Shortcut: WD:PC

Wikidata:Project chat

From Wikidata
(Redirected from Wikidata:Project Chat)
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.
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 2016/08.
Filing cabinet icon.svg
SpBot archives all sections tagged with {{Section resolved|1=~~~~}} after 1 day.





for permissions


for deletions


for deletion

for comment

and imports

a query



density vs. population density[edit]

I've noticed that density (P2054) has been used for "population density" - e.g. Rotterdam (Q34370), Rome (Q220), Lisbon (Q597), etc. Moreover, even worse, it is being used without any unit, so even if one accepted the wrong idea that density (P2054) can mean that, these numbers are still useless as it's not clear what they actually measure. I suspect most of those are produced by User:Titanopedia, but didn't check it.

So, should we have population density property? Should we keep those items around before the property is created and then migrate them? --Laboramus (talk) 08:16, 12 August 2016 (UTC)

A claim without unit and without source like that in Lisbon#P2054 is not worth migrating! -- Innocent bystander (talk) 14:06, 12 August 2016 (UTC)
Another thing: "population density" maybe is related to such a thing as "density" in English. But they are not at all related in other languages. If somebody only familiar with Swedish reads "density":1234 for UK they will finally understand why they meassure weight in "Stones". -- Innocent bystander (talk) 06:11, 13 August 2016 (UTC)
@Laboramus, Innocent bystander: I don't think we need a property for population density (Q22856). I mean, we already have population (P1082) and area (P2046), did we really need a third property who is just the division of the first two ? The only case I can see is if the source only give the density and not the population and the area, but it seems really very unlikely (or only if the source is bad, and then it wouldn't be a good idea to use it).
Indeed beware langages, I was puzled at first as in French the « densité » *can't* have a unit (as in French, « densité » is a false-friend for relative density (Q11027905)).
Cdlt, VIGNERON (talk) 16:49, 17 August 2016 (UTC)
@VIGNERON: I am not so sure that it is as simple as "is just the division of the first two". pop-density can be calculated both on the total area and sometimes on the land area. I am currently adding data about Swedish urban areas. The area of these were started to be reported in 1980 and both land and water area was then reported. Since 1990 only land area is measured and the water area is now not regarded as a part of the entity at all. For Swedish Municipalities there are four different areas reported, "land", "sea water", "water in the four great lakes" and "other lakes and watercourses waters". -- Innocent bystander (talk) 16:13, 19 August 2016 (UTC)
@Innocent bystander: sure, you need to to the right division with the right numbers (obvisouly, you don't divide either the 2010 population by the current area if there was a different area in 2010) but still, it's just a division, isn't it ? Cdlt, VIGNERON (talk) 18:19, 22 August 2016 (UTC)
Yes, in our templates at svwiki, we normally let the template do the division. The tricky part is when the latest updates of population isn't of the same date as the area. -- Innocent bystander (talk) 06:47, 23 August 2016 (UTC)
Wikidata:Property proposal/population density is currently open, you may wish to comment there. Robevans123 (talk) 10:33, 23 August 2016 (UTC)

Contribute to set a data quality framework for Wikidata[edit]

Dear Wikidata members, We are working on setting a data quality framework for Wikidata, as part of a research project carried out by members of the Web and Internet Science group of the University of Southampton.

Determining the quality of Wikidata is crucial for its future development. We believe that its community should have a primary role in defining what data quality means in Wikidata. Therefore, we would like to ask community members to contribute to our data quality framework draft by adding comments, suggestions, and concrete example of quality issues on Wikidata.

The draft has been published as a Request for Comment and can be found at this address: Data quality framework for Wikidata
Many thanks,
--Alessandro Piscopo (talk) 08:44, 12 August 2016 (UTC)

Hey :) Just for everyone's info: Alessandro has been working with us in the office for the past 2 weeks and it'd be great if you could support him in his work. I believe it will be valuable for Wikidata. --Lydia Pintscher (WMDE) (talk) 16:02, 12 August 2016 (UTC)
@Alessandro Piscopo: What about bringing quality statements (like the 1.0 classification) to Wikidata? Because they are language-specific they could be done as badges.--Kopiersperre (talk) 19:00, 12 August 2016 (UTC)
That would definitely be interesting. We should agree first what we mean with "quality" though. --Alessandro Piscopo (talk) 08:12, 16 August 2016 (UTC)

quality is in linking wiki links with wikidata statements[edit]

Hoi, attention to quality is good but I think the basics of what is perceived quality is in the occurrence of statements that describe links to other articles in Wikipedia. This allows for article level activitiy and work done in any language maps to work in all other languages. When we focus on what Wikidata is supposed to do in this way, most other quality considerations have a framework; the use that brings to being the data storage for Wikimedia projects. PS I blogged about this and welcome any arguments. Thanks, GerardM (talk) 12:21, 14 August 2016 (UTC)

@GerardM: interesting point of view. My doubt is: if Wikidata's quality should be considered in relation to what it can contribute to Wikipedia, don't you think that it may be limiting for the project? I think Wikidata might have much broader application that the mere support of Wikipedia. --Alessandro Piscopo (talk) 08:12, 16 August 2016 (UTC)
I see no reason why Wikidata should limit itself to being data storage for Wikipedia. If we have a recently discussed data set like >300.000 National Heritage buildings in the UK, most of that data isn't interesting for Wikipedia. On the other hand Wikidata works on integrating itself with OpenStreetMap and the from that point it can be quite useful.
There are also instances when Wikidata can be directly valuable. I use it for example as a multilingual dictionary for anatomy when I create my Anki cards.
I don't think that it's useful to think in terms of articles when looking at Wikidata. If an item has three links it might not be complex enough to make an article but those three links can still be very valuable to understand the structure of the underlying subject. ChristianKl (talk) 12:00, 23 August 2016 (UTC)
You forget the point to Wikidata. Yes, it can be more but the basics is that it supports Wikipedia and other projects. When it can bring substantial improvement in quality, both Wikidata and all the Wikipedias will benefit. This brings a practical and easily implementable difference that can be measured. It will brings us more contributors and this is imho more relevant than including external stuff. Thanks, GerardM (talk) 13:42, 23 August 2016 (UTC)

Stuff about railways[edit]

I've several questions regarding the relationships between railway stations. I'm specifically working with the Pearl River Delta at the moment, but I'm certain these can apply generally.

  1. What would be the appropriate connections to make between Shenzhen Railway Station (Q837327), Luohu Station (Q843947), Luohu Port (Q877115), Lo Wu Control Point (Q23498332), and Lo Wu Station (Q15169)?
  2. Is it necessary to have connecting line (P81) as a property of a railway station and as a qualifier to an adjacent station (P197)? (This question also applies to connecting service (P1192).)
  3. In a similar vein, does having one of the aforementioned connection properties require the inclusion of the other?
  4. Should two metro 'lines' be considered 'services' if they share the same trackage at any point? (Here I'm thinking of The Loop (Q2225459) and much of the Washington Metro (Q171221), but also of the concurrency of Line 3, Shanghai Metro (Q1326495) and Line 4, Shanghai Metro (Q1326504).)
  5. What's the hierarchy between metro/rail systems, their component lines, and their stations? Are stations part of (P361) lines part of (P361) systems? (Or are stations and lines both part of (P361) systems?)
  6. How would direction (P560) work for rail lines that are in loops? (Should we just pick two or three stations and use them for orientation?)

The property discussion pages are not terribly active, so I'm hoping there's some sort of existing consensus on these matters. Mahir256 (talk) 04:02, 15 August 2016 (UTC)

Try Wikidata:WikiProject Railways. Its mainpage is quite empty (feel free to fill it), but the talk is alive.--Jklamo (talk) 07:30, 15 August 2016 (UTC)
The simple questions to answer are 3 and 6. For 3, yes, they should be considered separate services if they are presented as separate services in reliable sources (e.g. system maps clearly treat the District line (Q211265) and Circle line (Q210321) as separate services at e.g. Cannon Street station (Q800615) even though they share the same tracks).
For 6, I'd use clockwise direction (Q16726164)/anticlockwise direction (Q6692036) or whichever cardinal direction is travelled in to reach the next station (e.g. the Circle line (Q210321) from Cannon Street station (Q800615) to Mansion House tube station (Q1477336) is west (Q679)), depending what reliable sources describe it as. Thryduulf (talk) 08:45, 15 August 2016 (UTC)
connecting line (P81) is defined as a qualifier, rather than a "main property". Danrok (talk) 03:09, 22 August 2016 (UTC)

New gadget to sort the statements on items[edit]

Hello everybody,

As the sorting of the statements on a item page had issues for a while, I'm glad to annouce that there's now a gagdet for it! Gadget-statementSort.js sort all the statements of an item, based on a properties ordered list.

This gadget have been created by Ladsgroup, using a previous script writen by Soulkeeper. Thanks a lot for your work!

You can now enable this gadget in your preferences. If you have any question about the gadget or if you want to suggest some modifications on the properties list, don't hesitate to ask Ladsgroup or leave a message below.

Bests, Lea Lacroix (WMDE) (talk) 09:39, 16 August 2016 (UTC)

Great gadget! @Ladsgroup: It would be nice if I could completely overwrite the default property list by a self-maintained list in a custom .js page in my userspace. Could you implement something like that? Thanks and regards MisterSynergy (talk) 12:37, 16 August 2016 (UTC)
I made phab:T143383 to keep track of it :) Best Amir (talk) 04:53, 19 August 2016 (UTC)
@Lea Lacroix (WMDE): Is this gadget restricted to special browsers, or what am I looking for? -- Innocent bystander (talk) 09:19, 20 August 2016 (UTC)
@Innocent bystander: No it should work in any browser and statements should be ordered the same way across all items when you enable it. --Lydia Pintscher (WMDE) (talk) 17:55, 25 August 2016 (UTC)

Large upcoming data import[edit]

Just a heads-up: In 2014, I created ~40K items for Grade I and Grade II* listed buildings in the UK. This list has grown to ~44K buildings by now. As discussed here, we are preparing to import the remaining Grade II buildings, using a current list from National Heritage. That would be ~342K new items. You can see an example of what these items will look like at Morgan Hall, The Lawns (Q26263429). Unless there are serious objections, I will commence item creation this evening or tomorrow. The import will be single-thread, so as to not overload Wikidata, and will be bot-flagged (because RC). --Magnus Manske (talk) 14:33, 16 August 2016 (UTC)

No objections, thanks for these huge donations. Sjoerd de Bruin (talk) 15:07, 16 August 2016 (UTC)
Splendid news. I look forward to working on this important data. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 15:08, 16 August 2016 (UTC)

Update: Import has commenced after positive feedback :-) View progress here (may be mixed with other unrelated edits). --Magnus Manske (talk) 15:33, 16 August 2016 (UTC)

Thanks for updating these! Hopefully it will inspire some people to use the geograph image import on Commons to illustrate them. Jane023 (talk) 15:35, 16 August 2016 (UTC)
WD-FIST recently gained the ability to match image-less items and Commons images via coordinates (100m radius). Pure coincidence, surely. --Magnus Manske (talk) 15:46, 16 August 2016 (UTC)
@Magnus Manske: happy to see that other countries are working on historic buildings too. National Heritage List for England number (P1216) has some constraint violations on Wikidata:Database reports/Constraint violations/P1216. Do you plan to work on that too? Some are quite easy to fix, like Listed buildings in Christleton (Q15979145) where some bot added a bunch of identifiers. Multichill (talk) 20:27, 16 August 2016 (UTC)
I tried to fix some of these in 2014, but someone said since (in this example) I can't make a query of Listed buildings in Christleton (because it is quite hard to automatically find that level of location data), it should stay in there. I do disagree with that; maybe we should try to get village-level information through some combination of the location name in the raw data (which is ambiguous), the larger region (which is not), and the coordinates. --Magnus Manske (talk) 20:35, 16 August 2016 (UTC)
Wouldn't it be better to include the lowest level of admin territory (civil parish) rather than the higher ones of district or county? BTW Christleton is a civil parish... Robevans123 (talk) 07:02, 17 August 2016 (UTC)
Of course it would be better. But on the "lowest level", there are many places that share the same name. And even if a Wikidata search only turns up a single one, how do I know it's the only one, and not just missing items for the others? National Heritage data doesn't come annotated with Wikidata item numbers, you know... --Magnus Manske (talk) 07:56, 17 August 2016 (UTC)

Update: Now trying to use more fine-grained located in the administrative territorial entity (P131). I was careful to get it right, but with these numbers, there is always a chance of some of them being wrong. Nothing that can't be fixed, but be aware just in case. --Magnus Manske (talk) 19:14, 17 August 2016 (UTC)

Cool - great to have the extra detail. Sorry for the extra work -:) Robevans123 (talk) 12:31, 18 August 2016 (UTC)
  • The example item doesn't contain "instance of" building. I think it would be great if you assign an instance of to the items you created.
A heritage status can change. It would be good if you add a "retrieved" source qualifier that tells the reader when the statement got created. It would also be helpful if reference url is filled as "" for the example item.ChristianKl (talk) 15:12, 19 August 2016 (UTC)

UPDATE This has now been completed. --Magnus Manske (talk) 23:12, 23 August 2016 (UTC)

Policy on Interface Stability: final feedback wanted[edit]

Hello all,

repeated discussions about what constitutes a breaking change has prompted us, the Wikidata development team, to draft a policy on interface stability. The policy is intended to clearly define what kind of change will be announced when and where.

A draft of the policy can be found at Wikidata:Stable Interface Policy. Please comment on the talk page.

Note that this policy is not about the content of the Wikidata site, it's a commitment by the development team regarding the behavior of the software running on It is intended as a reference for bot authors, data consumers, and other users of our APIs.

We plan to announce this as the development team's official policy on Monday, August 22.

-- Daniel Kinzler (WMDE) (talk) 14:50, 16 August 2016 (UTC)

It might make sense to announce breaking changes with [Breaking Change] or a similar tag, so that it's not required to read all mails that go through the mailing list. ChristianKl (talk) 09:59, 23 August 2016 (UTC)
Yes, that's sensible. Perhaps we'll add it to the policy. -- Daniel Kinzler (WMDE) (talk) 14:39, 23 August 2016 (UTC)

The policy is now official, see #Announcing the Wikidata Stable Interface Policy. -- Daniel Kinzler (WMDE) (talk) 14:39, 23 August 2016 (UTC)

Multiple items with identical sitelinks[edit]

Examples: Tasley (Q24668011) and Tasley (Q24668012). Not a good sign. --Magnus Manske (talk) 15:53, 17 August 2016 (UTC)

see Wikidata:True duplicates --Pasleim (talk) 15:57, 17 August 2016 (UTC)
Yes, that's a known problem, sadly it's not trivial to fix. The cause is that bots often double post entity creation requests to the API, so that both entities are created at nearly the same time, leading to our uniqueness constraints not working. Cheers, Hoo man (talk) 16:06, 17 August 2016 (UTC)
Why can't a bot merge items like that automatically? ChristianKl (talk) 15:28, 23 August 2016 (UTC)

Mixing up concepts[edit]

Should a Wikimedia disambiguation page (Q4167410) be used together with for example subclass of (P279) and country (P17), like on Abay District (Q364858)? I have a feeling that this causes conflicts and these concepts should be separated. We've had a similar discussion about templates earlier. Ping Infovarius. Sjoerd de Bruin (talk) 21:03, 17 August 2016 (UTC)

No. --Izno (talk) 21:36, 17 August 2016 (UTC)
But what we should do with pages (disambigs? set indices? lists?) like from ru:Категория:Страницы разрешения неоднозначностей:Населённые пункты? They have obvious properties like "containing settlements with the same name" and I am trying to mark it in Wikidata. --Infovarius (talk) 10:23, 19 August 2016 (UTC)
They are list or set index, but mark them as disambiguation. This makes no sense because disambiguation page don't must limit the content. The more correct solution is that change the kind of page. --ValterVB (talk) 11:02, 19 August 2016 (UTC)
  • Apparently (so my understanding) Russian Wikipedia uses disambiguation pages for all sorts of things. As these pages are linked from items with P31:Q4167410, all sorts of statements are added to such items, with the insistance that this "applies to part" ruwiki (a page could include a description of a capital, so the item would receive P31=capital).
    If pages on ruwiki are topical articles that combine various subjects, they should probably go on separate items and use some sort of P31 statement with "set index" and statements linking to the various topics combined. Merely adding this to items that are used for other things makes these items less useful for their original purpose.
    --- Jura 05:29, 18 August 2016 (UTC)
Same as Izno: no. Because a disambiguation item can list every kind of things. If a Wiki have the same disambiguation page but on their page they have also a different thing (for example a movie, a book a band with the same name) how we manage it? If you limit the content of a disambiguation, you have a list like this Lapwing (Q22293794) --ValterVB (talk) 07:12, 18 August 2016 (UTC)
Lapwing (Q22293794) is a good example! Yes, it is called "set index page", but it is a Wikimedia disambiguation page (Q4167410) too. So I suppose "set index pages" are a subclass of "disambig pages". And these pages also can be regarded as classes (with all appropriate statements), see discussion about greek characters. There are no such type of disambig templates in Russian Wikipedia yet. May be it's a solution - to create such type and move some such "disambigs" to "set indices"? --Infovarius (talk) 10:17, 19 August 2016 (UTC)
From en help: « A set index article is not a disambiguation page » and « (SIA) is a list article about a set of items of a specific type that share the same (or similar) name » so no, I don't think that a "set index pages" is a subclass of disambiguation but is a subclass of "list". --ValterVB (talk) 10:47, 19 August 2016 (UTC)
I see what you are afraid of, User:ValterVB, but I think usefullness is above this. See for example Mother (film) (Q4285420). It obviously contains only films. Why shouldn't we mark it as class of films? --Infovarius (talk) 10:23, 19 August 2016 (UTC)
Example: Mother (film) (Q4285420) is a disambiguation on, so it can be merged with others disambiguation item, for example with Mother (Q348342) (don't consider for now that already exist an ru sitelink) but this disambiguation page in en, or it or fr aren't about the movie but are about movie, song, videogame, play so don't have sense add mark it with a class of film. --ValterVB (talk) 10:47, 19 August 2016 (UTC)
I don't know why you are speaking about en/it/fr disambigs in Mother (Q348342), while ru:Мать (фильм) is a useful page itself. And it can have specific properties. --Infovarius (talk) 12:34, 19 August 2016 (UTC)
Because it's an example to explain the concept. I said "don't consider for now that already exist an ru sitelink", in this case, normally the page, of "ru:Мать (фильм) would be added to Mother (Q348342). If you want an real example you can see for example at The Great Gatsby (Q398510) the page in it's only a list of the film, it isn't possible add something of different, but if I check on other wiki I can see novel, soundtrack, movie and opera. --ValterVB (talk) 13:55, 19 August 2016 (UTC)
other examples Submission (Q2361990) Goldeneye (Q398775) Night Train (Q399204) --ValterVB (talk) 14:28, 19 August 2016 (UTC)

ItemDisambiguation limit at 100[edit]

The display limit for Special:ItemDisambiguation is set at 100. In cases of some ordinary names, e.g. John Campbell, this limit may be attained, or nearly so, for what is a reasonable request. In other words using such a page for normal disambiguation may be close to failing, and will fail as more items and aliases are added.

Could the number of hits be displayed? Could there be some fallback to a second page? It is highly desirable that this Special page should function as the global disambiguation equivalent for en:w:John Campbell, for example. Charles Matthews (talk) 06:12, 18 August 2016 (UTC)

As a work-around you can use this SPARQL query. --Edgars2007 (talk) 10:06, 18 August 2016 (UTC)

Thanks. Charles Matthews (talk) 07:06, 19 August 2016 (UTC)

Using Reasonator for disambiguation is much easier and informative.. Try John Campbell. Thanks, GerardM (talk) 05:56, 26 August 2016 (UTC)

A gadget for Wikidata completeness management[edit]

Dear Wikidata community,

we are happy to announce the next release of COOL-WD: a Completeness Tool for Wikidata, packed with new features, inspired by the engaging discussion with you all on the first release [1].

The main purpose of COOL-WD is to allow to create and manage completeness information about Wikidata, such as "Complete for all Switzerland's cantons" and "Complete for all of Obama's children". While previously one had to access an external tool to add and view Wikidata completeness information at, now a user script is available to enable adding and viewing completeness information directly inside Wikidata [2].

When the script is enabled, properties annotated as complete are marked by a green box, while all other properties are marked in yellow. To state that a certain property is complete, one can simply click on the yellow box to make it turn green. To add a reference URL for the completeness, one can click the small '(i)' icon next to the property name to add the reference URL. An example screenshot is available at [3].

Several other new features are:

- Completeness analytics: show the progress in data completion wrt class of objects of interest (, example screenshot at [4].

- Query completeness diagnostics: give an explanation (which completeness statements are used, and how) whenever (in)-complete query answers are given (, example screenshot at [5].

- Linked data publication of completeness statements, for instance, RDF description of the completeness statement of all cantons in Switzerland:

- RDF dump of over 10,000 completeness statements in COOL-WD collected from various sources:

Last but not least, a description of these features is to appear as a paper at COLD 2016 Workshop, which currently can be downloaded [6].

The tool is still a prototype, so we very much look forward to your feedback regarding how useful you consider the tool, and your ideas for conceptual or technical improvements!


Fariz, Simon, Rido, and Werner (Free University of Bozen-Bolzano, Italy)







Fadirra (talk) 11:33, 18 August 2016 (UTC)

Hello, what happen if a green statement gets changed/removed ? Maybe the completeness should be marked for review, to see if the modification is a vandalism or a real correction ? --Melderick (talk) 11:58, 18 August 2016 (UTC)
Hello! Currently, when a green (= complete) statement is clicked, then it will turn back to yellow (= potentially incomplete). This means basically that the statement is removed. I agree that marking would be a nice feature to add, as well as say a log/history feature for changes wrt. the statement and also voting feature for statements. - Fadirra (talk) 16:40, 18 August 2016 (UTC)

Reasonator is being weird, giving multiples of family members[edit]


Not sure whether its a problem with Reasonator or Wikidata but Reasonator is giving Nelson Mandela 3 copies of every family member.


--John Cummings (talk) 15:11, 18 August 2016 (UTC)

Yes, it does. It is still the only reasonable way of looking at the Wikidata data. Sad to say.. Thanks, GerardM (talk) 15:31, 18 August 2016 (UTC)
Thanks GerardM (talkcontribslogs), so its a problem with Reasonator? --John Cummings (talk) 16:31, 18 August 2016 (UTC)
See [1], it doesn't seem to have much attention. Sjoerd de Bruin (talk) 20:09, 18 August 2016 (UTC)
Very helpful, thanks Sjoerddebruin (talkcontribslogs). John Cummings (talk) 19:50, 19 August 2016 (UTC)
Sjoerddebruin (talkcontribslogs) The same can be said of the attention given to usability of Wikidata data. It does not get much attention and Reasonator is far better than anything else that is on offer. Thanks, GerardM (talk) 07:19, 20 August 2016 (UTC)

Unsourced and Wikipedia sourced P91 statements[edit]

What to do with unsourced and Wikipedia sourced sexual orientation (P91) statements? This really troubles me. The property has been used 3613 times, 2824 don't contain sources. Don't know how much have Wikipedia as source. Sjoerd de Bruin (talk) 21:28, 18 August 2016 (UTC)

613 has imported from (P143). Count. --Edgars2007 (talk) 21:39, 18 August 2016 (UTC)
Maybe there is the same problem for religion (P140) --ValterVB (talk) 06:44, 19 August 2016 (UTC)
If no source, delete. Snipre (talk) 09:39, 19 August 2016 (UTC)
No: If no source, find one and add it. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 10:56, 19 August 2016 (UTC)
We can't add sensible data on person without source. If some users add this data but don't add source makes an error. I think that the data is to be deleted. --ValterVB (talk) 11:08, 19 August 2016 (UTC)
If there is no source and the claim is plausible then look for a source. If you find a source, add it. If you don't find a source after looking and the claim is contentious or potentially so and/or your search was extensive and thorough then remove it. If the claim is not plausible, remove it. We really need a way of flagging the remaining cases where someone has done no or only a cursory search. Thryduulf (talk) 11:50, 19 August 2016 (UTC)
Nobody can force someone to look for sources, but Wikimedia Foundation, on biographies of living, asks you to add the sources, so if there are no sources for these data should be deleted. We can't keep sensible data indefinitely without source. --ValterVB (talk) 12:14, 19 August 2016 (UTC)
+1. WP require sources and WD aims to provide sourced data, so if people don't want to play the game, their contributions have to be deleted as useless and potentially subject to conflict. Snipre (talk) 16:15, 19 August 2016 (UTC)
So by this logic all statements on items about living people without a reference should be deleted (maybe except external IDs). I see no point in singling out this specific property. If it has to go, all does, or we accept that statements need to be decided about one by one (like Thryduulf said). And—still by this logic—the possible automatic statement deletion would not concern items about people who have passed away (still, talking about all properties, not only P91). – Máté (talk) 16:31, 19 August 2016 (UTC)
Not all the data are problematic, but certainly the data regarding religion or sexuality are more delicate and must be sourced. --ValterVB (talk) 16:46, 19 August 2016 (UTC)
Well, I'd add at least age, gender, residence, ethnic group, birth name etc. to the list of potentially just as sensitive data as sexual orientation and religion are. – Máté (talk) 16:51, 19 August 2016 (UTC)
(edit conflict) I think it is possible to define different priorities for properties in terms of how required sources generally are. Something like religion is pretty high up in most cases (I don't think it vital that a Catholic cardinal has source for religion, a US presidential candidate on the other hand probably is), however a handedness (P552) statement is almost never going to be controversial and so I don't think we should remove them without having looked for sources. External identifiers are almost always going to be self-sourcing, so we can treat them as completely uncontroversial. I'd suggest levels:
  1. always required, will be deleted if a source is not provided within a short time of the statement being added (should only be used for a very few properties and almost never when used for deceased people);
  2. almost always required, will normally be deleted when applied to living people or recently deceased people if a source is not provided but exceptions are possible based on common sense, especially for deceased people. (more than level 1, but not too many)
  3. Should be provided, statements should be accompanied by a source but they will not be routinely deleted without consideration of the circumstances (this should be default for non-external id properties)
  • low priority, statements should be accompanied by a source but they will not normally be deleted unless verification has failed or the statement is both implausible and applied to a living person (only things that will rarely be controversial should be at this level)
  • self-sourcing, no independent source is required (probably only applies to external identifiers). Thryduulf (talk) 16:58, 19 August 2016 (UTC)
we need a team to source statements. the people who periodically drive by and suggest statement deletion, are not collaborating and improving the data. "so if people don't want to play the game", they can take their deletion game elsewhere. there is no consensus for required references. Slowking4 (talk) 12:17, 21 August 2016 (UTC)
other self-sourcing statements are bibliographical data of Books or Texts, when the item is linked to the source (on wikisource for ex.). It seems very irrelevant to say that "Title" is so and so, and source on... the Book itself, which would be linked in the wikisource link... :) --Hsarrazin (talk) 18:45, 23 August 2016 (UTC)

Languages to add[edit]

Syriac I want to add the name in original language property to Q203179 for the text "ܥܕܬܐ ܕܡܕܢܚܐ ܕܐܬܘܖ̈ܝܐ" in Syriac, but I cannot add text in syc. Why is this list of values missing so many possible options? —Justin (koavf)TCM 05:07, 19 August 2016 (UTC)

It is possible languages are missing for monolingual text. You can create a ticket for it in phabricator or request addition on WD:DEV. Mbch331 (talk) 13:07, 19 August 2016 (UTC)

local Wikibase[edit]

I'm trying to use Wikibase in my local wiki and I am half way done. But I have problems still (I don't know if it's better to ask at developers forum?):

  • After some change I cannot edit or create claims. Specifically: for creation I click "edit", then I choose right property and get eternal "loading sign" instead of input field; for editing - just nothing happened after clicking "edit". Editing through API with my bot is possible though.
  • How to edit label and description in other languages? I've installed Babel extension, imported LabelLister gadget but nevertheless I see only 1 language.
  • Even in that language I cannot edit label+desc+aliases simultaneuosly: only one of them is saved at a time, then reload of a page is needed to save another.

Can anyone help? --Infovarius (talk) 09:54, 19 August 2016 (UTC)

@Hoo man: probably better for a dev to look at this one. --Izno (talk) 11:28, 19 August 2016 (UTC)
@Infovarius: I'm quickly replying point by point:
  1. Make sure you are running Wikibase master and MediaWiki master. Do you get any JS errors in this case? If so, please report them.
  2. Installing Babel and editing your user page should be enough. Please note, that you will need to run jobs (run maintenance/runJobs.php in MediaWiki's root directory) in order for your user pages categories to be written to the database, so that Babel can pick them up.
  3. See #1
Cheers, Hoo man (talk) 13:03, 19 August 2016 (UTC)
Thanks for answer. I've installed master version of Wikibase and get "Fatal error: Class 'Wikibase\DataModel\Entity\ItemId' not found in C:\xampp\apps\mediawiki\htdocs\extensions\Wikibase\lib\WikibaseLib.entitytypes.php on line 37". Then I revert Wikibase upgrade.
I've copied MediaWiki 1.27 over my 1.26. Now I see all required labels/desc/aliases(!), but editing leads to error "SyntaxError: Unexpected token < in JSON at position 0".
After rolling master version of Wikibase again I get "Fatal error: Class 'Wikibase\Lib\DataTypeDefinitions' not found in C:\xampp\apps\mediawiki\htdocs\extensions\Wikibase\repo\includes\WikibaseRepo.php on line 300". I revert Wikibase again.
In console I have: 1) Unknown dependency:; 2) ReferenceError: $ is not defined(anonymous function) @ Item:Q2:935 --Infovarius (talk) 15:42, 22 August 2016 (UTC)
Did you run composer update for both MediaWiki and Wikibase after updating them? For MediaWiki you will also need to run maintenance/update.php after applying the update. Cheers, Hoo man (talk) 15:50, 22 August 2016 (UTC)
Oh, that helps! All described problems are gone, thank you! During maintenance/update.php there was an error "Error: 1071 Specified key was too long; max key length is 767 bytes", but I see no problems in functioning yet. --Infovarius (talk) 15:46, 23 August 2016 (UTC)

Merge and purge[edit]

I noticed the behaviour of the merge gadget changed. It now asks for confirmation after merging to purge the page. This is a bug and I filed Phab:T143435 for this. Multichill (talk) 17:14, 19 August 2016 (UTC)

See also Wikidata:Contact the development team#Purge with confim. --Edgars2007 (talk) 17:24, 19 August 2016 (UTC)


Hello. Is there a way to find how many unique item are linked to Alki Larnaca F.C. (Q658131) that has for property subclass of (P279) the item Cypriot Cup (Q245970)? Or is there a way to find how many wikidata pages of the articles of w:el:Κατηγορία:Κύπελλο Κύπρου (ποδόσφαιρο ανδρών) are in Alki Larnaca F.C. (Q658131)? Xaris333 (talk) 18:12, 19 August 2016 (UTC)

Easy method: Special:WhatLinksHere/Q658131. More complex queries are possible with the Wikidata Query Service, but this would be something you’d need to learn first. —MisterSynergy (talk) 19:09, 19 August 2016 (UTC)

Hierarchical data examples[edit]

I am trying to find how hierarchical information is stored in wikidata? Most of the Olympics Sports pages have a Tournament Draw or Bracket like this or this. I would like to add this "who played who, at which stage" info, but I am a bit confused whether such info gets stored. The docs say lists and infoboxes are the main focus, so is this something that will be handled later? If its already being done could someone please point at a tutorial or examples for a beginner. Thanks! Quil1 (talk) 03:24, 20 August 2016 (UTC)

You could create an item:
Label: "W-j Kim vs. R Ega Agatha in round of 32 of the Archer at the 2016 Summer Olympics"
instance_of: "archery contest round at the Olympics"
participant: "W-j Kim"
participant: "R Ega Agatha"
is_part_of : "round 32 of the Archer at the 2016 Summer Olympics"
succeeds : "R Ega Agatha vs M Nespoli in round of 16 of the Archer at the 2016 Summer Olympics"
follows : "W-j Kim vs. G Sutherland in round of 64 of the Archer at the 2016 Summer Olympics"
follows : "R Ega Agatha vs. Y Xing in round of 64 of the Archer at the 2016 Summer Olympics"
This follows/succeeds? pattern stores the data in a Wikidata friendly way. However at the moment there's no easy way to integrate such data into Wikipedia. It's also possible that there are more specific properties than the one I listed here. ChristianKl (talk) 09:48, 26 August 2016 (UTC)

New Nomination[edit]

Hello.Please subscribe to The first Arabic user's nomination on Wikidata.Thank you --ديفيد عادل وهبة خليل 2 (talk) 10:50, 20 August 2016 (UTC)

Query Service to PetScan[edit]

Hi, this works in Query Service but it does not work when I copy it to PetScan. How can I run the same query on PetScan?--ԱշոտՏՆՂ (talk) 16:43, 20 August 2016 (UTC)

The output needs to start with ?item. This is the only output that will be used, so you could remove "label". Sample.
--- Jura 16:49, 20 August 2016 (UTC)
Thanks!--ԱշոտՏՆՂ (talk) 16:57, 20 August 2016 (UTC)

Aquarium life[edit]

aquarium fish (Q1448518) = "aquarium life" aka "aquarium animals" = "ornamental fish" is a categoría de Wikimedia aka Wikimedia list article? -- 00:36, 21 August 2016 (UTC)

No, it's not. It's a subclass of (P279) of fish (Q152), whose topic's main category (P910) is Category:Aquarium fish (Q8084627) and its Commons category (P373) -> Category:Aquarium fish. Strakhov (talk) 01:00, 22 August 2016 (UTC)

Property for listing teams than an individual has coached/managed[edit]

I was unable to find a property that could be used to specify the sports teams than an individual has coached/managed. I do not consider it ideal to use P:54 in such a context as it relates more to the teams an individual is associated with as a player, not as a coach. Does the property I am looking for need to be created or is there one that I simply have not found yet? Thanks, Lepricavark (talk) 00:12, 23 August 2016 (UTC)

Wikidata:Property proposal/head coach of is currently open, you may wish to comment there. Thryduulf (talk) 00:29, 23 August 2016 (UTC)
Thank you. I will do so. Lepricavark (talk) 11:07, 23 August 2016 (UTC)

How do I properly use Population property?[edit]

I try to use it but however, it always has a plus minus sign with "1". How do I fix this? MechQuester (talk) 06:07, 23 August 2016 (UTC)

If the number is exact, and there is not a range of valid values, write "+-0" after the digit entered. --β16 - (talk) 07:26, 23 August 2016 (UTC)

New gadget: currentDate[edit]

Hello everyone,

A new gadget has been added to our collection, this one (currentDate) automatically fills in the date of today when you use retrieved (P813). This one seems a perfect candidate to be enabled by default, we only need consensus for it. Thanks to TMg for creating this gadget!

Greetings, Sjoerd de Bruin (talk) 08:18, 23 August 2016 (UTC)

Seems a very useful gadget. Support making it default. Thryduulf (talk) 09:31, 23 August 2016 (UTC)
Agree. Robevans123 (talk) 10:35, 23 August 2016 (UTC)
I often add a "past date" to this property, but I still approve this! -- Innocent bystander (talk) 11:56, 23 August 2016 (UTC)
JFDI applies. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 14:17, 23 August 2016 (UTC)
at last ! I"ve been waiting for this soooo long !
support making it default too ! --Hsarrazin (talk) 19:08, 23 August 2016 (UTC)

Announcing the Wikidata Stable Interface Policy[edit]

After a brief period for final comments (thanks everyone for your input!), the Wikidata:Stable Interface Policy is now official.

This policy is intended to give authors of software that accesses Wikidata a guide to what interfaces and formats they can rely on, and which things can change without warning.

The policy is a statement of intent given by us, the Wikidata development team, regarding the software running on the site. It does not apply to any content maintained by the Wikidata community. -- Daniel Kinzler (WMDE) (talk) 14:37, 23 August 2016 (UTC)

Wikidata weekly summary #223[edit]

UBERON ID (P1554) should be datatype external id[edit]

There's a formatter url for UBERON ID (P1554) but currently the datatype is string, so it doesn't get automatically used. I think it would be benefitial to change the datatype to external id. ChristianKl (talk) 18:54, 23 August 2016 (UTC)

I don't think it will work with multiple formatter IDs. --Izno (talk) 19:07, 23 August 2016 (UTC)
@Izno: That hasn't stopped such properties like P:P2182 from being converted. -- Innocent bystander (talk) 05:23, 24 August 2016 (UTC)

Property to model that the biceps brachii "flexes elbow"[edit]

Currently enWiki displays in its infobox on that the biceps brachii has the "action" "flex elbow". Do we have an existing property to model this relationship? ChristianKl (talk) 19:11, 23 August 2016 (UTC)

I suppose you could say
< biceps brachii > use (P366) See with SQID < flexing >
of (P642) See with SQID < elbow >
but that seems very cumbersome and may not easily generalise. Thryduulf (talk) 21:51, 23 August 2016 (UTC)
That seems cumbersome so I produced a new property: ChristianKl (talk) 08:32, 24 August 2016 (UTC)

How to create a GUIDELINE for specific task or project here?[edit]

I was drafting here an Guidelines for external relationships to consolidate discussions and create an help page for new task-force people...

There are some example of that kind of Guideline (for an external task force linking ontologies)? Where the best place, here? Wikiversity?

PS: we are linking SchemaOrg with Wikidata... It is a startup project, and need some consolidation and collective consensus, to achieve the quality levels that we need.

Start a WikiProject. ChristianKl (talk) 08:15, 24 August 2016 (UTC)

"missing" items for series of ...[edit]

Within the realm of serial items. Some items cannot currently be given correct claims for instance of (P31), just because there is no existing class for it.

For example, we do have book series (Q277759) which is fine for books, but we do not have an item for say, a "series of events" which could have sub-classes such as "series of military campaigns".

Currently, we have a lot incorrect claims along the lines of Battles of Khalkhin Gol (Q188925), which is incorrectly claimed as a instance of a battle, despite being a series of battles.

So, I plan to create some suitable items for this. If anyone has any input or suggestions then please do comment. Danrok (talk) 00:32, 24 August 2016 (UTC)

For Battles of Khalkhin Gol (Q188925) in is a series but in other wiki is a single battle. Before change P31 is necessary to check and split the item. --ValterVB (talk) 06:49, 24 August 2016 (UTC)
Mmm tough problem at first sight. My first though is that "battles" and "series" do not mix very well all the time. Composition seems to be a better fit, if for one reason battles of some war can overlap and have sometime no obvious precedence order. The order might in some way related to a causality sequence, a battle can have been thought by one army because another one have been lost by the same army. A war is composed of several battles, and maybe several "sub-wars" ? A casestudy could be Hundred Years' War (Q12551) View with Reasonator See with SQID I guess. Does a war is a case of conflict that begins with a declaration of war (Q334516) View with Reasonator See with SQID and ends with a peace treaty of another way ?
May a war as a consequence cannot be composed of smaller wars and the composition holds at some higher level like an armed conflict that can be composed by some other armed conflict ? Can a battle be composed of smaller battles ?
A lot of questions and no answer, sorry /o\ author  TomT0m / talk page 07:35, 24 August 2016 (UTC)
The definition of battle is "part of a war which is well defined in duration, area and force commitment". I don't see why a series of battles can't be a battle according to that definition. It's also worth noting that different cultures speak differently about the same event and there's no reason why the English version should have preference. ChristianKl (talk) 10:23, 24 August 2016 (UTC)
Try significant event (P793) --Succu (talk) 21:18, 24 August 2016 (UTC)
part of (P361) is your best tool here, I think.
"Battle" is a pretty broad term, and a "battle" can include other "battles" - consider Battle of the Somme (Q132568) (1 July - 17 November 1916), which began with Battle of Albert (Q1992231) (1 July - 13 July), then Battle of Bazentin Ridge (Q2634717) (14 July - 17 July), and so on. These can all be "instance of: battle" and part of (P361) of the larger battle. They may also contain smaller events, which might also be labelled battles, or P31:engagement (Q6680005) (though I don't think many pages use these yet).
In some cases, battles might also be P361 of a military campaign (Q831663), a connected series of battles - so Battle of Kvam (Q20112888) is P361:Operation Weserübung (Q150939) which is P361:Norwegian Campaign (Q5084679)... and in the grand scheme of things, P361:World War II (Q362). So you can use P361 to go all the way up and down the chain.
In the specific case of Battles of Khalkhin Gol (Q188925), I'd consider whether P31:military campaign (Q831663) is the best way to go - they all fitted together as part of an overall series of battles. Andrew Gray (talk) 21:35, 24 August 2016 (UTC)
@Andrew Gray: military campaign (Q831663) Mmm is not this related to only one opponent actions ? A campain is how one side organized its actions imho. So it's in most case not appropriate as a battle is the sum of the actions of both side. It can be even a NPOV problem if you mix the two concept inappropriately. author  TomT0m / talk page 06:18, 25 August 2016 (UTC)

I think I picked a bad example, battles and wars are complex to define. Plus, there's the language problem, and the different way things are defined in different languages. Danrok (talk) 21:40, 24 August 2016 (UTC)
Just explicit the definitions you use. It's not exactly a problem specific to wars :) Wikidata should be definition based, not term based. author  TomT0m / talk page 06:18, 25 August 2016 (UTC)

Q18396215 to be merged with Q16983762[edit]

per similarity of the subjects covered. -- 10:48, 24 August 2016 (UTC)

Thanks, but you could have merged them yourself. Jared Preston (talk) 11:19, 24 August 2016 (UTC)
How could I do that? -- 12:11, 24 August 2016 (UTC)
Help:Merge. —MisterSynergy (talk) 12:18, 24 August 2016 (UTC)
It works! Thanks. -- 12:24, 24 August 2016 (UTC)

creating guidance on the process of importing data into Wikidata[edit]

Hi all

I've started a draft of guidance on importing data from external datasets into Wikidata, its very rough at the moment, I would very much appreciate some help.


--John Cummings (talk) 12:27, 24 August 2016 (UTC)

Perhaps something regarding sourcing and references? Danrok (talk) 12:41, 25 August 2016 (UTC)

Wikipedia corpus to Wikidata[edit]

Extracting data from corpus is a very promising and challenging problem, many research teams around the world are working on this problem, some of the examples: creating a module that make diagnose of a disease by scanning medical corpus, creating a voice assistance and much more. Wikipedia “as source of corpus” while wikidata is the “structured data” of that corpus. Offers a great environment to approach this problem.

Before getting into the details of this problem, I am interested to know how open is the Wikidata community to this problem? --GhassanMas (talk) 19:11, 24 August 2016 (UTC)

Wikidata imports a lot of information from Wikipedia already. In general Wikidata prefers to have data that cites sources from outside of Wikipedia.
Additionally there are the projects and that try to extract facts via machine learning and provide it to the PrimarySource tool where WikiData editors can approve or disapprove suggestions from the machine learning algorithms.
If you are an academic working on extracting facts from corpus data there's a good chance that you can work well with Wikidata, but it's worth to first understand the structure of Wikidata and how it plays with the Primary Source tool. ChristianKl (talk) 15:10, 25 August 2016 (UTC)
It is important to understand that the Primary Source tool is something that is not used for importing Wikidata and many other sources by the Wikidata community. Understanding this difference is vital. The biggest problem is not getting data into the PMS but finding people to consider the data in there. The statistics prove that the PMS is dysfunctional. Thanks, GerardM (talk) 05:54, 26 August 2016 (UTC)
"Not used" doesn't seem to be what the statistics says. It doesn't get used as much as desired but it still get's used. Even if only a subset of the data is evaluated by humans, an academic group that provides data via the primary sources tool can expect to get some feedback over what claims get approved and rejected.
I consider the PMS a work in progress. With increased data there's a higher probability that when I browse an item the PMS will suggest a statement or reference that I find valuable to add.
More data also means that there are higher returns to improving the UI of the PMS and thus improving usage. ChristianKl (talk) 09:44, 26 August 2016 (UTC)

Preferred sourcing for imported data already linked in identifiers?[edit]

I'm rewriting the script I use to generate items such as Thomas Thompson (Q26689403) to include sourcing. At the moment, every item has a History of Parliament ID (P1614) property, giving a clickable link to the main source. Given this, should I source the individual statements as stated in (P248):The History of Parliament (Q7739799), or as reference URL (P854):(the URL from P1614)? I'm leaning towards the reference URL (P854) approach but thought I'd better check which is preferred. Andrew Gray (talk) 19:50, 24 August 2016 (UTC)

ValterVB adviced me to use reference URL (P854) in similar cases, but we'd like to hear other opinions. --Epìdosis 20:04, 24 August 2016 (UTC) I've read again your message: in these cases I've always used stated in (P248) + reference URL (P854); the properties ValterVB adviced me never to use in references are identifiers (in this case History of Parliament ID (P1614)). --Epìdosis 20:08, 24 August 2016 (UTC)
per Help:Sources#Databases you would need to use stated in (P248):The History of Parliament (Q7739799) and History of Parliament ID (P1614) as source. --Pasleim (talk) 20:09, 24 August 2016 (UTC)
I didn't know it. Thank you, Pasleim. --Epìdosis 20:36, 24 August 2016 (UTC)
Interesting. Epìdosis's recommendation feels much more natural to me. The database approach it seems to require that we tag The History of Parliament (Q7739799) as P31:database, which would be wrong - it's a reference work that we record identifiers for, not a "database" like PubChem. It fits a lot better with the website approach in the section above (P248/P854). Andrew Gray (talk) 20:49, 24 August 2016 (UTC)
@Andrew Gray: So what is The History of Parliament (Q7739799) ? Because when I read the label of History of Parliament ID (P1614), I have "identifier on the History of Parliament website". According to that definition we have an online database. Snipre (talk) 21:01, 24 August 2016 (UTC)
Fair point - the difference here is a bit hazy :-). I wouldn't call it "instance of: database", though - and according to the help page that's a key element. Andrew Gray (talk) 21:17, 24 August 2016 (UTC)
@Andrew Gray: The first thing is to clarify the status of The History of Parliament (Q7739799): an item can't be at the same time a project and a reference work. Then if we consider that History of Parliament ID (P1614) is the identifier of the online version of The History of Parliament defined as a reference work then we can then follow the recommandations of Help:Sources#Databases. Here we have to better define what we want to link because The History of Parliament is now the same denomination for 3 things: a project (which is an organization), a reference work (published mainly as books) and a website build as a database. Theoretically we should have 3 items, each describing these 3 different concepts. Snipre (talk) 21:33, 24 August 2016 (UTC)
Hmm. I agree that "project" and "work" is a little confusing, but it's good enough for the moment (there is a project, it produces a work). We can fall back on only "work", though, if preferred. However, it's not a database nor is it built as a database - unless we define everything online as a database! I'm taking the information from there but I'm transcribing it by hand and then uploading with a script. Perhaps I shouldn't have used the word "imported"... Andrew Gray (talk) 21:43, 24 August 2016 (UTC)
@Andrew Gray: No, it's not good enough because if you try just to expand a little the information of The History of Parliament (Q7739799), we will have problems when adding specific properties. Can a project have a author (P50) property ? Can a reference work have a inception (P571) property ? If we follow your reasoning "there is a project, it produces a work", all data present in WD should be stored in the entity (Q35120) item as everything is an entity. We have to create enough items to identify correctly the concepts in order to avoid the current problem of using the same item for different purposes.
Then what is a database ?
* Systematically organized or structured repository of indexed information (usually as a group of linked data files) that allows easy retrieval, updating, analysis, and output of data
* A comprehensive collection of related data organized for convenient access, generally in a computer
* A collection of pieces of information that is organized and used on a computer
So the website can be considered as a database because one part of it at least is composed of structured information accessible by an automatic query. Snipre (talk) 07:56, 25 August 2016 (UTC)
By this definition, so can literally any website with information on it in an organised form. Wikipedia is a 'database'. I don't think it's a very helpful way of thinking about things, and to be honest I think the arbitrary distinction between 'website' and 'database' made by the help page just causes more confusion. Andrew Gray (talk) 20:43, 25 August 2016 (UTC)
If retrieved (P813) is intented to be added too, I'd say using the ID property for sourcing would be wrong, because the date won't be true if the formatter URL changes (?). Strakhov (talk) 23:06, 24 August 2016 (UTC)
@Strakhov: You have retrieved (and verified) the information stored in the statement at the given date rather than retrieved a particular URL. I can see your point, but I don’t think we should worry about this one. —MisterSynergy (talk) 08:39, 25 August 2016 (UTC)
Just food for thought. Not even sure if it was of the nutritive kind when I wrote it. :) Strakhov (talk) 11:43, 25 August 2016 (UTC)
In this case, the URL and ID property are fairly interchangeable anyway - the ID is a URL slug :-) Andrew Gray (talk) 20:43, 25 August 2016 (UTC)
From what I can see, the use of such properties like History of Parliament ID (P1614) in the references makes stated in (P248) redundant! In the page of P1614, there is a subject item of this property (P1629)-claim that links to The History of Parliament (Q7739799). That chain of relations is probably enough to describe this. In fact, that relation have we used on svwiki, when we decipher the references here at Wikidata. See for example note 1 at sv:Adelaide av Bourbon-Orléans. The use of both P248 and P227 here now gives two links to "Gemeinsame Normdatei", one of them are obviously redundant here, and I think it is P248! -- Innocent bystander (talk) 07:39, 25 August 2016 (UTC)
@Innocent bystander: No, stated in (P248) is not redundant because it gives you directly the item where you can find additional properties if needed. Without stated in (P248) you have to first retrieve the item connected with the property and then you can retrieve the data you want. If you take the time to look ãt the current templates used in the different WPs to cite sources, you can see that most of them require much more data than available in the sources section below a statement. So better to provide from the beginning the most related items of a source in order to reduce queries. Snipre (talk) 14:40, 25 August 2016 (UTC)
  • Pinging @Carcharoth:. He was interested in this stuff two weeks ago. Strakhov (talk) 21:14, 24 August 2016 (UTC)
    • Sorry, no idea what I can usefully say! Am following it with interest, though. Carcharoth (talk) 22:10, 24 August 2016 (UTC)


Demonstrating the options - all on an item which already has History of Parliament ID (P1614):1790-1820/member/thompson-thomas-i-1767-1818

Just P248 - item link to work
Just P854 - reference URL
P248 & P854 - item link and reference URL
P248 & P1614 - item link and property with identifier
According to Help:Sources

This structure avoids any change of the source structure in case of URL modification (if the URL change, you just have to correct the URL in the property History of Parliament ID (P1614)) and allows a nice link where the long URL can be hidden below the title value when using the source in Wikipedia (everyone prefers to see something like that Thomas Thompson instead of that ).

I'm reluctant to use P1476 here because it would involve a lot more effort (I'd have to call up and scrape a few thousand pages to get the title phrase for each statement), but I accept that's not a very good argument against it ;-) Andrew Gray (talk) 20:43, 25 August 2016 (UTC)
A year ago I wondered why a title (P1476) is at all necessary, since one could simply use the item’s label as a reasonable title for instance. However, User:Snipre came up with a convincing argument on Help talk:Sources: there are “Bonnie and Clyde problem”-like situations with Wikidata items and external database entries that do not have a 1:1 equivalence, and thus a Wikidata label is not necessarily identical to the title of a corresponding database entry. This does indeed require extra effort, but I think it is worth to do this work. —MisterSynergy (talk) 20:54, 25 August 2016 (UTC)

Follow-up questions[edit]

This is somehow related to the above discussion, thus I put it here. The structure of references is important to give data users (e.g. in Wikipedias) the ability to easily generate references based on our data. They (a) need to find all relevant information for a valid reference (e.g. in en:Template:Cite web), and (b) expect to find that information in always the same structure (i.e. our references are always composed of the same properties). However, we have a couple of different reference structures defined in Help:Sources (“Database”, “Webpage”, “Book”, etc.) and I have two follow-up questions:

  • How can one see which reference structure is actually used in a particular case? Take all properties and decide whether they form something useful?
  • Is there any technique known to query “incomplete” sources here at Wikidata, e.g. by using the Wikidata Query Service? This would be useful for reference maintenance.

I would be happy to hear about your ideas. Thanks, MisterSynergy (talk) 08:51, 25 August 2016 (UTC)

  • One large problem with adapting Wikidata-references with templates like "Cite web" is that we here technically allow 17 reference URL (P854) inside the same reference. (Nothing is technically stopping that option.) And no template on any Wikipedia is adapted to that. The template we use on svwiki in the article sv:Adelaide av Bourbon-Orléans (which I mention above) on is adapted to any number of reference-url's. The references will look very strange, but they are fully readable, all 17 of them. This template is not based on any present template, it was instead adapted to the open framework we have here on Wikidata. It has many flaws, but it is a start. -- Innocent bystander (talk) 10:33, 25 August 2016 (UTC)
Interesting. Are there any use cases for multiple reference URL (P854) (or identifier statements) in a single reference? Can’t we define those references as “technically invalid references” and make them show up on maintenance lists? Who picked the number of 17 and why? —MisterSynergy (talk) 10:47, 25 August 2016 (UTC)
I do not know if multiple P854 is a big issue, and it should probably be regarded as something that has to be maintained. I have at least seen references with two valid urls. My point is that unexpected use of properties are to be expected in our references. It is a good intention to maintain those, but my experience from Wikipedia and from our contraints-lists here at WD is that these maintainance-lists tends to be longer and longer by time. We have to take into consideration that we never will fix them all. One of the most common use of references here is the cases when an url is all that is found in the reference. If the webpage is in Armenian or the webpage is dead, very very few of us can set a "title" to such a reference. -- Innocent bystander (talk) 12:50, 25 August 2016 (UTC)
unexpected use of properties in references — I really feel uncomfortable with such a guideline, although I admit that it might be the best we can reasonably achieve here. I’m from dewiki, whose community is extremely skeptical about Wikidata, and at the moment they say: “There are no references at Wikidata, so we don’t use it!” Once we’ll get that right I see them saying: “Wikidata references are useless and messy, so we don’t use it!” —MisterSynergy (talk) 13:04, 25 August 2016 (UTC)
@MisterSynergy: Frankly speaking I can say that the use of data from WD by WPs is often the last problem of several contributors in WD. Just think about the continuing import of data from WP into WD without original sources or the different games which help to add data to WD but without adding the source too. Snipre (talk) 14:44, 25 August 2016 (UTC)

The closing of RFC without a summation is not a good practice[edit]

I find it unhelpful and both a little pointless and disappointing that some of the Wikidata:Requests for comment are simply closed as "no consensus" without a summation of the viewpoints. Where a conversation has taken place and someone is moved to close it, I see true value in the closure to explain the alternate viewpoints, and why how that person closing the discussion has determined the position of the community. People have invested in putting their points of view, and if someone cannot take the effort to summarise, then what are they doing closing the discussion? An open discussion is not problematic, it just looks untidy to some.  — billinghurst sDrewth 23:49, 24 August 2016 (UTC)

That is why lot of people fear to close RfC, as result there were open RfC for more than one year. I think that there is no benefit from RfC being open so long, as situation may change during the time. So I support timely closing RfC, even with no consensus result (and without a summation of the viewpoints). --Jklamo (talk) 07:24, 26 August 2016 (UTC)

Translation of Wikidata weekly summary[edit]

Hi, even if this discussion is about translation, I prefer to talk about that here rather in Wikidata:Translators' noticeboard. So, if Matěj Suchánek does not read here, he could be interested. My question is rather simple, as Wikidata is a multi-language project, would it be possible to translate the Wikidata weekly summary before they are published so that, user can read them in their native language? I know that TomT0m do this work on French Wikipedia a posteriori so I wonder if it would be possible, first technically, to translate these news a priori. If so, would it be possible for users registered here to receive this summary in their preferred language if the translation has been done in this language? Pamputt (talk) 18:50, 25 August 2016 (UTC)

I know the parsoid/wikitext team does a call for translation before publishing their status update, but I don't know how they dispatch it. author  TomT0m / talk page 18:53, 25 August 2016 (UTC)
Do you know the page where we can find this call? Pamputt (talk) 19:06, 25 August 2016 (UTC)
Recieved it in my mailbox for some reason, after a bit of digging it's visual editor team, not parsoid, and the mail can be found on the internet : author  TomT0m / talk page 19:12, 25 August 2016 (UTC)
(edit conflict) We could re-implement the logic of Tech News distribution. But if we really do, we need to overhaul the way the weekly summary is composed at first. Perhaps linking from the English weekly summary which gets distributed to a translated version would be sufficient.
Note that we were able to translate several status updates in 2014. Matěj Suchánek (talk) 19:14, 25 August 2016 (UTC)

Another information, what I am asking for already exists for Tech news. Pamputt (talk) 19:07, 25 August 2016 (UTC)

Special:Translate exists on this wiki, so it is technically possible to have the newsletter in multiple languages, and delivered in multiple languages, as you have seen with Tech News. The issue is always going to be timeliness of the production, the translation, and then the delivery.  — billinghurst sDrewth 07:12, 26 August 2016 (UTC)
And the wasted time that could be used for translating help pages and other stuff. Sjoerd de Bruin (talk) 07:47, 26 August 2016 (UTC)

Thanks for your feedbacks on this topic. I'm currently working on improving the Weekly Summary, from the content to the technical issues, and I'll be glad to hear your feedbacks and ideas about this :) Lea Lacroix (WMDE) (talk) 08:57, 26 August 2016 (UTC)

This is not necessary, imho. --Molarus 09:40, 26 August 2016 (UTC)