwapi:linkeditem ?item . option. --Yair rand (talk) 23:26, 27 April 2017 (UTC)
Yes, the example uses labels, because test host doesn't have sitelinks :) (for performance reasons). Proper query would use sitelink title, which we do have, but only in production data. These all will be worked out, but the point about enabling exctracting wikidata ID is imporant, thanks for noting it. I'll think about how to make it work. --Smalyshev (WMF) (talk) 19:38, 28 April 2017 (UTC)

unlink pages with lua errors[edit]

Half of the page Wikidata:List of properties/all is covered with lua errors, I suggest to use Wikidata:Properties instead.

Links can be found at prominient locations e.g. Help:Statements (section #Properties) d1g (talk) 06:57, 28 April 2017 (UTC)

Substitute the subpages then transclude to one. Have a process to build the properties on a daily basis. The page is useful (when it works) for doing a keyword find rather than trying to manipulate search.  — billinghurst sDrewth 09:02, 28 April 2017 (UTC)
There are two issues with Wikidata:List of properties/all:
  1. Lua errors: substituting will make the page only accessible in English
  2. The list has to be manually maintained. I count currently only 967 properties in the list, so around 70% of all properties are missing. --Pasleim (talk) 09:49, 28 April 2017 (UTC)
I Symbol support vote.svg Support removing it - link to the search page with Property selected instead. ArthurPSmith (talk) 12:39, 28 April 2017 (UTC)

Languages system does not work[edit] is disallowed to link to German as already does. Obviously this restriction does not work in the real world. So you English folks either need to reduce your vocabulary to match all other languages in order to make language references work... or just allow multiple mappings!

The intended and designed solution for this is to just keep the interwiki link in those rare cases where multiple articles from one language want to link to a single article in another language that does not link back to it. There is no need for any language to change in order to fit into Wikidata :) --Denny (talk) 15:55, 28 April 2017 (UTC)

Looking for a tool[edit]

Moin Moin together, i'm looking for a tool, but forgot the name. There was a list of all wikilanguages, in this list there were all articles without a Wikidata-item and from the tool you could generate a new Wikidata-item. Could somebody give me advice? Special thanks --Crazy1880 (talk) 17:46, 28 April 2017 (UTC)

@Crazy1880: This, maybe? Mahir256 (talk) 17:57, 28 April 2017 (UTC)
Dear Mahir256, yes, great, that is exactly this tool I was looking for. Saved immediately. Big thanks --Crazy1880 (talk) 18:31, 28 April 2017 (UTC)

How to asign a Property to an already created Mix'n'match catalog[edit]

Does anybody know how to add a property to an existing catalog? For the catalog, I want to asign the Property:P3887. Thanks. --Gerwoman (talk) 18:03, 28 April 2017 (UTC)

I think only Magnus Manske can do that for existing catalogs. Mbch331 (talk) 18:49, 28 April 2017 (UTC)

Property for the archives/collected papers of a politician or historical figure?[edit]

Hi, do we have a property for identifying a URL or institution that holds a person's collected papers or archives? Has anyone modeled this yet for a person in Wikidata? For example, in the United States, oftentimes politicians/members of Congress donate their papers to universities, libraries or to the Library of Congress, but it is inconsistent. It would be useful to track this in Wikidata. I imagine that this property would be somewhat like "award received" or "educated at" in that it may have multiple statements, with qualifiers. Sometimes the collection is online at a URL we can specify, but sometimes it is offline, and we can only point to the institution and a "finding aid" document, which is an inventory of what the physical collection looks like. Thanks for any leads/advice on this. -- Fuzheado (talk) 21:03, 28 April 2017 (UTC)

@Fuzheado: I always use archives at (P485) with qualifier inventory number (P217) and reference reference URL (P854). Jonathan Groß (talk) 21:24, 28 April 2017 (UTC)
Thanks much I'll give it a try and see how to specify finding aid. -- Fuzheado (talk) 21:31, 28 April 2017 (UTC)