Wikidata:Project chat/Archive/2015/07

From Wikidata
Jump to: navigation, search

This page is an archive. Please do not modify it. Use the current page, even to continue an old discussion.

Contents

Regex queries

Is there a way to do regex queries? Specifically, I'm thinking about something like Query: STRING[345:"tt\d{7}"] AND NOCLAIM[57] to give me items that have a IMDb ID (P345) starting with "tt" without a director (P57). Gabbe (talk) 09:14, 26 June 2015 (UTC)

It might be a sign that we should have several imdb properties. In this specific case, Query: claim[31:11424] and claim[345] and noclaim[57] would give similar results.--- Jura 11:48, 26 June 2015 (UTC)
Ah, but the items I'm looking for won't necessarily have the statement instance of (P31)  film (Q11424). Gabbe (talk) 11:58, 26 June 2015 (UTC)
... I see now that I can download https://wdq.wmflabs.org/api?q=claim[345]%20and%20noclaim[57]&props=345 and run the regexp on that on my own computer. Gabbe (talk) 12:41, 26 June 2015 (UTC)
SPARQL supports regexp queries. The query you wanted to ask is here: films without directors. This is the SPARQL test service on a low-powered server, and I needed to tweak the query a bit until it ran without a timeout. In particular, the way in which I check for the absence of P57 is a bit indirect (optionally finding it and then filtering to the cases where no value was found -- !Bound). There are more direct ways to say this in SPARQL ("NOT EXISTS" filter) but I got a timeout there. The query is limited to 100 results, but if you fix them, the data should catch up (it's synched with Wikidata), so maybe 100 at a time are enough. --Markus Krötzsch (talk) 07:27, 1 July 2015 (UTC)

Machine Synchrone

Not unrelated to my previous question: the (fr) article Machine synchrone is about both synchronous motor and alternator (same principle applied to different devices. As is, the article is linked to synchronous motor but can't be reached from alternator. Is there a way to improve this situation, since in this case a merge would not be acceptable? Evpok (talk) 13:09, 28 June 2015 (UTC)

@Evpok: Wikidata:WikiProject Cross Items Interwikis is looking at ways to solve this problem. Filceolaire (talk) 18:21, 1 July 2015 (UTC)

Q915373

Copied from Deletion requests
Never edited here on Wikidata before. The Problem is, there are two interwiki links which doesn`t match. The German and the Dutch articles are describing only successfull summer gold medalists, the other ones summer AND winter athletes. I assume the interwiki links shall be deleted and somebody should create a new item especially for the German and the Dutch articles. --César (talk) 18:14, 30 June 2015 (UTC)
end of copy

✓ Done, new item just for Summer Olympic Games (Q159821) is now found at no label (Q15360322) (Czech Wiki already had one). - FakirNL (talk) 05:54, 1 July 2015 (UTC)

Refresh main page

Someone here that wants to refresh the content of the main page (regularly)? It is a bit static now, a new highlighted WikiProject and a new example of a statement will help a lot. Sjoerd de Bruin (talk) 14:22, 1 July 2015 (UTC)

Cycling teams

Hi, it is again me. Today, I work about cycling teams. My example will be Russian Helicopters (Q15332177) (fr:Équipe cycliste Russian Helicopters in French). I am working on the fr:Module:Infobox/Équipe cycliste who will take datas from Wikidata.

@Jérémy-Günther-Heinz Jähnick: Classes are organized hierarchically. I think any professional team, whatever the sport, should be a subclass of the professional sport team class. It's doable doing something similar than {{SuperclassTree}} does, and a test to see if professional sport team in in the tree, with arbitrary access.
My proposition :
This makes the Russian Helicopters 2014 (Q18528960) automatically an instance of professional sport team.
Additionally I would say
< Russian Helicopters (Q15332177) View with Reasonator View with SQID > instance of (P31) View with SQID < Cycling team franchise >
. Instances of Cycling team franchise would then have a sponsor, a brand name (american style naming, I don't know how to name it in french). TomT0m (talk) 20:09, 2 July 2015 (UTC)
The final instance of statement isn't necessary that TomT0m has above. --Izno (talk) 21:04, 2 July 2015 (UTC)
@TomT0m: c'est un peu hard à comprendre, encore plus en anglais, mais je saisis le truc. Je viens de créer professional cycling team (Q20639847) et club cycling team (Q20639848), puis professional sport team (Q20639856) et amateur sport team (Q20639857), j'ai ensuite fait le lien avec subclass of (P279). Enfin, j'ai testé avec succès {{SuperclassTree}} sur la page User:Jérémy-Günther-Heinz Jähnick/Cyclisme.
professional cycling team (Q20639847) et club cycling team (Q20639848) serviront un peu plus tard sur la Wikipédia francophone, lorsque l'arbitrary access aura été activé. Est-il possible de dater ces informations avec start time (P580) et end time (P582) ? Certaines équipes amateur sont devenues professionnelles (c'est toujours un gros bordel vu qu'il y a toujours des cas particuliers). Utiliser has part (P527) pour lister les saisons d'équipes est-il correct (Russian Helicopters (Q15332177) > has part (P527) > Russian Helicopters 2014 (Q18528960)) ?
Est-ce qu'il existe une propriété proche de topic's main category (P910) qui pourrait me permettre de définir la catégorie d'une équipe (UCI Continental Team (Q1756006), UCI Professional Continental Team (Q382927), UCI ProTeam (Q20638319), UCI WorldTeam (Q6154783)), en jouant avec start time (P580) et end time (P582) comme qualificatifs vu que l'équipe peut changer de niveau, et qui pourrait également permettre de qualifier 2015 UCI Europe Tour (Q18342122) (une course peut être de catégorie 1.1, 1.2, 2.1...). Sinon, bonne nouvelle, j'arrive maintenant à remplir de manière assez complète les informations sur une équipe cycliste, je vais demander la création d'une propriété code UCI pour être complet. Jérémy-Günther-Heinz Jähnick (talk) 07:44, 3 July 2015 (UTC)
@Jérémy-Günther-Heinz Jähnick: En fait c'est gérable aussi avec instance of (P31) et subclass of (P279), cette histoire de niveau, au moins en ce qui concerne la classification des équipes. Par exemple si l'équipe des Russian Helicopter 2014 est une World Team, tu mets simplement
< World Team > subclass of (P279) View with SQID < équipe cycliste pro >
et
< Russian Helicopter 2014 > instance of (P31) View with SQID < World Team >
. Pour dire que World Team est un niveau de classification des équipes cyclistes, on indiquerait
< World Team > instance of (P31) View with SQID < niveau d'une équipe cycliste >
. Pour chercher tous les niveau, on ferait une reqûete du type Wikidata Query. Pour lister les niveaux d'équipe, j'aurai tendance à dire dans ce cas là qu'il vaut mieux utiliser une requête, ou coder un truc avec l'accès arbitraire : Wikidata Query, le tout trié par date de la saison ou date de début. Pour un tutoriel sur la classification, j'ai écrit Help:Classification (en anglais aussi, mais j'ai pas envie de traduire tant que j'ai pas un appui communautaire). Dans l'ensemble je pense que des exemples comme le tiens montrent que ça tient la route :) TomT0m (talk) 14:10, 3 July 2015 (UTC)
@TomT0m: j'ai un exemple concret en tête : il y a deux mois, j'ai travaillé sur les Quatre jours de Dunkerque 2015 dont l'élément Wikidata est 2015 Four Days of Dunkirk (Q18589873). Depuis le 1er juin, il est possible de lister les équipes participantes, ce que je viens de faire. On a donc sur Wikidata la liste des équipes et sur Wikipédia, une fonction de l'infobox calcule automatiquement le nombre d'équipes (une fonction similaire dénombre les étapes pour l'anecdote). Mon objectif est de parvenir à lister les équipes en différents tableaux suivant leur division, non seulement parce que c'est long de toujours faire des copiés-collés entre les articles pour récupérer les bonnes ligne, ou retrouver le pays d'une équipe, mais aussi parce que c'est bête de ne travailler que pour nous francophones (le travail de saisie de l'un devrait profiter à plus de monde). Je me doute que pour mettre ça en application il faudra l'arbitrary access sur la version francophone, mais il est prévu pour bientôt. Ça ressemblerait à quoi une ligne de programmation ? Grosso modo, pour une équipe Topsport Vlaanderen-Baloise 2015 (Q18746658), des informations telle que sa division, son code UCI et son pays devraient plutôt être centralisées sur Sport Vlaanderen-Baloise (Q135701), éventuellement avec start time (P580) et end time (P582) si des changements interviennent, pour éviter de dupliquer les informations, tandis que Topsport Vlaanderen-Baloise 2015 (Q18746658) listerait notamment les coureurs et les victoires. Ça paraît assez difficile question programmation (surtout que je ne m'y connaît pas trop, même si je commence déjà la base avec Lua), mais à l'usage ce serait enfantin, puisqu'inclure quelquechose comme {{Cycling teams listing}} (en anglais, pour passer simplement d'une version linguistique à une autre) rendrait immédiatement les tableaux avec les équipes listées. Suivant la course, ça ne prend que trois à quatre minutes (même si je comprends que les informations devront être remplies de manière uniformes, mais c'est déjà le cas pour les éditions de courses cyclistes). Jérémy-Günther-Heinz Jähnick (talk) 15:36, 3 July 2015 (UTC)
@Jérémy-Günther-Heinz Jähnick: La division, c'est calculé comment ? c'est pas constant sur une saison ? Dans ce cas ce serait mieux de rattache la division à l'équipe saisonnière.
Sinon il n'y a pas que le lua, il y a les requêtes aussi. On pourra écrire des requêtes du style "récupère moi tous les items qui sont des équipes ont une propriété participe à [telle item de course], trie selon [un critère] et formate leur leur numéro et leur sponsor, à la manière d'autolist2 et surtout avec überlistet en attendant : qui peut générer ça : https://en.wikipedia.org/w/index.php?title=User:Magnus_Manske/listeria_test Donc il faut plutôt dans ton cas se concentrer sur comment on écrit des requêtes (même si ça va changer :) ). TomT0m (talk) 16:49, 3 July 2015 (UTC)
@TomT0m: Globalement, le niveau reste le même au fil des saisons, par exemple cette petite équipe est de troisième division depuis sa création 2006. La majorité des équipes restent en troisième division (ou amateur) depuis leur création, il arrive parfois que certaines perdent leur statut, mais c'est rare (difficultés financières ou affaires de dopage). Il est possible d'indiquer la division aussi bien dans l'élément de l'équipe que dans celui de sa saison 2015, comme c'est également le cas pour le code UCI ou pour le pays. C'est pour ça que ma préférence va quand même à l'inscription dans l'élément de l'équipe. L'argument principal étant que c'est plus facile question maintenance pour mes collègues Wikipédiens qui ne connaissent pas trop Wikidata. Je suppose que remplacer instance of (P31) > cycling team (Q1785271) (comme des contributeurs ont écrit) par subclass of (P279) > UCI Professional Continental Team (Q382927) permet de générer une arborescence bien précise (et un bon rangement), tel que visible ci dessous. Même si tu as raison sur le fait que dans mon cas on doit se concentrer plutôt sur la rédaction des requêtes, je dois aussi veiller à établir une utilisation précise des propriétés, parce que si ça se mélange, les résultats ne sortiront pas comme il le faut. Jérémy-Günther-Heinz Jähnick (talk) 17:54, 3 July 2015 (UTC)

Vanishing property labels

This evening, when I open an item page, and I scroll down the page, the Property labels scroll to the right, until complete disappearance, which makes it rather difficult sometimes to guess what property is displayed.

Did someone else have the same problem ? was there some kind of GUI update recently ?

Thanks for your help :/ --Hsarrazin (talk) 19:39, 2 July 2015 (UTC)

confit/confiting

Do these really deserve two separate pages(?)? It seems like the Swedish wikipedia page wp:sv:confitering and wp:en:confit talks about the same thing. 79.136.60.114 21:48, 2 July 2015 (UTC)

Not my kind of subject. But there are two fr-articles: fr:Confiserie (technique) and fr:Confit here. So these items cannot be merged, but sitelinks can always be moved. -- Innocent bystander (talk) 06:13, 3 July 2015 (UTC)

What's a difference?

stated as (P1932) and named as (P1810) — what's a difference? -- Sergey kudryavtsev (talk) 07:40, 3 July 2015 (UTC)

It looks like they were created with different purpose, but they look mergeable. Start a thread at PfD and see what more people think. -- Innocent bystander (talk) 10:09, 3 July 2015 (UTC)
Not sure if that is a good idea. This chat or the properties talk page seems the better place.
P1932 seems for misspellings or at least peculiar spellings and the like, P1810 for artist's names. --- Jura 10:15, 3 July 2015 (UTC)
How the code in the present Module:Wikidata handle the time-properties, proofs that two properties with almost the same purpose is not a big problem here. -- Innocent bystander (talk) 14:14, 4 July 2015 (UTC)
If it was just that code ;). After some thought, feel free to merge them.----- Jura 14:36, 4 July 2015 (UTC)

user talk page bot archiving available?

Is anyone running the pywikibot archiving tool locally for user talk pages? Thanks.  — billinghurst sDrewth 01:23, 4 July 2015 (UTC)

Not that I know of. My user talk page hasn't really required archiving as far as I know.--Jasper Deng (talk) 06:56, 4 July 2015 (UTC)
User:Hazard-Bot/Archiving ? Thibaut120094 (talk) 10:15, 4 July 2015 (UTC)
Bot is inactive. Sjoerd de Bruin (talk) 10:36, 4 July 2015 (UTC)

Business contact information

I'm looking for non-copyrighted business contact information (phone number, email, website, etc) for as many American businesses as I can. Is this the right place to get this info and, if not, where would I find this. Thanks much.

No, Wikidata contains only important companies. Maybe you should try OpenStreetMap. But their data is copyrighted, it is licensed under the ODbL which requires for example attribution. I don't know if there are any other services. --Archer90 (talk) 14:12, 4 July 2015 (UTC)
White pages? --- Jura 14:27, 4 July 2015 (UTC)

Abbeys (again)

Hi I am back with another abbey question, this time about heritage names regarding religious orders. Please can someone tell me the difference between monastery (Q44613) and abbey (Q160742)? Also, how do these relate to institute of consecrated life (Q225446)? Thx in advance, Jane023 (talk) 11:36, 28 June 2015 (UTC)

Hm, how about: "An abbey is a monastery called 'abbey'. A monastery is an abbey called 'monastery'"? --- Jura 12:13, 28 June 2015 (UTC)
abbey (Q160742) is a special kind of monastery (Q44613). To create an abbay is necessary an official act from Holy See (Q159583) --ValterVB (talk) 13:06, 28 June 2015 (UTC) PS Source: it.wikipedia
enwiki adds some non-catholics: en:List_of_monastic_houses_in_England. --- Jura 13:18, 28 June 2015 (UTC)
To add to what Valter said: monastery (Q44613) is a much wider concept. There are Buddhist monasteries and Hindu monasteries, for example. An abbey (Q160742), however, is something exclusive to some variants of Christianity. Gabbe (talk) 17:12, 28 June 2015 (UTC)
Thanks for the quick responses! But just as in these days where the old Catholic churches are being turned into books stores and mosques, I am looking for a proper heritage term and am not particularly interested in which religious organization ordained what. Would it be possible to make all of these a subclass of something like "religious commune"? Because then I will just create the item if it doesn't already exist yet. Or do we say that an abbey is an abbey (with start and end dates) and a monastery (with start and end dates) and a nunnery (etc) ? Jane023 (talk) 19:13, 28 June 2015 (UTC)
The answer above is very clear. Abbey is a subclass of monastery is a subclass of organisation. If the organisation changed its status then record the status at different times with start/end dates. If you are talking about the building then abbey and monastery are both subclass of religious buildings. If you are not sure whether you are talking about a building or an organisation then look at the statements you want to make about the item. Are they statements about the organisation or about the building? if they are about both then rewrite them so they are only about one (Abbey (building) used by Dominican order from start date to end date or alternatively Abbey (organisation) geo coordinates from start date to end date). or split into two items. Follow the statements. they will tell you what you need to do. Filceolaire (talk) 01:54, 1 July 2015 (UTC)
Thanks for the answer! I am dealing with old heritage building ensembles that were at some point all of these things (nunnery, farm, monastery, abbey, etc). Do you mean to say that Abbey (building) is the same item as Abbey (organization)? Is it OK to have an item be an instance of building and of organization? Thx. Jane
I'm afraid not Jane. An item can be about the building or about the organisation but not about both. Statements linking the two can often be written either way (e.g. either as <organisation> had <residence>:<building> or <building> had <occupant>:<organisation>) but most statements apply to the building or to the organisation but not to both. Items about heritage buildings are probably about the <building>. Use occupant (P466) or use (P366) to list the various uses of the building, creating items for any especially interesting occupants for which there is good information available. Hope that helps. Filceolaire (talk) 01:49, 5 July 2015 (UTC)

Interesting! I guess this makes sense. I am working on a list of items that are pretty diverse in that some have practically no statements at all, and others have lots and lots of statements (abbeys that are part of a cathedral complex). Do you have a few examples that show abbey-building vs. abbey-organization? thx for your time on this. Jane023 (talk) 06:45, 5 July 2015 (UTC)

if the Earth is 6000 years old......

I came across an online copy of Bishop Ushers Annals Of The World which is the book, published in 1658, that worked out that the world was created on 23 October in 4004BC.

While it is undoubtedly wrong about the date of the big bang it is does give definitive dates for the births and deaths of each of the "begats" from Adam to Jesus. (I didn't say right, I said definitive). I think we should add these dates to all those biblical characters with the dates referenced to this book.

OK? Anyone want to help? Filceolaire (talk) 23:51, 1 July 2015 (UTC)

Hm, I had planned to add many of birth and death dates to these items in AM (which is very inconsistently converted to Gregorian) once the time datatype is extended to allow for other calendar systems. Add the calendar issues to the different opinions by different religions and religious sects, and these properties are likely going to have a very large number of values... --Yair rand (talk) 04:31, 2 July 2015 (UTC)
And some of those statements would tell that the heaven and earth was created in "the beginning" without having a specific date for it. Looks like a challenge to make a datatype that can handle that! :) -- Innocent bystander (talk) 06:59, 2 July 2015 (UTC)
Wikidata is broken for dates before AD 1, because the data model says the year before AD 1 should be entered as 0000, but the user interface enters it as -0001. I think there should be a moratorium on entering dates before AD 1 until Wikidata is fixed. Jc3s5h (talk) 07:29, 2 July 2015 (UTC)
According to age of the Earth (Q935310) Earth (Q2) was inception (P571) 4540 million years BCE. But this statement is lacking a reference. Giving James Ussher (Q333481) credit to him for the first estimate would be fine with me. But make it deprecated. --Succu (talk) 21:04, 3 July 2015 (UTC)
We could add determination method (P459):Bible (Q1845) as a qualifier; but not until BC dates datatype have been sorted. Filceolaire (talk) 23:00, 3 July 2015 (UTC)
Both statements are a litte bit different:
James Usshers date is a „real“ date on which earth was created. So inception (P571) fits.
The statement in Earth (Q2) should be read as 4540 million years old. This is more like a time span and not a (more or less exact) point in time. So inception (P571) fits not very well.
--Succu (talk) 18:45, 4 July 2015 (UTC)
inception (P571) takes a Wikidata time type as a value, but that is a calendar date, which would be either the Julian or Gregorian calendar. Both of these calendars count actual sunrises. But the days were much shorter in the distant past (although the ability to reconstruct the day length at any given time is limited). Methods such as radioactive decay that are used to measure the age of the Earth give results in years of constant length, just as the atomic clock time scale International Atomic Time or the ephemeris time scale based on the independent variable in the laws of physics that determine the orbit of planets are years of constant length. To the best of my knowledge, Wikidata does not have a data type for durations of atomic time (or equivalent), nor is there a property to give the age of something in atomic time. Thus Wikidata has no property with which to state the age of the Earth. Jc3s5h (talk) 19:59, 4 July 2015 (UTC)
All these corrections of the atomic timeline inserting a leap second (Q194230) are really uggly. :) We are awaiting the datatype quantity with unit. And new properties using this. Earth (Q2) will get two different claims. --Succu (talk) 20:13, 4 July 2015 (UTC)
What would the two claims be (in general terms, since the data types or properties don't exist yet)? Jc3s5h (talk) 20:43, 4 July 2015 (UTC)
I'm sure you'd better served with item datatype .. --- Jura 20:47, 4 July 2015 (UTC)
Claiming to be 4540 million years old does not need a calendar. Only an implementation of time (Q11471) as a quantity with unit. --Succu (talk) 21:04, 4 July 2015 (UTC)

Adding television series episodes

So here's the page for "Star Trek: The Original Series":

https://www.wikidata.org/wiki/Q1077

There's a link on that page to "list of Star Trek: The Original Series episodes":

https://www.wikidata.org/wiki/Q1257895

However, I don't see any actual episodes referenced in the list. :-)

Can someone point me to an example television series which properly lists episodes?  – The preceding unsigned comment was added by ‎Dharmatech (talk • contribs).

Wikidata:Showcase items doesn't have one yet. Maybe there is one on Special:WhatLinksHere/Q1983062
To see what properties others use, try http://tools.wmflabs.org/wikidata-todo/related_properties.php?q=claim[31%3A1983062]
Not sure if it's optimal, but you might find it interesting: https://tools.wmflabs.org/reasonator/?q=Q317766 --- Jura 21:17, 4 July 2015 (UTC)

Wikidata weekly summary #165

Atheism

The use of religion=atheism has - rightly - been deprecated on en.Wikipedia; yet we have cases of religion (P140)=atheism (Q7066). How can we best avoid this? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 18:34, 28 June 2015 (UTC)

Replace with lifestyle (P1576), run a bot every month and tell people about it. Sjoerd de Bruin (talk) 18:46, 28 June 2015 (UTC)
It sounds strange to me to describe atheism as a lifestyle and I'm also not sure why we would want to separate it from the religion (P140) property either. To me it seems like it should be religion (P140) = no value (with some sort of qualifier if we want to describe the exact form someone's lack of religion takes). - Nikki (talk) 19:23, 28 June 2015 (UTC)
Atheism is not a religion. (It has been said that "atheism is no more a religion than 'not collecting stamps' is a hobby"). Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:53, 28 June 2015 (UTC)
Debatable, but assuming you're correct, I could point out that anarchy is not a mode of government, but I could totally see adding it as one here on Wikidata. --Yair rand (talk) 21:04, 28 June 2015 (UTC)
You're welcome to read the debates in en.Wikipedia; I don't intend to rehash them here. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 23:04, 28 June 2015 (UTC)
@Pigsonthewing: enwikipedia may have any opinion they like. They can (and probably should) set a filter to the templates that import such statements, instead of trying to influence their opinions to other projects. -- Innocent bystander (talk) 02:38, 29 June 2015 (UTC)
I'm not aware of "enwikipedia trying to influence their opinions to other projects"; however I do believe that the points made in the discussions I mentioned apply more widely. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 09:21, 29 June 2015 (UTC)
Andy, please provide the location of that discussion. Saying "en deprecated this use" is unhelpful without reasons why and since you are electing not to supply any, your request for change is unhelpful. --Izno (talk) 13:27, 29 June 2015 (UTC)
That's still weird: Who says atheism (Q7066) is a "lifestyle" either? What would an "athiest" lifestyle look like? (Same goes for agnosticism (Q288928).) We should be asking ourselves how these claims are relevant to the items they're being added to; that would probably help decide what property it's about. There doesn't seem to be a property for belief system (Q5390013); maybe religion (P140) should be about belief system (Q5390013) instead of religion (Q9174). Anything of class religion (Q9174) would be a subclass of belief system (Q5390013), right? --Closeapple (talk) 02:30, 29 June 2015 (UTC)
There is some inconsistency here, yes. We recently had a long discussion (on several places) if the Catholic church was a religion. I do not know if we came to any conclusion in that case. Split this property makes it even worse, I'm afraid. I would say that the lifestyle of an atheist is "materialism". But I am not aware of that we have any item that describes that word it in the way I intend it here. And that statement today have only me as source. Some athiests do not like to be attached to the word religion at all, and there can be doubts if it really is. But can an athiest be "religious" in the same way as we use the word for non-athiests? My opinion is that they can. Ricky Gervais (Q23517) looks to me as a passionate preacher of that religion. He spends a lot of time to tell others that "atheism" is the only true "faith". In that way, he is more religous than many true religous people around me. -- Innocent bystander (talk) 03:40, 29 June 2015 (UTC)
Add it as a constraint on P140 also. --Izno (talk) 19:13, 28 June 2015 (UTC)
@Izno: How would that be framed? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:53, 28 June 2015 (UTC)
Create a new property philosophy for person ? Snipre (talk) 07:04, 29 June 2015 (UTC)
I think the present constraints may take care of this. --Izno (talk) 13:21, 29 June 2015 (UTC)
We should rename religion (P140) to world view (Q49447), allow atheism (Q7066) as value and let enwiki to have any kind of filter they want to. I don't like second property option. -- Vlsergey (talk) 07:41, 29 June 2015 (UTC)
Procedural pings: these users have all edited Property talk:P140 regarding atheism and have not participated above. @Gloumouth1, Danrok, AS, Touriste, Sabbut, Wylve: @Infovarius, Tobias1984, Cbrown1023, Emw, fnielsen:.

My preference would be to redefine the property to be inclusive of atheism, as per P140's talk page. There are several suggestions there for such a renaming. --Izno (talk) 13:21, 29 June 2015 (UTC)

Symbol strong support vote.svg Strong support religion (P140) needs to be redefined. Call it "world view" or "holds belief system". I am not sure if that solves all the cases, but it would help. For example there are atheists that still are members of a church. But in that case member of (P463) seems more appropriate. Then there are also countries where you inherit your parents' religion on your birth certificate. So there was no choice there to join the religion. No idea how we should call that. - In any case this property should be used only if the person himself, or a notable source has spoken about that "belief system". No mass editing based on hunches. --Tobias1984 (talk) 14:37, 29 June 2015 (UTC)
The article world view (Q49447) doesn't say much if anything about religion. Also, religion (P140) = Catholic Church (Q9592) is not the same as world view = Catholic Church (Q9592). For one, Catholic Church (Q9592) is essentially an organisation, which is made up of people, many of whom have conflicting views. In any case religion (P140) = Catholic Church (Q9592) is not so correct either, the religion is Christianity (Q5043). We might claim member of (P463) = Catholic Church (Q9592), but how do we verify that given that many churches do not have a formal membership system? There are baptism records, but all that tells us is that the person was baptised. We can't make assumptions based on that. Danrok (talk) 16:25, 29 June 2015 (UTC)
@Danrok: Christianity (Q5043) is not a religion because e if this is a religion what is Catholicism (Q1841) or Protestantism (Q23540) ? And just think about the denomination of French Wars of Religion (Q673175) that subdivisions inside Christianity (Q5043) are more than details. Snipre (talk) 19:13, 29 June 2015 (UTC)
How about "Life stance"? Lymantria (talk) 17:21, 29 June 2015 (UTC)
More than a "life stance" it is a "spiritual stance", or as other said a "view". In that frame we could fit emergentism (Q3724133) or physicalism (Q269114) as valid options.--Micru (talk) 17:29, 29 June 2015 (UTC)
Current literal translation of P140's Russian label into English is "attitude(relation) to religion", so there is no problem with mixing atheism and religions. I guess there are few more languages, that already have solved this problem by slightly redefining this property. --Lockal (talk) 17:29, 29 June 2015 (UTC)
I completely agree with Lockal and Izno pov. If Atheism is not strictly speaking a religion, it is certainly a militant pov about religion, not a lack of religion, it cannot be simply indicated by attribute none.
I think we should redefine religion (P140) to accept atheism and agnosticism, which are positions regarding religion, and not simply philosophical pov. Why not add a definition religion or pov on religion (incl. atheism) ?
world view (Q49447) certainly cannot fit in French... it is a strict definition of the German-term :/ --Hsarrazin (talk) 18:42, 29 June 2015 (UTC)
Symbol strong support vote.svg Strong support rename religion (P140) to make it more inclusive, in order to avoid endless debates about some "world views" which could be considered as religions by some people and not by others. It is the same reason why discoverer or inventor (P61) was named "disovered by" only at the very beginning, and was generalized after. --Gloumouth1 (talk) 21:51, 29 June 2015 (UTC)

Atheism is representable with

< Richard Dawkins (Q44461) View with Reasonator View with SQID > religion (P140) View with SQID < no value Help >

. I think he can be defined as an activist, so I'd like a political conviction property to say he is active is the political atheism movement for example. TomT0m (talk) 16:39, 30 June 2015 (UTC)

+1. But stay the problem of agnostism. Religion = some value ? Snipre (talk) 18:28, 30 June 2015 (UTC)

no, the lack of religion is not atheism, or, more precisely : someone answering to "what is your religion ?" "none" doesn't mean the person would define oneself as atheist
no... Atheism is not "no value", it's "no God"... it's a direct and firm refusal of the existence of God/s which is much different from the mere indifference to religion. - it's as Symbol oppose vote oversat.svg Strong oppose of God, not Symbol wait.svg Wait (which would be Agnosticism) and not Symbol neutral vote.svg Neutral - which is the "no answer".
moreover, there is a strong problem of definition of atheism, according to who claims it : accusing someone of atheism (generally an insult - see Pascal) is not at all the same as someone claiming to be an atheist. If P140:no value could represent a position, it would probably be agnosticism rather than atheism. :/ --Hsarrazin (talk) 20:30, 30 June 2015 (UTC)
We should rely on definition, not on dubious analogies. We know what beeing theist is. Begin an atheist is the opposite of this position, that's all. Wikidata would not really know what to do with connotations, and insulting someone is not really a reliable source, as far as I can tell. What would be a plausible counter argument would be whether or no http://www.cairn.info/zen.php?ID_ARTICLE=DIO_205_0069 buddism is a theism (seems not) and if it is a religion. TomT0m (talk) 20:49, 30 June 2015 (UTC)
Atheism is a direct and firm refusal of the existence of God/s is a definition, as far as I know ;)
Religion is, by definition, a question of beliefs, conviction, and not a question of facts, and truths, even if believers are convinced that they detain Truth.
as such, the conviction that God does not exist is as much a faith that the contrary conviction.
I did not mean to insult anyone, I meant that we can only use atheist/atheism to describe someone only if s/he describes oneself as such… and people who describe themselves as atheists are strongly opposed to the existence of god/s, which is a position much stronger than having "no religion".
The Buddhism argument is right though… is buddhism a religion or a philosophy ? … many people would say it's NOT a religion, many others that it is. Why can't we simply enlarge the definition of religion (P140) to have all thems (atheism/agnosticism/buddhism) as accepted values ?
if religion is a valuable data to store about people, don't mix atheists (who have a determined opinion) with people who just don't care. - It's important that all opinions can be stored, not just considered as "no value". --Hsarrazin (talk) 21:52, 30 June 2015 (UTC)
I can see a need for two different properties here - one for "belief system" and another for the "church (organization) that a person is a member of". So a person can be a Baptist with allegiance to the Southern Baptists. Hope this helps. Filceolaire (talk) 01:39, 1 July 2015 (UTC)
The right way to do this is to use religion (P140) = no value. However you define atheism, this applies in every case. I´d define atheism as some kind of philosophy. And yes, there are (confessing or non confessing) atheists, who are members of churches or other organisations, some might even have leading positions. So you might have no religion, but be a member of Southern Baptists or Catholic church or whatever. So there must be a separation between personal religiosity and membership to an organisation.--Giftzwerg 88 (talk) 13:56, 1 July 2015 (UTC)
P140 = no value is not linked data, meaning its utility is much lower than if we correlate to the actual item representing atheism. --Izno (talk) 15:36, 1 July 2015 (UTC)
We should not worry too much about that as linked data evolves, and anyway Wikidata needs mappings to other models. TomT0m (talk) 16:17, 1 July 2015 (UTC)

I think changing religion (P140) label to something like "religion or belief system" - or introducing a new property "belief system" and putting novalue to religion (P140) for atheists both makes sense. Probably the latter even more sense, since one can explicitly be non-religious but not be an atheist. That would mean that religion (P140) = novalue is not by itself a description of an atheist, but that's why the second property can be useful. --Laboramus (talk) 02:07, 5 July 2015 (UTC)

Oppose – Firstly, who said that atheism is a religion? A religion is a belief of a person who wants to make a link between him and God. Atheism has no God. Atheism is just a belief. „Religion” is an word from latin („religare”- vb, I,→ means reconnection). So „religion” is a reconnection between man and God. I'm not interested by this topic but I hope that would not affect other Wikipedia languages. If for you atheism is an religion, that means that you have an American culturea and that discussion must be on en.Wikipedia not on wikidata. --Valimali67 (talk) 10:27, 5 July 2015 (UTC)

Wikidata:Verifiability

Since the stability of our information must be good in order for Wikipedias and other projects to trust our data, can we please ratify this now (if I see enough support here for the version as-is I will initiate an RfC; otherwise let's please work on drafting it)? This is badly needed both to create Wikidata:Living people and be more specific with Wikidata:Notability, both important policies. To this end I think it might also be important if a "citation needed" tag is possible for tagging dubious information.--Jasper Deng (talk) 02:01, 3 July 2015 (UTC)

Since we just had a similar discussion on the English Wikipedia, I would like to make a point here. There is information which actually does not need reliable sources. For instance, adding the property "Commons category" can be perfectly sourced as "imported from Commons", though Wikimedia Commons is normally nt a reliable source by our standards. Moreover, it can not be sourced in any other way for the majority of the properties. According to the current proposal, if interpreted literally, this info should be removed. I guess we must be very clear on the point that some information can not (and should not) be sourced in the usual way.--Ymblanter (talk) 06:02, 3 July 2015 (UTC)
I think we then need to make a distinction between "meta" information and "real" information where the former is derived from our projects' categorization system (sitelinks, categories, etc.) and the latter is derived from outside sources (such as someone's birthdate).--Jasper Deng (talk) 06:25, 3 July 2015 (UTC)
@Ymblanter: On the contrary, in that example I would say that Commons seems like a reliable source for the kind of claim it supports. In fact, isn't it the ideal reliable source? Gabbe (talk) 12:17, 3 July 2015 (UTC)
I actually agree with Jasper. It is a perfect "meta" reliable source, but in general it is not. Another example is authority control, which takes all the data from the same source and as such does not need a source for every item. Once these cases have been sorted out, I guess we can go on in the direction of acceptance.--Ymblanter (talk) 14:42, 3 July 2015 (UTC)
Not sure if any statement in the source section that is qualified with "imported from" should be considered as reliably referenced as such. At best, the statement doesn't need a reference (e.g. VIAF id) or it indicates where such a reference can be found. --- Jura 14:58, 3 July 2015 (UTC)
Absolutely not, most of these statements do need references and do not have any. My point is that we should learn to discriminate between properties which need to be reliable sourced (say by Wikipedia standards) and those where "imported from" is sufficient.--Ymblanter (talk) 18:28, 3 July 2015 (UTC)
Often, what has been imported from Wikipedia is wrong. I have changes many GeoNames ID (P1566) in the last days. When I have changed them, I have added stated in (P248) in the source-part instead of "imported from". Maybe we also should do that with statements that we have confirmed as correct? -- Innocent bystander (talk) 18:42, 3 July 2015 (UTC)
I suppose in theory you could list the geonames id (or worse, its url) a second time in the sources sections. This way, you would have an authoritative reference for the identifier. ;) This doesn't really advance us though. --- Jura 18:53, 3 July 2015 (UTC)
I have downloaded parts of the database, not the url. The url is not very helpful. -- Innocent bystander (talk) 19:03, 3 July 2015 (UTC)
BA candidate.svg Weak oppose it's too late now to enforce such limitation on Wikidata. Would it be enforce as rule we will need to stop use Wikidata as primary storage for non-meta information, take everything back to make sure it won't be deleted from user view, as it regularly happens with images on Wikimedia Commons due to enforced regulations. IMHO the best way to handle such non-source information is local policy like the one on German Wikipedia when local community can filter out all statements without sources by LUA, or, even stricter, to limit number of trusted sources. From other point of view in ideal database it shall be done (from the very WD beginning), may be even by automatically removing all statements from non-meta properties without sources, and may be from the very beginning "it's from enwiki" shall not be accepted by source. But again now it's too late for that IMHO. -- Vlsergey (talk) 08:02, 3 July 2015 (UTC)
Well, some statements that do not look like meta-statements, may in fact be meta-statements. London (Q84) P31:city P17:UK are som basic statements that defines the item itself, and separates it from the horse with the same name and from the city in Canada. It's the Wikipedia-articles that defines those things and it is the ok to have enwiki as a source for that very statement.
There is also some contradictions between our principles and the verifications. I here add "P131:Municipality of Whatever" and add source for that. Our principles tells me I should add the "lowest level of hierarchy" for the administrative division. There are lower levels than municipalities, but I have no source for that statement. This principle is also based on that P131 is considered as a transitive property. But it is only transitive as long as you limit it's use to nations with a simple hierarchy. -- Innocent bystander (talk) 10:36, 3 July 2015 (UTC)
Also all linked guidelines shall be copied from enwiki to wikidata or to meta. I don't like that fact that "common" project would rely on local community rules or guidelines. -- Vlsergey (talk) 08:02, 3 July 2015 (UTC)
  • Pictogram voting comment.svg Comment The general idea is sound, but many of the more specific parts of the current version of the page seem debatable. "Authoritative sources of information include: .. encyclopedias, .. guidebooks,". Even the CIA World Factbook sample in the introduction is sub-optimal. "Sources that are not usually authoritative" seems to write in a complicated way that sources should be public (not online). Not sure if the content beyond sections 1 and 2 is of much use. Details like "This property replaces P387, which is now obsolete." are probably pointless even on a dedicated help page. --- Jura 10:34, 3 July 2015 (UTC)
Pictogram voting comment.svg Comment I agree that we should aim to have references for every statement, but I couldn't support this the way it's currently worded, because it sounds very harsh, deletionist and nothing like how Wikidata actually works right now. Saying that we're going to remove almost all unsourced statements sounds like we're going to start ruthlessly deleting three quarters of our current data (according to [1]), when, fundamentally, I don't think unsourced statements are actually a problem. The very fact we can have references means we already know which statements have no source, so, in theory, there's no reason for us to ever delete unsourced claims because we can always just filter them out (if people want a way to only get referenced statements, we should make that possible). In practice, I think most people would agree that we should delete implausible statements unless they can be sourced, but for plausible statements, we should focus on finding and adding references (regardless of whether we were the original contributor or not), not deleting the statements just because the original contributor didn't add a reference. I think we should encourage people to add references by helping them understand why it's a good idea and by making it quick and easy to do, not by having policies which threaten them with having their contributions deleted if they don't. - Nikki (talk) 22:29, 3 July 2015 (UTC)

We need more than sourcing without automated checking it is worthless. --FischX (talk) 00:11, 4 July 2015 (UTC)

Pictogram voting comment.svg Comment I'm going to take the pragmatic view here and ask this: Is the lack of a Verifiability Policy an actual problem for Wikidata at the moment? Back in Wikipedia's early childhood (and to a lesser extent still today) people showed up at a specific article, saying stuff like "I know for a fact that the Earth is a cube and you can't prove otherwise!". This resulted in many lengthy and fruitless discussions about what "the truth" is in several topics. Replying that everything an article says has to be attributable to a reliable source helped quench many of those discussions. The Verifiability Policy on Wikipedia thus helps solve (or at least mitigate) an actual problem. So what exactly is the problem (real rather than hypothetical) that would be solved by having a verifiability policy on Wikidata? Gabbe (talk) 07:39, 4 July 2015 (UTC)

@Gabbe: Uhm, yes. For one, I have had to, in multiple instances, hide potentially libellious information about living people. It is a foundation-level principle that we have to do our job to protect them, and that means strong verifiability for information. Also, many content disputes (e.g. about whether to merge item x with item y) could be resolved more efficiently if sources could be checked. If we are to be a viable knowledge base, then we must commit to the verifiability of our information.--Jasper Deng (talk) 09:41, 4 July 2015 (UTC)
Jasper if we have libellous info with no reference then I hope your first step would be to look for a reference. This usually isn't hard - there is usually a reference in the wikipedia article that you can copy over. If No reference is found the I agree libellous info should be deleted.
For non-libellous info for which there are no references I would hope you follow a similar path - checking for sources and references and adding them to wikidata, thus making wikidata better. If a quick look fails to find a reference then that is not, in my opinion, enough to justify deleting the info. Deleting info which we have every reason to believe is correct just because it doesn't have a reference recorded in the appropriate place does not make wikidata better - it makes it worse because it makes it more difficult for other to improve.
Tools to make adding references are needed on Wikidata; at present it is a laborious process which often involves having to create new items for the book, it's author and it's publisher before you can start to enter the reference info, even if the info is stored in a wikipedia template. We can do better and when we do lots more references will be added.
at least that is how I see it. Filceolaire (talk) 01:18, 5 July 2015 (UTC)
@Filceolaire: I think you misunderstood what I meant by "libellious". "Libellious" here means unsubstantiated information that is hence truly libel. I'm talking about information that is specifically a gross violation of the principle of protection of living people; this categorically excludes information where appropriate sources are found.--Jasper Deng (talk) 09:32, 5 July 2015 (UTC)

Pictogram voting comment.svg Comment(apart from tl;dr) If you want referencing and verifiability, then you need to 1) make that evident on each WD Q-page with a standard linking on what is required and how to do it; 2) identify whether research is permitted as the WPs do not allow for research; 3) identify how research can be stated and what we are going to do about sources that are not within WD, and predominantly not there; and 4) make it easy to add multiple references to multiple items by a better means of data addition. … Such that if you are going to enforce a higher standard, and more work, then you need to make it easier/accessible.

I have added many data claims undertaken from research of published and unpublished sources, some freely available and some behind paywalls for lesser known writers at enWS. There I put my research hits/info on the respective Author_talk: ns pages for the person, but I do not normally add the reference tag here (too hard, too time-consuming). Where I do research for local Q-page, not associated with another site, then I will put the research on the Q_talk: page [2]. Hope that feedback is of assistance.  — billinghurst sDrewth 02:10, 5 July 2015 (UTC)

I agree with your points and think that a full RfC should be initiated to iron these out.--Jasper Deng (talk) 09:33, 5 July 2015 (UTC)

Store information for government hierarchy

Does wikidata store (or can it store) data on the government hierarchy to be used in Wikipedia infoboxes? I have been working on articles in the w:Bas-Rhin department (equivalent to state or province) and realized that in March 2015, the number of arrondisements (sub-unit of a department) were reduced from I think 7 to just 5. The boundaries were completely redrawn...it's not like the eliminated arrondisements were just divided between neighboring ones. Cantons which are mainly just districts for election purposes were also adjusted. The lowest government unit is the commune, which is basically a town and its surrounding rural area. There are 527 communes in Bas-Rhin...far too many to manually adjust. Plus, most articles are in 20+ Wikipedia languages! I don't know whether other deparments have undergone similar restructuring. On 1 January 2016, several regions of France (government level between national and department governments) will be merged. This will affect several thousand infoboxes for French communes, cantons, arrondisements, and departments on just the English Wikipedia.

I do not understand much about code and how to run bots, except that the purpose of Wikidata is to keep such relationships between articles across Wikimedia projects. Basically, can such relationships be kept and adjusted in the infoboxes across Wikipedia versions? I believe the French Wikipedia has already adjusted all of their infoboxes (I will check), so perhaps a bot could import the relationships from WP-fr, then another bot could adjust the infoboxes on WP-en and other WP languages? AHeneen (talk) 19:55, 3 July 2015 (UTC)

Hi AHeneen. The first step is to add the information to wikidata. An item for a 'department' can have multiple values for maps etc. each with start dates and end dates. The current value should be marked preferred.
If it is not practical to add the before and after data to the same item (because too much has changed) then a new item can be created for the new department, using 'replaced by' and 'replaced' to link one to the other (with qualifier 'applies to part' if needed).
Once the info is in Wikidata then it can be harvested in an info box. Where you have one wikidata item with statements for both before and after then the infobox in wikipedia will just get the preferred values (which should be the current values). There is some work going on at the moment looking at infoboxes where you have one wikipedia articles (say for the department both before and after) but two wikidata items (e.g. one for before and another for after). Hope this helps. Filceolaire (talk) 22:51, 3 July 2015 (UTC)
Thanks for the response. Is there a page on Wikidata for bot requests (to import the information) or do I need to make a request on Wikipedia? AHeneen (talk) 12:25, 4 July 2015 (UTC)
See Wikidata:Bot requests. - Nikki (talk) 12:40, 4 July 2015 (UTC)
Thanks. AHeneen (talk) 12:45, 5 July 2015 (UTC)

Manny Pacquiao (Q486359)

How do I remove unsuitable Aliases? I thought there was an ❌ to click on. Or is it the way the page is displayed on my iPhone? (I'm using desktop view not mobile.). Thanks. --Senator2029 (talk) 07:59, 5 July 2015 (UTC)

Just hit "edit" on the top of the page and remove them?--Ymblanter (talk) 08:05, 5 July 2015 (UTC)

AGROVOC FAO vocabulary

As part of Wikidata:WikiProject Food, I'm looking at controlled vocabularies, and the main one is produced by the FAO : AGROVOC.
It has 32000+ concepts in 21 languages. (http://aims.fao.org/standards/agrovoc/concept-scheme)
It's tightly linked with various sources Wikidata already uses, and I was wondering how to proceed to import a massive and diverse source.
You can find all the kinds of things described by AGROVOC, the databases it's linked to at: http://aims.fao.org/standards/agrovoc/linked-open-data

My understanding is that it would be a candidate for Mix N'Match, but there might be a dozen of properties to create before hand. Am I right ? Comments welcome. --Teolemon (talk) 12:53, 5 July 2015 (UTC)

Adding a vocabulary helps when the identifiers are "mixed and matched" certainly. How do you propose we benefit from it once we have done this? Thanks, GerardM (talk) 13:08, 5 July 2015 (UTC)
Yes it is a good idea..
according to http://aims.fao.org/standards/agrovoc/linked-open-data, the perfect "mix and match" might be quite consequent given the sheer amount of exact match connexions they have already made (RAMEAU, Library of Congress…). What we'd get is multilingual labels, multilingual aliases, ingredients, a hierarchy of food concepts, links to other authority control DBs.
Here's the cheese item for instance: http://aims.fao.org/skosmos/agrovoc/en/page/c_1507
We could also set other properties based on the concepts they have (see Top Concepts at http://aims.fao.org/standards/agrovoc/linked-open-data)
--Teolemon (talk) 13:30, 5 July 2015 (UTC)

WikiProject Linguistics

I have just created Wikidata:WikiProject_Linguistics, currently aimed at improving coverage of linguistic typology (Q192625), if anyone's interested. (incidentally, I've also created it if you're not interested Ü) Popcorndude (talk) 01:27, 6 July 2015 (UTC)

Creepy! I was thinking of starting WikiProject Linguistics a few hours ago and then I check here and someone else literally just did. :) I've created Wikidata:WikiProject Linguistics/Participants (added you to it too) and started Wikidata talk:WikiProject Linguistics. - Nikki (talk) 03:41, 6 July 2015 (UTC)

Apparently siWP has years as items, rather than co-joined

Someone may wish to double-check, though it seems that siWP has items for each year that are separate to the existing years. It maybe that Google translate does not do a good rendition and they are different, or it may be that they are duplicate. Anyway, someone more attuned may wish to review, and, if required, organise the merges.  — billinghurst sDrewth 03:17, 6 July 2015 (UTC)

Goal: Establish a framework to engage with data engineers and open data organizations

Hi, I would like to introduce you a quarterly goal that the Engineering Community team has committed to:

T101950 - Establish a framework to engage with data engineers and open data organizations

The problem is that we are missing a community framework allowing Wikidata content and tech contributors, data engineers, and open data organizations to collaborate effectively on this use case:

  • Open data organization has a subset of interesting data that could be used to improve Wikimedia wikis after being added to Wikidata.

The solution, we believe, is a basic framework agreed with the Wikidata community and documented, offering a process that addresses main questions and obstacles. Imagine a framework like GLAM applied to data.

If you are interested, get involved! We are looking for

  • Wikidata contributors with good institutional memory
  • people that has been in touch with organizations willing to contribute their open data
  • developers willing to help improving our software and programming missing pieces
  • also contributors familiar with the GLAM model(s), what works and what didn't work

See the initial announcement and its discussion in the Wikidata mailing list. You can subscribe to the related Phabricator task to stay up to date.--Qgil-WMF (talk) 13:26, 3 July 2015 (UTC)

Too early: WD doesn't have all datatypes and very few policies about how structure the data. It is not possible to have a discussion when important parts of the system are still in development. 141.6.11.23 15:25, 3 July 2015 (UTC)
VIAF seems to have that. --- Jura 14:37, 4 July 2015 (UTC)
It's all about iterations. It is probably too early to offer a full package that just works, but it is not too early to offer the best framework we can have today. By having a first and very small wave of open data organizations trying to play this game, we will be able to identify blockers and their priority (technical and also social), and we will be able to inform better Wikidata developers and the rest of the community.--Qgil-WMF (talk) 14:15, 6 July 2015 (UTC)

Biographical vs autobiographical memoir

There seems to be an issue brewing so I bring here the question of whether memoir (Q112983) should remain solely a subclass of autobiography. A memoir can be either biographical (written by someone other than the subject) or autobiographical (written by the subject), so it seems to me that the subclass should reflect this so connect data correctly. Take A Memoir of Thomas C. James, M. D. (Q19828336), for instance. It is a memoir, an account of a person's life written by either that same person, or another person close to subject. Therefore, it has an instance of memoir. As I had noticed that memoir was being defined on Wikidata as solely autobiographical, I attempted to remedy this a few different ways, all of which were reverted. However, while I'm against creating a new memoir item for non-autobiographical memoirs, if that's what needs to be done, then that's what I'll do. Thoughts? Hazmat2 (talk) 02:31, 6 July 2015 (UTC)

Hazmat2, Beware of arguing from dictionary definitions. Wikidata items are about concepts and can have labels in hundreds of languages and these may not match perfectly. The labels and descriptions try and describe the items but the statements about the items are based on the underlying concept, not on any one language label. Be very wary of changing the definition of an item by changing it's description without checking that the labels and descriptions in other languages and statements (including statements on other items linking to this item) are also changed to match.
The en:biography article does not specify that auto-biographies are not biographies. It is a general article about various types of biographical works. The en:memoirs article describes memoirs as a very specific type of (auto)biography. If you feel wikidata has a need for a class for biographical works which are not written by the subject then that is a new item and, as far as I can see it doesn't match many WP articles.
If you feel that the English language label "biography" is inappropriate for an item which is the class of all biographical works then discuss alternative English language labels for this item. Please do not change the English language description to match the English language label while ignoring all the other labels and all the ways the item is used in statements. I will ping Andreasmperu too. I hope this helps. Filceolaire (talk) 12:33, 6 July 2015 (UTC)
@Filceolaire: Sorry for the confusion; I was referring to changing memoir (Q112983) to include both autobiography and biography, or simply biography (which can include both). The other issues were already worked out. You bring up the good point of being inclusive to all languages, which I was reminded of. In English, a memoir is not strictly autobiographical. A memoir is less specific than that and includes people who write about others they have personal knowledge of. In the spirit of being inclusive to all languages, I think it's better to have memoir be a subclass of either biography or both biography and autobiography. Hazmat2 (talk) 15:00, 6 July 2015 (UTC)

mistake noticed on george steinbrenner entry (reinstatement...para 3). am new on computer and wiki. don't know how to edit it out.

i don't how this works as i am new on computer and wikidata. i don't even know how to 'send' this note but i can identify the error and hope it eventually gets read and corrected.

on the entry 'george steinbrenner (1930 - 2010) there is a mistake in "re-instatement and championship years" in para 3 it reads that in 1995 the yankees made the play-offs for the first time since 1981 (that part is correct) but then it goes on yo state that then they - the yankees won the world series in 6 games. the yankees were not in the world series in 1995 as they lost to seattle in the first play-off round, 2-3. it was in 1996 that the yankees defeated the braves in 6 games.

i could re-write the paragraph and correct the error but i don't know the wiki procedure on this. cheers, flynn e-mail: johnjflynn0@gmail.com – The preceding unsigned comment was added by Johnny finnegan (talk • contribs).

@Johnny finnegan: It sounds like you are talking about a Wikipedia article, in which case you want to go to the talk page of the article in question.--Jasper Deng (talk) 09:57, 6 July 2015 (UTC)
Comment copied to en:Talk:George Steinbrenner

Extract One Entity in json

Hi, I am a new user of wikidata and I wanted to extract data for some entities I found this page: https://www.wikidata.org/wiki/Wikidata:Database_download to get dumps and others extracts but the files are very big and I can't use them like this Is there a way to extract only one entity in json or rdf formats

regards,  – The preceding unsigned comment was added by Melbux (talk • contribs).

https://www.wikidata.org/wiki/Special:EntityData/Q20000000.json and Wikidata:Contact_the_development_team#Concept_URI --- Jura 12:09, 6 July 2015 (UTC)
See Wikidata:Data access for a bit more information about the first URL Jura mentioned. As I said on the second URL, it supports various RDF formats as well. I haven't used them so I don't know how well they work, but it supports the extensions rdf, n3, nt and ttl. - Nikki (talk) 12:14, 6 July 2015 (UTC)


Yes, it is what I want. Thx

Why can't fictional characters not have family?

Fictional characters aren't allowed to have a family according to the constraint violations for several properties. A character like Ender Wiggin (Q2846779) is in several reports, because he has a father (P22), mother (P25), no label (P7) and no label (P9). Either those properties should be removed (and we will lose information) or the constraints should be adapted to include fictional characters or we need separate properties of parents and siblings of fictional characters or we keep the current way it is and will unresolvable constraint violations. Mbch331 (talk) 08:03, 30 June 2015 (UTC)

@Ash Crow:: changed it Special:Diff/205932509 back in March. Maybe he can explain. --- Jura 09:31, 30 June 2015 (UTC)
When I run into issues like that I usually just change the constraint from class=human (Q5) to classes=human (Q5),fictional character (Q95074). Popcorndude (talk) 14:11, 30 June 2015 (UTC)
@Popcorndude: Thanks for the tip. Followed your advice and altered the properties. Mbch331 (talk) 14:34, 30 June 2015 (UTC)
Another option is class=person (Q215627) which includes humans and fictional characters and sibling duos and biblical characters (there is some dispute if biblical character is a subclass of human or of fictional human but everyone agrees they are a subclass of person). Person doesn't include thoroughbred horses however. I expect that someone will eventually dump the studbooks into wikidata so we will have to sort that. Filceolaire (talk) 01:28, 1 July 2015 (UTC)
No. --- Jura 07:34, 1 July 2015 (UTC)
The bible contains a continuum of characters, from Tiberius (Q1407) who I think nobody claims to be fictional, to characters that in the text itself is described as fictional, like "Wickedness", a women mentioned in Zechariah. -- Innocent bystander (talk) 14:10, 2 July 2015 (UTC)

On the subject of biblical humans, I was wondering about Abraham and the Story of Abraham. I think we need to separate the human aspects of biblical persons from their biblical stories, because if I am talking about a concept "Abraham takes a knife to kill Isaac and his hand is stayed by the Angel of God", then that is part of the Story of Abraham (as well as being part of Isaac's life). It can be created as a subclass of biblical art as a valid genre, and a subclass of bible stories going up to the Old Testament, but I don't think it is part of Abraham himself, though you might want to crawl up the tree of stories to get to Abraham. Has anyone thought about how to model that? Thanks in advance, Jane023 (talk) 12:56, 2 July 2015 (UTC)

Then the story of Abraham is … a story. It's main character is Abraham, the biblical character. I don't really if scientific history has anything to say about the historical figure of Abraham … TomT0m (talk) 14:00, 2 July 2015 (UTC)
But at the same time no scientific history can deny the existence of Abraham. A lot of persons don't let a trace of their existence in the history, does it mean that they are fictional ? Snipre (talk) 15:32, 2 July 2015 (UTC)
What's objective to me is that he's a biblical character, and nobody will deny that. Now for the non biblical part what matters is the other sources. Believers will believe he's real, so the instance of person claim would have to be sourced by the bible. Non believers will deny this claim and require another one. In the case we want to separate the character as it appears in the bible and the one as studied by historians, we will use fictional analog of. Sourced, why not, by the historians. This seems objective. Different pov, different sources. The rest will be an interpretation matter. TomT0m (talk) 16:05, 2 July 2015 (UTC)
except that Wikidata seems to use the VIAF definition of person i.e. person includes fictional characters (even funny animal (Q2369882)). 'Human' is the class wikidata use for persons whose existence is considered to be confirmed with other classes like 'Biblical character' and 'legendary human' for persons who might or might not have existed and 'fictional human' and 'mythical character' for persons we are pretty sure never existed. At least that is how I understand it. Filceolaire (talk) 01:33, 5 July 2015 (UTC)
↺ Not sure if you read the start of the thread: Mbch331 started this as the constraint only based on "person" didn't work anymore. AshCrow had changed the subclass on fictional character somewhen in March, breaking most constraints.
↺ Biblical characters still work as "person" is still there. --- Jura 09:47, 5 July 2015 (UTC)
@Jura1, Filceolaire: The constraint system can't really well account for POV discrepancies :/ The only mechanism that could be used for this is the exception mechanism. I think ideally what we would need is "guarded constraint" by source, for example "according to the bible, Any biblical character (that is said to be real in the bible) is a real character", that would check only the constraints on statements that are sourced by the bible ... @Lea Lacroix (WMDE): @Ivan A. Krestinin: @Denny: any opinion ? TomT0m (talk) 10:04, 5 July 2015 (UTC)
What world is used as basis of Wikidata? Is it real world, bible world or The Master and Margarita (Q188538) world? Behemoth (Q4759868) is real black cat in The Master and Margarita (Q188538) world. Additional problem is dividing historical persons. We have no enough information to determine its reality often. Additional question: is bible reliable source for statement "Abraham is real historical human in real world"? — Ivan A. Krestinin (talk) 20:15, 5 July 2015 (UTC)
"What world is used as basis of Wikidata" The sourcable world. If somebody said something, then Wikidata says he said that, Wikidata is not about truth ... Coherentism (Q1778809)
But lets return back to constraints. {{Constraint:Type}} can have multiple classed. For example please see Property talk:P9. no label (Q14943515) can be added as special class if it is not subclass of human (Q5) or fictional character (Q95074). Lets make our discussion more concrete. What exact constraints are unresolvable? — Ivan A. Krestinin (talk) 20:15, 5 July 2015 (UTC)
The problem is how to manage inconsistent claims actually, and different povs. In that view I think the philosophical view of coherentism Coherentism (Q1778809) (anything is true if it is consistent with someone's system of belief). As a consequence, there can be inconsistencies beetween claim, and a constraint who is correct in one set of belief (let's say a source or a family of source) but inconsistent with another set of statements. As a result the claims would violate the constraint, but that's normal because they belongs to another point of view of the world ... The problem I have with weakening the constraints by adding, for example both human and fictional humans to have a family is that it does weaken the constraint : a fictional human should have a fictional family, and a real human should have a real family. With the constraint as is, it would just allow a real human to have a fictional family and as such cross the boundary. TomT0m (talk) 20:37, 5 July 2015 (UTC)
Real persons need to be able to have a family, fictional characters need to be able to have family. It should however be impossible to mix the two of them (so fictional characters can only have family members that are also fictional characters and real people can only have real people as family members), if that can be translated into constraints that would be a good solution. Mbch331 (talk) 20:47, 5 July 2015 (UTC)
And, because everyone loves edge cases, what about things like historical fiction? How would we represent the family connections of a fictional nefew of Genghis Khan (Q720), for example? Given a sufficiently complex constraint system we could have human (Q5) must link to human (Q5)s and fictional character (Q95074) can link to either human (Q5)s or fictional character (Q95074)s, but only fictional character (Q95074)s are required to link back. Popcorndude (talk) 21:36, 5 July 2015 (UTC)
Regarding the edge cases: usually, the fictional character, if it is based on a real character, still gets an own item. Example: Gaius Julius Caesar (Q5517445) and Julius Caesar (Q1048). Then the fictional character can have their own statements, and is clearly distinguished from the real-world person.
Besides that: if the constraints do not support our use cases, then the constraints have to be amended - not the other way around. Constraints should now not be a way to introduce all the strictness that we intentionally have kept out of Wikidata's design. --Denny (talk) 15:04, 7 July 2015 (UTC)

Freebase identifiers are waiting for what ?

Hoi, I am really sorry to see that a multitude of Freebase identifiers are waiting for someone to come by and say that they are correct. It is bloody stupid. Who in his right mind wants to do that and WHAT is the added value of such folly? In my opinion all of them should be added. They are as likely as anything to be correct and a manual phase is unlikely to add any value. Thanks, GerardM (talk) 22:32, 4 July 2015 (UTC)

It's not only identifiers and they are not 100% correct, just like your edits in the past. A second opinion is always handy. Sjoerd de Bruin (talk) 22:55, 4 July 2015 (UTC)
There is nothing we have that is 100% correct. This kind of content is certainly as good as what we have. We can remove content where we deem it to be wrong easily enough. Professor for instance is NOT an occupation. It is an academic rank. GerardM (talk) 06:29, 5 July 2015 (UTC)
Incorrect example. "Professor" can be both occupation and academic rank in some countries (Russia for example). It should be different entities thought. -- Vlsergey (talk) 10:15, 5 July 2015 (UTC)
When in a language "professor" means teacher on a university, it does not make professor an occupation in English. Professor is a rank. It is not an occupation in the general sense of the word. In some countries, professor is reserved for universities in others it is not. Calling them an "educator" when they teach makes sense. Informing what university they are occupied with makes sense, informing when they made what rank of professor makes sense. Calling their occupation "professor" does not. GerardM (talk) 10:54, 5 July 2015 (UTC)
Already professor in Sweden (Q19788990) is a multi-purpose-item, that should be split in different items to be useful. -- Innocent bystander (talk) 12:05, 5 July 2015 (UTC)
Wikidata is not English only. Making an argument about word be ready for counterargument. "Professor" as occupation making sense in some languages. It also make sense to write it different in English. It also make sense to inherit such professor entity from "educator" (but not to replace with!). -- Vlsergey (talk) 13:51, 5 July 2015 (UTC)
Wikidata_talk:WikiProject_Freebase#Items_about_people:_Freebase. --- Jura 10:45, 5 July 2015 (UTC)
tl,dr candidate ... Your reference does not answer my point. The point is that we SHOULD include all these references without waiting for a human to "verify". GerardM (talk) 10:50, 5 July 2015 (UTC)
I'm not sure if you are referring to the (1) "primary sources tool", (2) a general import of statements, or (3) a mapping between WD Qid and FB identifiers?
The "primary" sources tool requires users to "verify" additions and adds statements as well as secondary and tertiary sources.
A more general import of statements could be done once the mapping at Wikidata:WikiProject Freebase/Mapping is deemed complete. (it maps Wikidata properties to Freebase "types"?). It should also improve suggestions by the "primary sources tool".
Both rely on a mapping between Qid and Freebase identifiers. According to Wikidata_talk:WikiProject_Freebase#Items_about_people:_Freebase, it doesn't seem part of the plan to improve this. Given that the percentage of mappings can be fairly low, this has a large impact on (1) and (2).
Obviously, at some point, one could just import remaining statements from Freebase into new items. --- Jura 11:01, 5 July 2015 (UTC)
Hmm, according to Wikidata_talk:Primary_sources_tool#3_x_primary_sources_gadget one now also gets suggestions for additional (3) based on Samsung's. --- Jura 14:22, 5 July 2015 (UTC)
The requirement for verification of the primary sources tool is a fallacy. It is just garbage inherited from Wikipedia think. Share and compare is more powerful and, it does not waste resources like this stupid requirement does.
Yes, this is a challenge to explain how this helps us grow our data quickly and reliably. Thanks, GerardM (talk) 13:04, 5 July 2015 (UTC)
Once a basic mapping is done, it might be worth working through a series of properties for all items of a given field. --- Jura 14:22, 5 July 2015 (UTC)
Existing errors in Wikidata are not an excuse for adding more errors. --Yair rand (talk) 17:37, 5 July 2015 (UTC)
I always "verify" Freebase identifiers. this exercise in futility does add no value. GerardM (talk) 00:57, 6 July 2015 (UTC)
I have just removed Freebase ids from the Primary Sources tool. According to the amount of data (around 3 millions ids) it was silly to wait for someone to review everything. My fault. I'm trying to figure out what are the best options to upload this mapping to Wikidata and I'll consult the community about it. What I'll do is to use this mapping (merged with the data currently in Wikidata) in order to feed the Primary Source tool without waiting for the possible importation of this mapping into Wikidata. I'll put this data in an other Primary Sources dataset (called something like "freebase-test") in order to be able to remove it quickly if the suggested data are too bad. Thank you all for beeping so interested in the Freebase data importation, Tpt (talk) 21:04, 6 July 2015 (UTC)

Wikibase Quality Extensions have launched

Now we can view the constraint report of an item, for example, Special:ConstraintReport/Q183.

See mw:Wikibase Quality Extensions for details.--GZWDer (talk) 16:52, 6 July 2015 (UTC)

How does one mark the "known exceptions", and is there a log for them? --Yair rand (talk) 18:31, 6 July 2015 (UTC)
Exceptions are added using the "exceptions" parameter of the constraint, e.g. see the unique value constraint on Property talk:P525 which then shows up in yellow as an exception on Special:ConstraintReport/Q117728. - Nikki (talk) 05:32, 7 July 2015 (UTC)
Please explain the results for Q2302314 I totally do not understand what it says and consequently I cannot use it. Thanks, GerardM (talk) 06:03, 7 July 2015 (UTC)
no label (Q2302314) has 3 statements with 11 constraints on them. 2 constraints on named after (P138), 6 constraints on country (P17) and 3 constraints on instance of (P31). For the definition of the constraints see the corresponding property talk pages. All 11 constraints on no label (Q2302314) are fulfilled therefore everything is green. --Pasleim (talk) 06:33, 7 July 2015 (UTC)
So in essence I was fed too much information. Arguably it is better to only provide info on errors. A catch all message when everything is green is imho to be preferred. GerardM (talk) 06:59, 7 July 2015 (UTC)
Ah, I was trying to add a property id and work through the list for that property. :) --- Jura 06:07, 7 July 2015 (UTC)
For P776 there is a single-value constraint. But in, for example Veda och Mörtsal (Q7917894) there is two values. But they are not contemporary and therefor we have preferred value to one of the them. It would be nice to have a "single-value in best value-constraint", instead of using exceptions in such cases. -- Innocent bystander (talk) 07:03, 7 July 2015 (UTC)
There are many "soft" constraints that are helpful to stop errors if you look at a list of 50 or 100 exceptions for a property used on 10000 or 100000 items, but aren't helpful if you look at any of the 50/100 items. --- Jura 07:11, 7 July 2015 (UTC)

Looking for a bot for localities of Mexcio

Hi! i am looking for a bot that could work on locality of Mexico (Q20202352). INEGI locality ID (P1976) has been created recently, thanks a lot to Pigsonthewing.

As preparation to add these, it would be nice to have a bot add labels. The label of the place should be the same in English and in Spanish. If Spanish is present, the bot could copy it to English.

Also if "sh" (Serbo-Croatian) is present (A bot in shWP added created a lot of article for these items), the bot here could copy the part before the comma to English.

Is there any bot specialized in labels? Eldizzino (talk) 20:56, 6 July 2015 (UTC)

Wikidata:Bot requests might be a better place for this. - Nikki (talk) 05:28, 7 July 2015 (UTC)
Amir runs a bot with some regularity that adds missing labels. GerardM (talk) 06:05, 7 July 2015 (UTC)

@Nikki, GerardM: - Thank you! I used Bot request, maybe later will also need Amir. Eldizzino (talk) 14:30, 7 July 2015 (UTC)

A comet who is also a planet ?

Paperoastro Sarilho1 YsogoCekli829 Manlleus Meodudlye, with only limited amount of time to spend in the foreseable future. Romaine - like adding observatory codes + mapping usage (like the *observatory codes) mikeu Jc3s5h VIGNERON Harlock81 Ptolusque Pictogram voting comment.svg Notified participants of WikiProject Astronomy:

Scientist discovered planets that looks like comets. [planet 1]

Could we say :

< GJ436b > subclass of (P279) View with SQID < comet (Q3559) View with Reasonator View with SQID >
 as well with
 ? The question is open :)

TomT0m (talk) 10:59, 26 June 2015 (UTC)

I have not read the source here yet, but You do not have to look outside the Solar system, to find such anomalies. Ceres is maybe both a comet and a dwarfplanet, since it looses mass. Some prefer the term "active asteroid" or "rock comet" instead. And Ceres is not the only case. There are several cases of official classification of both an asteroid and a comet. (Not to be confused with asteroid belt comet.) The boundaries between comets and other kind of solar system objects is not as sharp as many believe. The centaur-class of asteroids are often comets that have not shown any coma and tail yet. Since the orbit of centaurs isn't stable, it's only a matter of time until they come close enough to the sun.) And Damocloid asteroids are often considered as comets that have lost their tail and coma (i.e. dead comet). -- Innocent bystander (talk) 11:37, 26 June 2015 (UTC)
Nit: instance of (P31). Comeon, Tom. :( --Izno (talk) 14:14, 26 June 2015 (UTC)
And I think there is an item for Exocomet, somewhere out there! :) -- Innocent bystander (talk) 15:33, 26 June 2015 (UTC)
exocomet (Q2855300)! -- Innocent bystander (talk) 15:38, 26 June 2015 (UTC)
See no label (Q2868657). I know there is at least a handfull of comets that officially are described both as a comet and a minor planet/asteroid. There are also a group of objects that have the behavior of both, without "official recognition", one of them is Ceres. -- Innocent bystander (talk) 16:57, 27 June 2015 (UTC)

Izno, Innocent bystander, if you „know” references to scientific literature would help. --Succu (talk) 21:49, 7 July 2015 (UTC)

@Succu: I have no access to scientific literature today, but I will from later this year, when I will study halftime at the University. Other kind of sources can be found in articles like Rock comet (Q15949801).-- Innocent bystander (talk) 04:54, 8 July 2015 (UTC)
  1. http://www.tdg.ch/geneve/actu-genevoise/planete-chevelue-comete/story/13658952

calendar model screwup

Hi everyone,

I have some bad news. We screwed up. I’m really sorry about this. I’d really appreciate everyone’s help with fixing it.

TLDR: We have a bad mixup of calendar models for the dates in Wikidata and we need to fix them.

What happened?

Wikidata dates have a calendar model. This can be Julian or Gregorian and the plan is to support more in the future. There are two ways to interpret this calendar model:

  1. the given date is in this calendar model
  2. the given date is Gregorian and this calendar model says if the date should be displayed in Gregorian or Julian in the user interface

Unfortunately both among the developers as well as bot operators there was confusion about which of those is to be used. This lead to inconsistencies in the backend/frontend code as well as different bot authors treating the calendar model differently. In addition the user interface had problematic defaults. We now have a number of dates with a potentially wrong calendar model. The biggest issue started when we moved code from the frontend to the backend in Mid 2014 in order to improve performance. Prior to the move, the user interface used to make the conversion from one model to the other. After the move, the conversion was not done anywhere anymore - but the calendar model was still displayed. We made one part better but in the process broke another part badly :(

What now?

  • Going forward the date data value will be given in both the normalized proleptic Gregorian calendar as well as in the calendar model explicitly given (which currently supports, as said, proleptic Gregorian and proleptic Julian).
  • The user interface will again indicate which calendar model the date is given in. We will improve documentation around this to make sure there is no confusion from now on.
  • We made a flowchart to help decide what the correct calendar model for a date should be to help with the clean up.
  • We are improving the user interface to make it easier to understand what is going on and by default do the right thing.
  • We are providing a list of dates that need to be checked and potentially fixed.
  • How are we making sure it doesn’t happen again?
  • We are improving documentation around dates and will look for other potential ambiguous concepts we have.

How can we fix it?

Wikidata calendar model decision tree

We have created a list of all dates that potentially need checking. We can either provide this as a list on some wiki page or run a bot to add “instance of: date needing calendar model check“ or something similar as a qualifier to the respective dates. What do you prefer? The list probably contains dates we can batch-change or approve but we’d need your help with figuring out which those are. We also created a flowchart that should help with making the decision which calendar model to pick for a given date.

Thank you to everyone who helped us investigate and get to the bottom of the issue. Sorry again this has happened and is causing work. I feel miserable about this and if there is anything more we can do to help with the cleanup please do let me know. --Lydia Pintscher (WMDE) (talk) 17:37, 30 June 2015 (UTC)

Comments

So, should the start time for Universe (Q1) say "13798 million years BCE Julian" now? What would be the normalized ISO interpretation of this date? --Laboramus (talk) 18:43, 30 June 2015 (UTC)
I would not want us to try to be specific on such things. It will just invite the worst in many people and, it will be wrong given the next theory anyway. GerardM (talk) 19:23, 30 June 2015 (UTC)
Such things happen, datums and conversions between them are always difficult. A big thanks to Lydia and the team for fixing this bug, and for the how-to for our correction of the remaining errors. Jeblad (talk) 20:45, 30 June 2015 (UTC)
"Before 1583" is dubious, the Gregorian calendar was introduced later in many countries. In theory counting days as MJD could bypass the calendar issues on earth, but I guess that won't fly. For timestamps (PHP, W3C, etc.) I'd interpret yyyy-mm-dd as w:Proleptic Gregorian calendar for anything from 0001-01-01 to 9999-12-31, but that's only me. How about a required qualifier (Gregorian vs. Julian) for all dates before, say, March 1, 1923? –Be..anyone (talk) 23:21, 30 June 2015 (UTC)

So far I see no mention of the need to allow (require?) the editor to enter the before and after fields in the user interface, and no ability to enter the time zone. These abilities are necessary in order to change a day in New York to a day in Greenwich. Jc3s5h (talk) 04:51, 1 July 2015 (UTC)

  • Currently dates are stored in Proleptic Gregorian. Why? Because wikidata can not store date "29 February 1900" (it will be replaced by "1 March"). Will it be the same from bot/scripts point of view? I.e. will there be store model/format/interpretation change, or just user interface? -- Vlsergey (talk) 06:38, 1 July 2015 (UTC)
    • It is an interpretation change if you previously interpreted the calendar model to be about what the date should be displayed as. If you interpreted it as what the date is given in then there is no change in the interpretation and it'll just be a bit of user interface improvements and clarification. --Lydia Pintscher (WMDE) (talk) 22:39, 7 July 2015 (UTC)
  • I don't quite understand diagram "how to fix". Consider situation with ruwiki where dates were presented in articles and infoboxes but no direct reference was provided. It was okay for ruwiki to have data without source, it was okay to copy data to Wikidata, it was okay to remove data from wikipedia article because wikidata have it now and can provided it via LUA / infobox / templates. But according to diagram Wikidata is going to remove data because no source provided. I'm sure local community won't be happy with that. The same situation with replacement to Gregorian without calendar-mentioned-sources for dates after 1583. This will change date representation in articles. Local community will be really furious about this. Why don't we just mark those dates with some flag? For example . -- Vlsergey (talk) 06:51, 1 July 2015 (UTC)
    Actual deletion should only affect a rather small number of dates for all I can tell. For most of them we should be able to find a source or it can be assumed from context. --Lydia Pintscher (WMDE) (talk) 22:42, 7 July 2015 (UTC)
    "it was okay to copy data to Wikidata". This is not correct: WD always requires sources, it was just because some bots were doing some forcing to import data without sources that WD let them working. Here is a good example of what happens when you go ahead without enough reflexion. Local community took some decision but the problem is that we don't have one local community but some local communities and decisions of one local community don't engage the other ones unless it is a common decision. We have to find a solution but a common decision will have the priority. Snipre (talk) 12:21, 1 July 2015 (UTC)
    1. "WD always requires sources" -- it doesn't look like this is what really happens on Wikidata, you know. I see nothing even close to it. 2. Well, common decision != "wikidata decision" != "wikimedia commons decision". Anyway, I proposed a solution that allow not to break anything in any community (i.e. mark elements without sources and not to delete them). We will need some kind of "source required" flag anyway, why not to create it right now? -- Vlsergey (talk) 13:54, 1 July 2015 (UTC)
    We can't prevent contributors to add data without sources according to wiki concept, especially when they don't take the time to read the basics about WD, but this doesn't imply than WD has to accept these edits. You propose a solution but please don't start your proposition saying "local community won't be happy with that" or "local community will be really furious". Snipre (talk) 11:45, 2 July 2015 (UTC)
    Well, how can I help in situation when people seriously propose "delete claim" without thinking about local communities? Yes, ruwiki won't be happy about deleting claims. Yes, it will be furious if one start to delete information from infoboxes. Any kind of rule that implies deleting information that was accepted in Wikipedia (and was moved to WD as primary storage) will means turning away from using Wikidata as primary storage. Nobody at local wiki will take care of using "source-backed-(wiki)data" in 1% of infoboxes if all other infoboxes will need to preserve all local data, just in case they will be delete from WD. It would be much simpler not to use Wikidata for infoboxes at all. -- Vlsergey (talk) 12:01, 2 July 2015 (UTC)
I'm also against deleting useful informations from Wikidata just because it actually has no source. No one is forced to use such informations, it is very simple to filter out dates without sources. --Archer90 (talk) 17:55, 1 July 2015 (UTC)
We don't delete the dates because they don't have no source, but we delete them because we assume they are wrong because we don't have any proof they are correct. Snipre (talk) 11:45, 2 July 2015 (UTC)
"it has no sources" => "it may be wrong" => "delete it". I see no big difference here. -- Vlsergey (talk) 12:03, 2 July 2015 (UTC)
There is one difference: we know that the data import was not done with a strict rule for date conversion, so the problem is the calendar conversion and not the dates themselves. Even if I assume that the dates are correct, I have a problem to keep them in the new storage structure because I have to convert them. Snipre (talk) 15:22, 2 July 2015 (UTC)
Dates import by my bot was done with strict rule for dates conversion -- they are imported according to storage data model, i.e. in proleptic Gregiorian calendar. Of course, I do know several bot owners who did mistake importing Julian dates as Gregorian ones and vice versa. Still this not require any kind of cleanup-by-delete. Any project that want to "merge" with Wikidata will be able to resolve it on per-case basis by comparing it's dates with Wikidata and fixing discrepancies. We don't need to remove anything unless those discrepancies really found. From ruwiki side I can be sure that most of imported dates are correct, at least they are displayed correctly even with conversion. Also, as far as I understood Lydia, there will be no conversion, just UI fixes. -- Vlsergey (talk) 17:56, 2 July 2015 (UTC)


The flowchart should be modified. Looking at the Julian side of the chart, where the source does not specify a calendar, and the date in the source is the same as the date in Wikidata, we need to ask if the date is related to the Roman Empire, and if the date is before 1 March AD 8. If so, the date must be presumed to be in the actual Julian calendar (as observed by the people of Rome) rather than the proleptic Julian calendar. There is no precise conversion available, so the before should be 1 day and after should be 2 days.

If the situation is otherwise the same, but the date is before 1 January 45 BC, the date must be presumed to be in the Roman calendar, which is not supported, and there is no reliable conversion from the Roman calendar to the proleptic Julian calendar. These dates should probably have the precision changed to year and before = 1 year, after = 1 year. Jc3s5h (talk) 18:02, 1 July 2015 (UTC)

In these cases it would be nice if the reference could be given and if a quote (P1683) could be included as a qualifier with the wording of the source so the information is there if we get suitable properties later. Filceolaire (talk) 23:58, 1 July 2015 (UTC)
Any chance you can make an adjusted version of the diagram? The svg is on Commons. I am not sure I understand the changes you are proposing so can't adjust it. --Lydia Pintscher (WMDE) (talk) 22:46, 7 July 2015 (UTC)

What about dates with precision less than a year - such as decades, centuries, millions of years? For decades it may make some sense to set Julian when it's before 1582, but for a date with millions of years precision it doesn't really. Should we have some cutoff there? --Laboramus (talk) 04:55, 3 July 2015 (UTC)

Yes we already exclude anything that is precision year or less. --Lydia Pintscher (WMDE) (talk) 22:48, 7 July 2015 (UTC)

I suggest, first of all, to provide some dates as simple and correct examples. Thanks. --Accurimbono (talk) 09:27, 3 July 2015 (UTC)


So finally here we are. This problem about dates was reported in phab:T87312 (and then phab:T88437) back in January this year. I don't understand why it took six months simply to acknowledge that there is a problem: in these six months the problem just got bigger :(. Anyway, that's not important now. @Lydia Pintscher (WMDE):, can you please be more specific on this point:

  • "We are improving the user interface to make it easier to understand what is going on and by default do the right thing:" does this mean that if I insert a date prior to 15 Oct 1582, the UI will preselect "Julian calendar" as default?

Also please provide some simple example about how to "fix" a date. Thank you Candalua (talk) 11:16, 3 July 2015 (UTC)

  • Yes. That should select Julian by default.
    How to fix a date: Take a potentially broken date, walk through the flowchart and depending on what it tells you do one of the following: 1) leave it as is, 2) change the calendar model, 3) delete the statement. I will provide a diff link when we have marked all dates that need looking at. Will post about that in a few mins here. --Lydia Pintscher (WMDE) (talk) 22:52, 7 July 2015 (UTC)
  • In response to Accurimbono's request for a simple correct example, I will describe what I consider to be a correct example (but it is fictitious to protect the privacy of real people). The TimeValue would have the following fields:
  • Date = +2015-01-21T00:00:00Z (The "Z" is a lie.)
  • timezone = 360 (minutes)
  • precision = 11 (days)
  • before = 0 (the given time is at the beginning of the possible birth period)
  • after = 1 (the birth could have occurred any time during the 24 hour period beginning at midnight)
  • calendarmodel = URI for Gregorian calendar
  • Also add the reference: http://www.austindailyherald.com/category/neighbors/births/

Of course, it is not possible to enter this with the user interface in its current condition.  – The preceding unsigned comment was added by Jc3s5h (talk • contribs).

I'm asking for an working example, not a theoretical one. Somethink simple to make reference to.
Some examples relevant for Julian/Gregorian dates with a source.
Like the birth date of Petrarch (Q1401) i.e. "20 July 1304 (Julian calendar)" (Source is already in the property, Julian date to be checked) --Accurimbono (talk) 13:41, 3 July 2015 (UTC)
To the best of my knowledge, there are no working examples. Taking Accurimbono's suggested example, Petrarch (Q1401), and looking at the most recent edit of the birth date we see the following
  • Timestamp +1304-07-28T00:00:00
  • Timezone +00:00
  • Calendar Julian
  • Precision 1 day
  • Before 0 days
  • After 0 days
This claims that Petrarch was born at 00:00 hours July 28, 1304, Julian calendar, Universal Time, plus or minus 0 days. In other words, exactly at midnight. But the source is undoubtedly referring to the local time in the birthplace, Arezzo, where the numerical value of the time is about 1 hour greater than Universal Time. So the example does not agree with the source. If you're counterargument is that we really don't want to be that fussy, and people should just use common sense, I would say that (1) computers don't have common sense and (2) if we want to be less fussy, our data model should be drastically changed to reflect the level of care, or lack thereof, that we expect from editors who enter data. Jc3s5h (talk) 15:21, 3 July 2015 (UTC)
You are wrong. This doesn't claim that he was born at 00:00 because the precision is 1 day. It says nothing about the time, this is just the stored string value, could be anything or nothing, but this doesn't mean that he was born at 00:00. Don't talk about time when the precision is day. --JulesWinnfield-hu (talk) 19:41, 3 July 2015 (UTC)
JulesWinnfield-hu, have you read the [Data Model]? It says that "A TimeValue represents a point in time..." A day is not a point in time. Midnight at the beginning of July 28, 1304, Julian, Universal Time is a point in time. "After" is defined as "integer. If the date is uncertain, how many units before the given time could it be? the unit is given by the precision." "Before" has a similar definition. If before and after are both set to zero, there is no uncertainty and the value is exact. Jc3s5h (talk) 20:01, 3 July 2015 (UTC)
A day is a day and a point in time without beeing more precious. No second, minute, hour or timezone is involved, neither a uncertainty. Just a day with a month in a year. --Succu (talk) 20:12, 3 July 2015 (UTC)
Succu's concept of a day is completely incompatible with the data model. That user's concept of a day is something like "a calendar day as observed by the inhabitants of a particular place". Calendar days are conceptually so different from points in time that we would have to add a new data type to the calendar model and change everything in the database to the new data type. You just can't represent "a day named November 22, 1963 in the Gregorian calendar, anywhere on Earth" with the same data type that can represent "12:30 pm Central Standard Time (USA) on Friday November 22, 1963". Jc3s5h (talk) 20:22, 3 July 2015 (UTC)
You mean „completely incompatible“ with your interpretation? I did not mention the calendar model. --Succu (talk) 20:37, 3 July 2015 (UTC)
The meaning of data in a computer database is seldom self-evident. When a description of a data model is provided, the data should be interpreted in accordance to the model. Do you claim that the data currently in the database complies with the data model? Jc3s5h (talk) 21:00, 3 July 2015 (UTC)
Your implications are based on this data model? --Succu (talk) 21:29, 3 July 2015 (UTC)
Yes. Jc3s5h (talk) 22:00, 3 July 2015 (UTC)
In your interpretation from february 2015? --Succu (talk) 18:56, 4 July 2015 (UTC)
It's not just my interpretation; I believe it reflects the consensus of the discussions on the bugs listed at https://phabricator.wikimedia.org/T87764
Also, the edits are mostly just clarifications; the only real change is allowing Julian dates. Jc3s5h (talk) 19:30, 4 July 2015 (UTC)
Not? Reflecting an internal Phab discussion as the „truth“? Really? --Succu (talk) 21:46, 5 July 2015 (UTC)
Recording an event in time on a dayly basis is normal to most of us and reflected in common sources as well. Besides the apple, do you want to establish a timeframe within Isaac Newton (Q935) was accepted as a member of Royal Society (Q123885)? --Succu (talk) 22:13, 5 July 2015 (UTC)

Marking dates for checking

So far no-one has indicated a preference on my question if you'd like a list on a wiki page or a qualifier added to the statement with the date. I'll therefor go with what I think is easier/less work for you. I'll ask someone on my team to open a bot approval request to add qualifiers to the statements in questions saying something like "instance of: Wikidata date needing calendar model check". --Lydia Pintscher (WMDE) (talk) 22:57, 7 July 2015 (UTC)

Probably both would be useful. Can we add the instanceof as you said, and then generate a list from it (with name and label of the item and date(s) involved)? Then one can go through the list, and remove the instanceof once a date has been checked. --Candalua (talk) 10:14, 8 July 2015 (UTC)
Maybe we could also mark the statements as deprecated, so that they can no longer be used on clients. Matěj Suchánek (talk) 10:37, 8 July 2015 (UTC)
The skill set and tools needed to check tools include a knowledge of calendars, knowing which reference works are reliable, and access to suitable reference works. Knowledge of Wikidata is less important. The phrase "add qualifiers to the statements in questions saying something like" is meaningless to the kinds of people we need to fix dates (with some exceptions, of course). You would have to provide step-by-step instructions on how to use the Wikidata site to find the articles that need attention. If the process is complicated then a web page would be better.
Of course the other tool that is needed is a user interface that can enter a correct birth or death date, which the current user interface cannot do. Jc3s5h (talk) 11:40, 8 July 2015 (UTC)

calendar model screwup - even deeper (about time zone)

See Wikidata:Project chat/Archive/2015/06#Wikidata is incapable of representing birth and death dates where it says: "Maybe a solution would be that the interface allows to store the information that is found in the sources? UTC before that existed is a derived information." - That also includes that the calendar model might be unknown. Eldizzino (talk) 14:13, 7 July 2015 (UTC)

Using the example of User:Jc3s5h "+1304-07-28T00:00:00" is probably not found in the sources, is it? Eldizzino (talk) 14:19, 7 July 2015 (UTC)

And keep in mind: Storing an event that is fixed to a local time, as UTC with offset is not correct. Offsets can change, e.g. by announcements before the occurrence, or by discovering of new information about the "true" offset. Eldizzino (talk) 14:24, 7 July 2015 (UTC)

"Timezone +00:00" is not a legal time zone, but it is a zone of same offset (minute precision?) from a certain time (UTC?). Are seconds as offset not allowed? Eldizzino (talk) 14:25, 7 July 2015 (UTC)

As far as what one would find in a source, there is no need to find a publication that was written near the time of the event. I'm not positive about WikiData, but Wikipedia prefers that historical items be supported by fairly recent reliable sources, such as books and journal articles by historians. In such a source, dates will usually follow the convention that the year begins on January 1 (even if some other date was observed in the place of the event). If the Julian or Gregorian calendar was in force in the place of the event, English language sources will normally use whichever calendar was in force. I think this is true for other European languages, but I'm not certain. In most cases, the source will also specify the place of the event, so with some care, the offset from Universal Time can be determined.
The data model requires that the time zone offset be stored in minutes (an integer). When you see an offset written as 00:00, it means 0 hours and 0 minutes. In places where time zones were in force, they (almost?) always were a multiple of 15 minutes, and usually an even number of hours. In the time before time zones, it's easy enough to figure out the offset to within a few minutes, but it would be difficult (and probably not useful) to figure it out more accurately than the nearest minute. Jc3s5h (talk) 15:38, 7 July 2015 (UTC)

Quality metrics

Hey folks :)

One of the major points that came out of the RfC about Wikidata usage on German Wikipedia and many similar discussions on Wikipedia is the uncertainty about how good the data in Wikidata actually is. Editors on large Wikipedias often get a negative view based on seeing "imported from some Wikipedia" or no references. It needs to be addressed because it is a major issue for us in terms of having our data actually used. We are already addressing this on many fronts both in editing and development work but we need to show that it is actually working to increase trust in our data. To do this I'd like to spent some time on getting metrics we can use both for ourselves and in communications. I think the 3 following metrics can give us a good picture of the situation:

  • percentage of statements with a source that is not another Wikimedia project (We already have that at https://tools.wmflabs.org/wikidata-todo/stats.php thanks to Magnus. I believe this metric is extremely important right now but will become much less important in the future.)
  • percentage of statements which fail a constraint check
  • percentage of items with a quality score higher than X for scoring system Y (We'd look into various options for scoring items. The first one I'd look at with a researcher is taking a number of items and feeding them into a system like Mechanical Turk to get them checked this way.)

Does this make sense to you? --Lydia Pintscher (WMDE) (talk) 23:20, 7 July 2015 (UTC)

  • For me the issue is not trying to prove to wikipedia editors that our data is good quality. The objective should be to create a workflow where it is a lot easier for a wikipedia editor to add references from a wikipedia article to a wikidata item so wikipedia editors can improve our data until it is of the quality they need. Clicking on an infobox should let a wikipedia editor edit the corresponding wikidata item, including adding references, even if this means creating new items for reference works. Most de:WP editors won't be interested in doing this but some will, if we make it straightforward for them to do it. At the moment adding a reference to wikidata is too much work so I spend my time doing other stuff. Filceolaire (talk) 01:01, 8 July 2015 (UTC)
That is what I mean with we are already addressing it. But that won't help us if we then can't show how things are getting better. --Lydia Pintscher (WMDE) (talk) 01:40, 8 July 2015 (UTC)
@Filceolaire: You want to attract wikipedians to contribute to wikidata but for that wikipedians have to be interested in Wikidata. But as they are looking for high quality data they are not interested in the current version of WD. So starting from that reasoning your proposition of a better tool for contributing from WP is not interesting. You have to provide first a small dataset of high quality in order to attract those people. Snipre (talk) 07:25, 8 July 2015 (UTC)
  • The most useful statistics I can think of right now are statistics for individual properties rather than the entire dataset. Other Wikimedia projects using Wikidata are going to be using specific properties, so it would help them determine whether they think the property they're interested in is reliable enough to use or not and whether it's getting better or worse. Some properties need references more than others too (e.g. identifiers for linking datasets versus statements about the item itself like date of birth), which is something which seems to be lost when looking at statistics for the entire dataset. Other statistics which might be nice to take into account are the number of unique sources per item and the number of references per statement. - Nikki (talk) 04:54, 8 July 2015 (UTC)
What we have to look for depends indeed on the type of property.
For constraint checks, it's probably worth to differentiate between the ones that identify errors and the ones that work more like property suggestions for an entire list of items.
If the objective is to reassure people, maybe the interface shouldn't be presenting the import log in the reference section. --- Jura 06:22, 8 July 2015 (UTC)
  • Percentage is not a good metric, number of statements is better. Because percentage metric is like a preposition for deleting particular claim instead of improving it. -- Vlsergey (talk) 07:13, 8 July 2015 (UTC)
Another metric shall be "quality of sources". For example GND is not a good source for dates (many mistakes), but very popular in Wikidata. Britannica would be better, but rarely used. Year of publication of source may be another metric as well. -- Vlsergey (talk) 07:15, 8 July 2015 (UTC)
  • Not really convinced:
  • Percentage of statements with a reference: it doesn't say too much about quality. You can use the first webpage you find to source your statement so using Google instead of Wikipedia, I don't think this brings more value.
  • A measure of the quality of the references used is better. But this is difficult because depends on the field and needs assessment by contributors which have some knowledge and expertise in that field.
  • I will propose a mix of the two propositions above: percentage of statements with reference but with a differentiation according to the type of source used. Books, scientific articles and databases should have a higher value than media or url. Then as most Books, scientific articles and databases will have their item on WD we can later attribute a note to these different sources on WD and weight the presence of reference with the quality of reference. Snipre (talk) 16:02, 8 July 2015 (UTC)
As I mentioned elsewhere, all instances of P373 (Commons Category) which are sourced are sourced to Wikimedia Commons. This is perfectly valid, and should not be counted as anything negative. There are also other similar properties, sunc as Main Category.--Ymblanter (talk) 16:32, 8 July 2015 (UTC)

property label/alias uniqueness

Hey folks :)

Right now the property parser function on Wikipedia and other clients only works with the label of the property. People are asking us to make it also work with aliases. In order to do that labels and aliases need to be unique across properties. There are unfortunately a number of properties where the label and aliases are not unique. Those need to be fixed before we can enable the feature. Bene has created a page that lists all properties that need to have their aliases changed: https://tools.wmflabs.org/bene/alias-uniqueness/ If you could have a look at this and fix some that'd be much appreciated. --Lydia Pintscher (WMDE) (talk) 23:26, 7 July 2015 (UTC)

Lydia, having aliases for properties that are not unique is not a bug. It's a feature. It means that if you enter a property name that is ambiguous then the UI will show you multiple properties, with descriptions to help you pick the most appropriate property. Can't the property chooser in Wikipedia work the same way? Filceolaire (talk) 00:47, 8 July 2015 (UTC)
There is no chooser in that sense. You type {{#property:capital}} or similar. --Lydia Pintscher (WMDE) (talk) 01:38, 8 July 2015 (UTC)
Tell the "People" it does not look like a good idea! -- Innocent bystander (talk) 07:01, 8 July 2015 (UTC)
+1 Snipre (talk) 15:42, 8 July 2015 (UTC)
Making parser functions works with the label of the property is bad idea from the beginning. First of all, WD access shall be from LUA or templates, not from articles. It shall never be "{{#property:capital}} or similar" (because of qualifiers, ranks, sources!). Second, using property label makes code vulnerable to label change. Third, it makes code/module/article/template much harder to translate to another language. -- Vlsergey (talk) 07:23, 8 July 2015 (UTC)
I agree with the translation problem. That said, the parser function could have a substitution semantics, with stuff like substituting the alias with the Pid (or Qid, a parser function for items could be great) and adding a comment in the wikitext like <! -- property alias/label -- > for wikitext reader. TomT0m (talk) 08:47, 8 July 2015 (UTC)

New Authority control validation tool

Hello everyone,

I am proud to announce a new cross-wiki authority control validation tool: https://tools.wmflabs.org/kasparbot/ac.php The website collects different types of problems which occurred as a result of my bot's activity of migrating authority control information from different Wikipedias to Wikidata. Feel free to solve the problems which are listed there and to discuss the tool and possible improvements.

Regards, --T.seppelt (talk) 12:45, 8 July 2015 (UTC)

Subdividing "is a list of"

I propose to subdivide is a list of (P360). into things like:

  • is a list of works by
  • is a list of objects owned by
  • is a list of places in the administrative territory

Any thoughts, before I make formal proposals? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 15:18, 7 July 2015 (UTC)

Is a list of is used extensively to register queries that are executed in Reasonator. There are many thousands of them. They are used on list and category items.
PS Reasonator already has ways to generate those lists and such lists should be available when the long awaited Wikidata Query makes its appearance. Thanks, GerardM (talk) 15:37, 7 July 2015 (UTC)
I'm sure Reasonator can be made to work from sub-properties of P360 (as say, instances of an item such as "Wikidata lists type", as well as P360 itself. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 21:50, 7 July 2015 (UTC)
@Pigsonthewing: I think it's a terrible idea. We already have classification properties, and creating a class Work by X can already be create, with
< Work by X > subclass of (P279) View with SQID < Work >
, this class can itself be classified per Help:Classification as
< Work by X > instance of (P31) View with SQID < class of all the work by someone >
. Then you can use the item in the
subject > is a list of (P360) View with SQID < Work by X >
. TomT0m (talk) 18:15, 7 July 2015 (UTC)
I'm unable to fathom what this has to do with the use of P360. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 19:50, 7 July 2015 (UTC)
Any motivation for this change Mr. Mabbett? --Succu (talk) 21:28, 7 July 2015 (UTC)
We have no satisfactory way, at present, to show that, for example, items like list of public art in Birmingham (Q16249030) relates to Birmingham (Q2256); and (to a lesser extant) Pink Floyd discography (Q756304) would be better as a "list of works by Pink Floyd" than "performed by Pink Floyd", since Pink Floyd perform the works, not the discography. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 21:50, 7 July 2015 (UTC)
Andy, I still think most 'list of' WP articles would be better treated as a class so a Pink Floyd discography would be <subclass of:recorded works><performed by:Pink Floyd> and list of public art in Birmingham (Q16249030) is <subclass of:public art><located in:Birmingham>. Filceolaire (talk) 01:12, 8 July 2015 (UTC)
Isn't this just what you need? Edoderoo (talk) 20:53, 8 July 2015 (UTC)

Property P1351

I am trying to use the property number of points/goals/set scored (P1351) and I can´t put the number "0", automaticaly puts "0±1", how can I solve it? --Ezarate (talk) 18:35, 7 July 2015 (UTC)

Type "0+-0". Popcorndude (talk) 18:38, 7 July 2015 (UTC)
We need to get rid of that default +-1 really. --AmaryllisGardener talk 19:26, 7 July 2015 (UTC)
Agreed, I've never understood why it's there. I've never had a need for it. --Stryn (talk) 19:29, 7 July 2015 (UTC)
+1 -- Innocent bystander (talk) 19:35, 7 July 2015 (UTC)
See the related bug--GZWDer (talk) 18:26, 8 July 2015 (UTC)
This is indeed a very annoying bug! 10 stroopwafels for him/her who will solve this. Edoderoo (talk) 20:55, 8 July 2015 (UTC)

Property like "referring to"

I have seen it somewhere though cannot remember were, for article Ralph of Coggeshall (DNB00) (Q19091989), it is not actually an article, it is instead a published pointer to find the article elsewhere in the volumes. Could someone please identify that property, and I will document it for the DNB project. Thanks  — billinghurst sDrewth 00:49, 9 July 2015 (UTC)

Or we can just look to remove something like Q19091989 and call it non-notable. I don't think that removal would be an issue for the enWS community.  — billinghurst sDrewth 00:50, 9 July 2015 (UTC)

Possibility to pull in the entire iw link list of "Article A" for "Article B"

Hi, all. I am doing some work these days on the Ladino (Judeo-Spanish) Wikipedia. Like some other languages (Serbo-Croatian comes to mind, but there are others), Ladino can be written in two different alphabets—in this case, Latin and Hebrew. There are articles in this wiki in both scripts, though Latin alphabet predominates. To give an example, there is an article on astronomy (Q333), lad:Astronomiya. There is also an article lad:אסטרונומייה, which is more or less the same article in Hebrew script. Understandably, the WD item points to the Latin-script version, and as far as it goes the Hebrew script article title is the Hebrew language label for the WD item (with a spelling variant).

Ideally, it would be worthwhile for wikis that legitimately have parallel articles in different scripts to be able to map them both to the wikidata item. How easy or difficult that would be I don't know; how much that might open a can of worms in other ways, I also don't know. However, I'm wondering: is there any way that I can import the entire iw list of articles so that on the Hebrew-script page one would see them? In other words,

  • In all other languages, clicking the "Ladino" link would take you only to the Latin-alphabet version, but
  • You would be able to see all of the iw links of Q333 not only in lad:Astronomiya but also in lad:אסטרונומייה.

Thanks. StevenJ81 (talk) 18:21, 8 July 2015 (UTC)

Until we have a fully working solution to cases like this, see how said to be the same as (P460) is used in no label (Q12715471). In Nynorsk Wikipedia, some articles are written in Högnorsk, why they have two articles in some subjects. nnwiki is not alone in this manner. -- Innocent bystander (talk) 18:56, 8 July 2015 (UTC)
@StevenJ81: please add your usecase on WD:XLINK. It's the project where we figure out how to sort out such interwikis problems. TomT0m (talk) 20:16, 8 July 2015 (UTC)
@StevenJ81: Do not use said to be the same as (P460), do not join WD:XLINK, but look at Special:Search/Bonnie Clyde problem [3]. Visite fortuitement prolongée (talk) 21:10, 8 July 2015 (UTC)
@Visite fortuitement prolongée and @TomT0m: Chers amis:
  1. This really isn't a Bonnie-and-Clyde problem. I have often used the Bonnie-and-Clyde shortcut approach in other situations. It doesn't really work here. Both articles exist, both articles essentially have the same name, and both articles cover substantially the same ground. It's just that they're written in two different alphabets. But they're really duplicates, albeit legitimate ones.
  2. I added the use case at WD:XLINK.
  3. @Innocent bystander: Thank you. I'm not doing that yet. I have not put the Hebrew-alphabet version in Wikidata at all, yet, and for some pretty good reasons:
    • Its title is a (very slightly) variant spelling of the word that is already the Hebrew label for Q333, and rightly so. I'm not even sure I'd know what to call the second WD entry, unless something contrived, like the Hebrew-alphabet version of lad-he-Astronomy.
    • By all rights it would be merged if it could be. But people will try, I guarantee it. I just don't want to fuss with that problem for now.
    • P460 carries the legend of "said to be the same ... but disputed." No one disputes this. They really are the same, or aspire to be. They're just. different. alphabets.
Frankly, from the perspective of someone on Wikidata (or one of the Wikipedias), both articles do not need to appear as links. If one clicks on the Ladino article in Q333, which is in the Latin alphabet, the first thing you'll see at the top of the landing page is a pair of tabs that will let you toggle between the Latin- and Hebrew-alphabet versions. (I'm in the process of making sure these exist on all paired pages at ladwiki.) What I'm more interested in doing is to be able to place a template, or a module call, or something, on the Hebrew-alphabet page that would essentially put all of Q333's iw links on the Hebrew-alphabet page, almost as if that list of iw's were itself a property call. StevenJ81 (talk) 21:37, 8 July 2015 (UTC)
That is fully possible to do. And that is a good reason to use "said to be the same". In this case the level of "dispute" between lad:אסטרונומייה and lad:Astronomiya is ~0%, and that is stated as P31:Wikimedia duplicated page (Q17362920) in the nnwiki-item I linked above. When you finally have Arbitrary Access on ladwiki, you can with a module import the sitelinks of the item linked by P460 to use them as interwiki on the hebrew page. -- Innocent bystander (talk) 06:02, 9 July 2015 (UTC)
I may ask for help on that. I think I get arbitrary access on July 29. One question: does the Heb-script version have to be in Wikidata for this to work? Thank you. StevenJ81 (talk) 12:48, 9 July 2015 (UTC)
If you add the Qid of the article of "lad:Astronomiya" into the template of the hebrew page, you do not need that. But if you uses P460 in the item for the Hebrew page, the template can find the correct item by itself. -- Innocent bystander (talk) 12:54, 9 July 2015 (UTC)
OK, @Innocent bystander. I'll ping you on or after July 29 to get more help with that. Thanks. StevenJ81 (talk) 18:49, 9 July 2015 (UTC)

Error with extraterrestrial coordinate locations (P625)

Hello,

I'm looking for a way to display Wikidata data in a future infobox on the French language Wikipedia about geograpical features on extraterrestrial bodies, such as quadrangles or craters. Unfortunatly, geohack generated URL has two annoying bugs preventing me to using the coordinate location (P625) property efficiently : "globe" is always set to "earth" and "language" is always set to "en" (examples: Diacria quadrangle (Q3055620) and Endurance (Q1341055)). A proposal in 2013 already talked about the first issue (and the fix has been said to be "easy" to do), but nothing as been done about it for years. Could somebody help me to fix those errors/bugs, please? I just know how to do simple edits here and I'm totally at loss when I have to deal with such complexity. :/ Feldo (talk) 11:28, 9 July 2015 (UTC)

(edit: crossposted in French to improve to possibilities to fix this quickly Feldo (talk) 11:37, 9 July 2015 (UTC) )

I've created phab:T105321 for the globe always being Earth and phab:T105323 for the language always being English. - Nikki (talk) 12:34, 9 July 2015 (UTC)
@Feldo: What do you mean by "the language always being English". -- Innocent bystander (talk) 12:35, 9 July 2015 (UTC)
@Innocent bystander: The geohack link always redirect to the English language version of geohack pages (like here) instead to act accordingly with the interface language set by the wiki/registered user. Feldo (talk) 12:41, 9 July 2015 (UTC)

Announcing WikibaseQuality Constraint Reports

  • Already posted above #Wikibase Quality Extensions have launched. Artem Korzhimanov (talk) 13:34, 7 July 2015 (UTC)
  • Congrats to the team. Good work! --- Jura 13:38, 7 July 2015 (UTC)
  • Thank you very much for your hard work! Tpt (talk) 16:37, 7 July 2015 (UTC)
  • Great! What about setting/gadget to show link straight in item UI? --Jklamo (talk) 21:21, 7 July 2015 (UTC)
    • It'll be even better ;-) We'll show the issues right next to the statements. That'll come in the next version. --Lydia Pintscher (WMDE) (talk) 22:31, 7 July 2015 (UTC)
    • I have written some lines of JS that probably do what you described. https://www.wikidata.org/wiki/User:Jonas.keutel/ConstraintReport.js. You can use it in your common.js. This script displays a link to the constraint report special page on wikidata.org and also a link to a page that performs cross-checks agains GND. This will soon be live here, but currently, we only run it on our labs instance. Jonas.keutel (talk) 12:54, 9 July 2015 (UTC)
      • Thanks! --Jklamo (talk) 14:31, 10 July 2015 (UTC)
  • Very nice and useful. Good work folks :) Jean-Fred (talk) 08:58, 8 July 2015 (UTC)

Importation of Freebase content : what should we do with identifiers?

As you may already know, I am working on feeding the Primary Sources tool with data from Freebase. But before moving forward further I would like to have the community input on what should be done with Freebase identifiers and IDs of external databases that are stored in Freebase.

Freebase identifier

Current state of the mapping between Wikidata and Freebase (numbers are approximations):

  • Freebase contains 48M topics and Wikidata 14M of items.
  • Wikidata contains 1.1M of Freebase ids. These ids are mostly based on a mapping published by Google.
  • An other mapping have been published by Samsung with 4.4M of links. It contains 3.2M extra links compared to Wikidata and 6,000 conflicts (the mappings gives two different Wikidata item ids for the same Freebase topic id). The Samsung dump mostly includes the current Wikidata mapping (only 9,000 extra links in Wikidata).
  • Using both mapping we map 9% of all Freebase topics and 32% of Wikidata.
  • It is currently planned to feed the Primary Sources tool with a merge of the two existing mappings, the Wikidata one being preferred in conflicting cases.

Advantages to have more Freebase ids in Wikidata, e.g. by importing the Samsung dump:

  • Allows Freebase API users to migrate to Wikidata more easily
  • Have a nice place to proofread/improve the mapping used in order to feed Primary Sources

Disadvantages:

  • Maintenance cost

Possible ways to import the Samsung dump:

Bot import
Advantages: quick and easy
Disadvantages: difficult to track errors except having a wikipage listing conflicts
Mix'n'match
Advantages: easy to use interfaces to proofread/enhance matching
Disadvantages: Not directly into Wikidata, dataset will maybe be too big for the software.
Primary Sources
Advantages: safe, everything should be validated
Disadvantages: boring and slow work, we should maybe use contributor effort for something more useful

What do you think about it? Should we import the Samsung dump into Wikidata? And, if yes, how should it be done? Tpt (talk) 21:20, 7 July 2015 (UTC)

Google's initial backward mapping seems very restrictive (requiring links to several WP) and I don't quite understand why that is so (a lot of content isn't overlapping). Many (English?) Wikipedia entries seem to have been used to start Freebase entries (even include part of the WP description), but somehow these are not all included. We could attempt to re-do a mapping or rely on Samsung's.
For Samsung's maybe some sampling should be done and, if this is convincing, upload the identifiers to Wikidata. If sufficient sampling has already been done through "Primary Sources", we could move ahead. --- Jura 05:41, 8 July 2015 (UTC)
I have just uploaded to the Primary Sources tool 8M of new statements from Freebase (in the dataset "freebase-testing"). These statements have been created using the Samsung mapping. I believe that the number of wrong statements because of mapping issues will give a good idea of the mapping quality. Side remark: the Primary Sources backend is currently experiencing some performance issues. Tpt (talk) 23:44, 8 July 2015 (UTC)
It seems there are many more suggestions -- or maybe my selection of items to check is better ;). Did you add more property mappings as well?
Obviously a large part is things we haven't imported yet from WP. Still, it's easier to do it with the tool.
BTW, the reject button doesn't seem to work (the suggestion re-appears). Also, there are statements with incorrect date qualifiers. Currently one can just accept or reject the entire statement. --- Jura 12:04, 10 July 2015 (UTC)
I have added 8M of new statements last Tuesday using the current state of the mapping. There are some problems with the reject button caused by performance problems of the backend. It is also why sometimes the random button or the statement load fails. Yes, I should maybe implement an edit button that would allow to add/remove qualifiers before saving the statement. But it'll require a not negligible amount of development work. Tpt (talk) 16:47, 10 July 2015 (UTC)

Other IDs

Freebase contains quite a huge number of external identifiers. For example, there are 17M links to MusicBrainz, 5.3M ISBNs, 0.8M links to IMDB. With the existing mappings we could import into Wikidata 138,000 new Discogs artist ID (P1953), 91,000 new IMDb ID (P345), 85,000 new Open Library ID (P648)...

There are some possible ways to do the importation:

Bot import
Advantages: Requires less contributor time
Disadvantages: Introduces data into Wikidata that have not been locally curated
Primary Sources tool
Advantages: Simple to setup + human validation
Disadvantages: Takes a lot of effort
Hybrid approach, upload to Primary Sources in order to check the quality of the data, then use a tool that would do an automated “approve claim” for all Primary Sources statements matching a given query
Advantages: allows to easily see if the data are good + we do not consume too much contributor effort
Disadvantages:?

My personal preference is for the hybrid approach with the creation of a new Feature for Primary Sources, maybe relying quick statements, that would allow authorized users to do massive imports. If there is no oppose to it, I would proceed with the hybrid approach. Tpt (talk) 21:20, 7 July 2015 (UTC)

Sure. It's just that in its current mode, the Primary Source Tool doesn't allow to check in a targeted way.
As for specific properties some contributors have spent a lot of time cleaning them up, we need to make sure we don't re-upload what has been removed. --- Jura 06:48, 8 July 2015 (UTC)
I agree with you, mass import using already well curated properties like VIAF is something we should avoid. Yes, a feature like "random item with this property" would be very useful for this use case. I'll consider to work on it if there is a consensus about the use of Primary Sources. Tpt (talk) 23:46, 8 July 2015 (UTC)

Citation needed

Have we hade any thoughts about marking statements in need of sources?

I today changed a statement in the item for Kraków. A sister city (P190) normally is between two political entities and Gothenburg (Q25287) is not a political entity. Thereafter I went to the homepage of Göteborg City, to get a source to this statement. But I cannot find any source for any sister city (P190) other than Turku, Finland; Århus, Denmark and Bergen, Norway. -- Innocent bystander (talk) 15:22, 9 July 2015 (UTC)

Is the mark in the statement defined as "0 references" not enough as mark ? If we can avoid to transform our items into Christmas trees I think we will be more happy. Snipre (talk) 15:50, 9 July 2015 (UTC)
@Snipre: It said 1 reference: "imported from plwiki" before I changed it. It did not help to have a source here, it's still wrong. -- Innocent bystander (talk) 16:39, 9 July 2015 (UTC)
  • We need some way to mark statements as "citation needed", "disputed", "non-reliable source", "improve reference", etc. It shall be something like qualifiers, but not qualifiers. From technical point of view it may still be qualifiers, but with different UI representation from viewer / editor point of view. -- Vlsergey (talk) 21:59, 9 July 2015 (UTC)
    • Agreed. Qualifiers are supposed to help, amongst other things, indicate the quality of the data.--Jasper Deng (talk) 03:24, 10 July 2015 (UTC)
  • How about properties with type of date like "Reliable source required since" or "Claim disputed since"? By default it should be "today", i.e. date of adding the qualifier. Other option is to create two qualifiers like "claim issues" and "claim has issues since", first with entity like "citation needed", "disputed", "non-reliable source", "improve reference", etc., second -- with date. Second can be added by bot automatically. -- Vlsergey (talk) 05:36, 10 July 2015 (UTC)

Source links formatting?

Is it possible, or will it be possible to format source links? They are pretty ugly as a naked link in Wikipedia articles. Check out hu:Aaron Yan, the second ref. All the sources are neatly formatted and then there is the source link from wikidata, sitting there naked. It would be nice if we could use cite templates or a similar structure would be built in for the "source link" property. Teemeah (talk) 11:59, 10 July 2015 (UTC)

You need to add a title to the reference in wikidata and then pull that title into your template. Then you can use standard [https://www.example.com standard linking] where the URL is the URL and the title is the second part. If on the other hand this is being sucked into a template like the citation templates on en.WP, then you need to set the title parameter (or equivalent) in the template after adding the data in Wikidata, --Izno (talk) 13:30, 10 July 2015 (UTC)
@Teemeah: See help:sources. Snipre (talk) 16:45, 10 July 2015 (UTC)

Coordinates precision

@Multichill: Latitude and longitude values must be integer multiples of the precision, at least theoretically. Precisions like 0.39915187561824 have no sense. The purpose of the precision is the possibility to tell that 2000 is precise to the tens, or a coordinate is precise to the arcseconds if it's a round value to the arcminutes. Please stop flooding Wikidata with this kind of precisions. --JulesWinnfield-hu (talk) 19:50, 6 July 2015 (UTC)

There was some discussion about it in Property talk:P625#Special precision topic. Maybe coordinates data type need more discussion as it is already in progress with date/time data type. Paweł Ziemian (talk) 20:17, 6 July 2015 (UTC)
Wikidata has a duty to data users to make it clear what any non-trivial value means. Since latitude and longitude are widely reported in several formats, it is non-trivial to figure out what the precision value means. Unless someone can show me where the meaning of the precision of a latitude or longitude is clearly presented to users, I think all such precision values should be deleted. Once the community agrees on what they mean and present the meaning to users, the values could be recomputed and added back. I was able to find in mw:Wikibase/DataModel that precision is represented in degrees, and also that when one edits a location in the user interface, it is shown that the precision is in degrees. The precision value is a fixed point decimal, with 3 digits before the decimal point and 9 digits after. Jc3s5h (talk) 21:08, 6 July 2015 (UTC) modified 22:50 UT.
@JulesWinnfield-hu: either you put a message on my talk page directed at me or you post a general note here. I consider it quite rude to post your message like this here in this location. Go read the old topic and phab:T89218. Multichill (talk) 21:53, 6 July 2015 (UTC)
There is no such thing as special precision. Could be almost any value, but the coordinate values must be integer multiples of the precision. Otherwise is nonsense. It has nothing to do with meters. It's the precision of a numerical value. Please don't add wrong precisions. --JulesWinnfield-hu (talk) 22:39, 6 July 2015 (UTC)
I see no requirement that coordinate values must be integer multiples of the precision. Please show where this requirement is stated. Jc3s5h (talk) 22:53, 6 July 2015 (UTC)
The precision is an attribute of the stored value, not an independent property. If the value is not an integer multiple of the precision, it makes no sense. Read my explanation in the first comment. --JulesWinnfield-hu (talk) 23:12, 6 July 2015 (UTC)
Strictly speaking, the stored value is in an internal format that is not accessible to data users or Wikibase editors. The nearest we can get to it is the JSON representation, which can be exported for any item. This is described at mw:Wikibase/DataModel/JSON. The latitude, longitude, and precision are given in degrees, including a decimal fraction after the decimal point. There is no requirement in the data model that the value be an integer multiple of the precision. Considering that the latitude and longitude may be found by many different methods, and the precision may be found by many possibly different methods, there is no reason why anyone would want to create such a restriction. If you are saying the only reason why the value must be an integer multiple of the precision is because, in your opinion, it is the only approach that makes sense, I disagree. If you found this requirement somewhere in Wikidata or Mediawiki, please indicate where. Jc3s5h (talk) 23:42, 6 July 2015 (UTC)
You are in error. The precision is the precision of the numeric value. The only reason it exists is to be able indicate higher precision (lower precision value) than the implicit precision. To be able to give 10, 1 or 0.001 if the value is 100 for example. It's not the precision of something else. If the value is 100, 22.005 the implicit precision is 0.001, but if it is 100.000, 22.000 then 100, 22 is stored with the precision 0.001 which has to be set manually. The implicit precision for 10°7' is 1/60 but you can set 1/3600 which means 10°7'0". That's all the precision exists for. You cannot enter a random value. This is not about where it is stated or what the community thinks about, this is just what the precision means. --JulesWinnfield-hu (talk) 00:51, 7 July 2015 (UTC)
JulesWinnfield-hu I am an engineer so I deal with precision. It is perfectly possible to have a measurement which is 10.37mm +-0.1mm. This means that the average of the measurements taken is 10.37mm but the various measurements varied between 10.27mm and 10.47mm (I will omit the whole discussion of what type of average and how to use standard deviations to define uncertainty intervals). I agree with you that a reading of 10.36987654mm +-0.1mm would mean that most of the digits in the first number are not significant or useful but that doesn't mean they are wrong. Sometimes you get this type of result when you convert units. Best to keep the extra digits in case you have to convert back some day. Multichill is not wrong here. You are. Filceolaire (talk) 01:32, 8 July 2015 (UTC)
For a practical example of expressing uncertainty of a horizontal position (although not the kind of position likely to be added to WikiData) see the US National Geodetic Survey datasheet at http://www.ngs.noaa.gov/cgi-bin/ds_mark.prl?PidBox=AE8289. The position is given in degrees, minutes, and seconds [46°46'29.10992"(N) 092°05"37.38770"(W)] but the uncertainty is given as 0.33 cm. It would just be silly to expect that after converting the uncertainty in cm into degree and the degrees-minutes-seconds into degrees that the latitude and longitude would be exactly divisible by the uncertainty. Jc3s5h (talk) 02:05, 8 July 2015 (UTC)
@Filceolaire, Jc3s5h: Please read my previous comment. It has nothing to do with meters, it has nothing to do with uncertainty, and random values shouldn't be added any more. There is no conversion, there is no measurement. The precision is necessary only because we store the value numerically and not as a string literal. 22 and 22.000 are stored in the same way. 10°7' and 10°7'0" are stored with same 10.116666666666667 and have to be distinguished with the precision. This is the only reason of the precision. What you talk about is completely different. --JulesWinnfield-hu (talk) 11:14, 8 July 2015 (UTC)

────────────────────────────────────────────────────────────────────────────────────────────────────We store two angles and the uncertainty of those angles. I'm an engineer, and I know perfectly well what the string 10°7'0" means, if written by someone with an engineering or scientific background. It means the effort of writing a full and proper description (something like 10°7'0" with a normally distributed error, +/- 2", with 95% confidence) wasn't justified. It means that there is some uncertainty in the seconds digit. When this is converted to decimal degrees, there is some uncertainty in ten-thousanths digit, so the uncertainty is at least +/- 0.00028°. All this effort to try to represent numbers so the original string could be reconstructed is completely non-standard and unconventional; I refuse to accept the concept. Jc3s5h (talk) 11:33, 8 July 2015 (UTC)

I'll pile on as another engineer and say that Jules, you are wrong. --Izno (talk) 18:03, 9 July 2015 (UTC)
I kindly ask all engineers from Wikidata to ask Multichill and others not to add wrong precisions to Wikidata like here Property talk:P625#Mismatched display of location data. For decimal values add the precision corresponding to the number of decimals, for dms values add 1/60, 1/3600___ accordingly, without rounding the latitude, longitude and precision values, just like the UI does. --JulesWinnfield-hu (talk) 00:19, 10 July 2015 (UTC)
I agree with some of what Mulitchill is doing, and some of what JulesWinnfield-hu suggests. I think the precision should reflect the actual precision of the location, to the extent that is known. It is not simply the fraction corresponding to adding one to the last digit shown in the source (unless the number of digits printed in the source is the only information about the uncertainty). I disagree with the approach of the user interface in creating a multi-digit precision such as 0.00027777778 for 1/3600; the precision shouldn't need more than two significant figures. The hope of easily reconstructing the numbers from the source is an illusion. The numbers from the source could be a different datum (like NAD 83), a Cartesian grid (like UTM), etc., so it can take some work to verify the claim. But the claim, including the precision, should be verifiable from the source. We shouldn't have edits that provide a precision that have no apparent relationship to any numeral in the source.
Unfortunately discussion of this on Wikidata, Phabricator, and the mw:Wikibase/DataModel has been so sporadic that no quorum has formed to decide if the precision should be the the uncertainty in the measurement process, or an area that roughly corresponds to the size of the thing being located, or perhaps the greater of these two. Jc3s5h (talk) 01:45, 10 July 2015 (UTC)
You divert the discussion. Please don't bring in uncertainty, measurement, area and such things. This is pure theory, mathematics and computer science. We have a data 9°46'40" without any of aforementioned attributes and we have to store it somehow. This can be done by storing 9.7777777777778 with the precision 0.00027777777777778 and the value will be integer multiple of the precision. If you want to display the dms value you will get exact 9°46'40" just like in the source as you can see here Special:Diff/228264478. If you don't do like this you won't get back the value you want to store. This is not complicated, this how the UI works (couldn't other way) and this is what the precision is for. Engineers please don't bring in measurement theory, I know those terms, but this is not about that, not the purpose of this attribute, it's just a tool to be able store the value. Also to differentiate 10°7' from 10°7'0". --JulesWinnfield-hu (talk) 10:22, 10 July 2015 (UTC)
JulesWinfield-hu, you want pure math, fine. In mathematics, a single counterexample disproves a hypothesis. The example http://www.ngs.noaa.gov/cgi-bin/ds_mark.prl?PidBox=AE8289 because it is impossible to enter that point into Wikidata in a way that allows one to reverse the entry process in the way you describe. This counterexample disproves your hypothesis. Jc3s5h (talk) 11:10, 10 July 2015 (UTC)
It seems like you didn't catch my message. Why don't you (plural) see the evident? --JulesWinnfield-hu (talk) 11:20, 10 July 2015 (UTC)
Why don't you take us step by step how you would enter my example into WikiData? Jc3s5h (talk) 11:36, 10 July 2015 (UTC)
When the system was designed, the maximum precision was arcseconds with 3 decimals. So if you round the input to that precision, store it as I say, you will get the exact display of the value you want to store. Now the constraints were lifted but even like this you will get something similar. See Special:Diff/228250501/228302562. --JulesWinnfield-hu (talk) 13:06, 10 July 2015 (UTC)

────────────────────────────────────────────────────────────────────────────────────────────────────The sample data is in a different datum, and must be transformed to WGS84 before it can be added to WikiData. If you decide to use the user interface, the data would have to be entered as in this diff but a user skilled with the API could enter it in decimal degrees, and specify the precision as 0.00000001°, which would be more conventional and closer to the uncertainty from the source converted to an angle, 1.1 arcsecond. So having many digits in the precision does not help the reader understand the uncertainty as stated by the source. It only tells us what uncertainty the editor decided to write on his scratch paper before entering it, possibly while having his arm twisted by the limitations of the user interface. Jc3s5h (talk) 16:58, 10 July 2015 (UTC)

This is another off topic that wasn't my intention to cover. My call is not to flood Wikidata with wrong precisions with a faulty Python script. If we have a WGS84 data, considering storage limitations, the only meaningful precisions are 10n, 1/60 and 1/3600*10n consistent with the stored value. Again, no uncertainty. Just a number and the task to store it. --JulesWinnfield-hu (talk) 20:15, 10 July 2015 (UTC)
How does it hurt you, when the precision is out of your desired set of values? Where and what are the unwanted consequences? Paweł Ziemian (talk) 21:05, 10 July 2015 (UTC)
You can't say if it is 10°7' or it is 10°7'0". All data users which use precision to display the correct value. The UI itself, see Property talk:P625#Mismatched display of location data. Why would we want to store wrong, meaningless precisions? --JulesWinnfield-hu (talk) 21:18, 10 July 2015 (UTC)
Yes, I can. I just choose the best desired precision myself, which is the largest one and less or equal of the given invalid precision. Paweł Ziemian (talk) 21:29, 10 July 2015 (UTC)
It is not correct because you would choose 10°7', but if we store arcseconds precision then the correct display is 10°7'0". --JulesWinnfield-hu (talk) 21:37, 10 July 2015 (UTC)
But Wikidata does not tell how to display value, rather how it was measured. In this example the precision is about 24', so it is naturally for me that the location is given as 10°7'. Do you want to say how far is a train from station A in millimeters? Paweł Ziemian (talk) 21:49, 10 July 2015 (UTC)
If you are able to store 10°7'40" you have to be able to store 10°7'0" too. And if we store 10°7'40" we will display 10°7'40" and not 10°7'. --JulesWinnfield-hu (talk) 22:09, 10 July 2015 (UTC)
The user interface is only one option for entering data. The user interface should be regarded as an unusual way of displaying data. The main way of displaying data should be in software independent of WikiData, chosen by the reader, or in Wikipedia infoboxes (if the data ever gets reliable enough for that). So we should be concentrating on putting data in that will garner the respect of the skilled data consumers who will decide if WikiData is a success. Weird departures from the normal practice of technical professionals isn't likely to garner that respect. I don't give a rat's ass about how the data looks in the user interface. Jc3s5h (talk) 00:08, 11 July 2015 (UTC)
Why don't you take us step by step how you would store and afterwards display 10°7'0"? The UI does it in a theoretically correct way. I don't understand your objections. And this is not the main problem. Please appeal against the faulty Python script and the totally wrong precisions. --JulesWinnfield-hu (talk) 09:57, 11 July 2015 (UTC)

────────────────────────────────────────────────────────────────────────────────────────────────────I personally have not had a need to extract large amounts of position data from WikiData. If I did, I would probably write a program that does an automated web query using the URL https://www.wikidata.org/w/api.php?action=wbgetclaims&entity=Q178114&property=P625&format=jsonfm to extract the location (P625) data for Washington Monument (Q178114). I would parse the response, part of which looks like this:

"value": {
                            "latitude": 38.889464,
                            "longitude": -77.035228,
                            "altitude": null,
                            "precision": 1.0e-5,
                            "globe": "http://www.wikidata.org/entity/Q2"
                        },

Since I would be handling the data with a program, not looking at each entry manually and trying to puzzle out whether the creating editor was working in decimal degrees, degrees and decimal minutes, degrees-minutes-seconds, grad, radians, time equivalent of longitude, feet, meters, etc, I would just use the precision to figure out how many digits to keep in the latitude and longitude, and round to that number of digits. If I didn't have some special requirement of how to display the data, I would probably just write it as comma separated values (CSV) and look at it with Excel. In this case, I would round to the 5th digit to the right of the decimal, and save the latitude and longitude as 38.88946 and -77.03523. Jc3s5h (talk) 13:45, 11 July 2015 (UTC)

Pair of values latitude, longitude give us point (Q44946) on sphere (Q12507). Since point has no size, but any geographical object is not a point, but has some size, we need something that show us how big the object is. That is the main purpose of the non-standard precision field. Roughly speaking it gives latitude arc, whose arc (Q147572) sets diameter (Q37221) of disk (Q238231) that more or less cover the pointed object. This is how I understand it. Paweł Ziemian (talk) 14:45, 11 July 2015 (UTC)
In these edits to the data model a user named Denny, who used to work for Wikimedia Foundation, added precision and dimension. Dimension is exactly what we want to represent the size of an object. I don't know if it was ever implemented in the database.
In these two edits User:Adue removed the dimension without any explanation. In the JSON output I placed above, you can see that no dimension value is reported out of the databse. So my interpretation is that the developers, without consulting the community, decided that the dimension of an object is not a suitable quantity to represent with a globecoordinate. If you want to represent the dimension, open a feature request in Phabricator to restore dimension to the data model and implement it in the databaase. Jc3s5h (talk) 17:08, 11 July 2015 (UTC)
Maybe they want to kill two birds with one stone. There is high correlation between dimension and precision values. Meantime I don't cry for lack of one of them. However someone could expand the data model with a few examples about coordinates, including non-trivial precision values. Paweł Ziemian (talk) 18:18, 11 July 2015 (UTC)

French communes with wrong department

There are 50 French communes with conflicting "in administrative unit" claims in [4] - they are in one region tree and also in the tree of a department outside that region. The ones I've checked were just assigned to a wrong department (Hautes Pyrénées) but assigned to the right canton (in Landes department). I'm not sure about which is the most efficient way to fix it.--Pere prlpz (talk) 08:39, 10 July 2015 (UTC)

@Pere prlpz: Can you please, specify a few examples of communes that are wrong? -- Innocent bystander (talk) 10:16, 10 July 2015 (UTC)
Hoi, the best way of doing is this by only including the lowest level administrative unit. From this higher units may be inferred. Reasonator does this already for a very long time. GerardM (talk) 17:42, 10 July 2015 (UTC)
But what should we do when there is no simple hierarchy? I have struggled with that question since I came here, but still do not know any simple answer. -- Innocent bystander (talk) 18:58, 10 July 2015 (UTC)
@Innocent bystander: Examples: Q1103103, Q1112272 and all the comunnes find here by Autolist.
Please notice that administrative divisions hierarchy in France is quite simple - except for the canton-commune levels. Therefore, no commune should belong to two different departments or regions.
And I agree about leaving just the lower level if redundant. Now, the only doubt with those 48 is communes is whether to just remove department or to check first in Wikipedia if the error is the canton or the department. In the handful of instances I checked, the department (Hautes Pyrénées) was always wrong.
A best approach could be checking in Wikipedias using a bot or an automatic tool (Autolist again?).--Pere prlpz (talk) 08:57, 11 July 2015 (UTC)
I think I got it. I used [5] to check against fr:Category:Commune des Landes and double check the number or articles against its Catalan equivalent and now I'm removing department claim with -P131:Q12700.--Pere prlpz (talk) 09:14, 11 July 2015 (UTC)
Nice that you found the wrong statements. In Sweden (Q34), the hierarchy is not simple at all. The Municipality -> County - hierarchy is simple. But the parishes have cases when parts of a parish can be found in several municipalities or even counties. Civil parishes are even worse. I do not know how it will be with the new districts, but it's said that the borders of the new districts are fixed, they will not change in the future. A change in borders between two municipalities (something that happens from time to time) will never change the borders of a district. It will therefor look very complicated in the future. But the the most terrifying issue is the provinces. Is a county P131 a province or is a province P131 a county? The answer to that is the same as the answer to if an electron is a particle or a wave. Some has suggested that we should create items like "the part of parish X that is located in municipality Y but not in province Z". Technically that would maybe work, but it looks awful both to create such items and to use them. My place of birth would then be "the area located in parish X that was transferred from Municipality Y to City Z 1940". -- Innocent bystander (talk) 12:36, 11 July 2015 (UTC)
For administrative divisions with parts in two different upper level administrative divisions, the most important thing is not to claim both upper level divisions as P131, because it would produce wrong results in tree queries. Anything in the divided administrative division would need to carry also the upper level adminsitrative division.--Pere prlpz (talk) 17:52, 11 July 2015 (UTC)
Läppe (Q10572342) is a splendid example. Until 1994-12-31, the village was split between Örebro and Vingåker Municipality and therefor also two counties. It was also split between Södermanland and Närke province. 1995 the border between the municipalities changed, and the village is now in only Vingåker and therefor also only one county. But it is still in both of Närke and Södermanland province. Vingåker Municipality is today in at least two provinces. But parts of Läppe and some of the surrounding area is probably the only part of Vingåker that is located in Närke. -- Innocent bystander (talk) 18:53, 11 July 2015 (UTC)

Is Claims Id case sensittive

Hello, Is Claims Id case sensitive, some ids are upper case other are downcase should I uppercase them or leave them like I find them in dump files regards,

Leave them as you find them! :) ·addshore· talk to me! 16:16, 12 July 2015 (UTC)

(medieval) manuscripts, shelfmark/call number, etc.

Hello, I've been editing Q2258615 to try to get a good example of a medieval manuscript in Wikidata, but can't seem to find any properties for the shelfmark (generic term) or call number (more specific term) that describes how to identify an item in a library. Have I missed this or should I add a new property? Thanks! Mscuthbert (talk) 10:27, 12 July 2015 (UTC)

Glancing through Wikidata:WikiProject Books, I suspect inventory number (P217) is what you're looking for. Gabbe (talk) 11:01, 12 July 2015 (UTC)
Thank you! I added aliases for shelfmark and call number to that. Mscuthbert (talk) 13:39, 12 July 2015 (UTC)

Wikidata weekly summary #166

Meetup at Wikimania

Hey :) Wikimania is approaching soon. Does anyone want to organize a Wikidata meetup? It was very nice to have a chance to chat with fellow editors last year. --Lydia Pintscher (WMDE) (talk) 22:59, 7 July 2015 (UTC)

If someone organizes a meet up I'll probably be there! Multichill (talk) 07:01, 9 July 2015 (UTC)
I've now set one up: https://wikimania2015.wikimedia.org/wiki/Wikidata_Meetup Hope to see many of you there :) --Lydia Pintscher (WMDE) (talk) 00:57, 13 July 2015 (UTC)

Wikidata for links and red links in Wikipedia

I realised that Wikidata is ready to provide link functionality in Wikipedia. I blogged about it. The point is very much that through templates it already provides a better functionality than red links. The new idea is that it can also provide a better functionality than Wikilinks. This is because by "right clicking" a link, access to articles in other languages and or Reasonator like functionality is possible.

This is functionality where Wikidata is ready because it only relies on the interwikilinks. It does not rely on statements. This is where Wikidata is absolutely ready to do a superior job. What do you think? What did I miss?? Thanks, GerardM (talk) 07:52, 10 July 2015 (UTC)

Yes I think I agree. Until now, many Wikipedians have been trying to make Wikidata change its behavior about sitelinks whereas no one has spent much time rethinking the use of blue links, red links, and redirects on Wikipedia. I think there could be more colors to links, like maybe a brown link takes you to Wikidata? Jane023 (talk) 09:51, 10 July 2015 (UTC)
I doubt that I will be able to convince "my" wikipedians to link to Wikidata or Reasonator, when local articles are missing. But that may change in the future.
One challenge is definitly to identify a method to create red links by the help of Wikidata. If somebody was born in "Perth, Western Australia", and we do not have any local article about it. How would such a link look like and be created? To construct "Perth" ..", " .."Western Australia" may give us a link to the wrong article. This since there is more than one Perth in Western Australia. And there can be redirects we cannot imagine exists, who links to something completly different. Maybe construct something like [[Redlink Q123456789|Perth]]?
And how will we know when a link is considered obsolete. How do we add a Notability-policy into the AI of our Templates? -- Innocent bystander (talk) 10:59, 10 July 2015 (UTC)
The point is not only that it will do a better job on red links, it will also do a better job on ordinary links. Think about it, when you have a link and you are more fluent in Dutch, why read the English article? When there is no article, obviously any language you know (the #babel functionality) will do. The beauty is that editors do not have to do much. It is just functionality that is replaced.
When you convince "your" wikipedians, the point for them is NOT that it refers to Resonator or Wikidata. The point is that it does not link to nothing. The point is that disambiguation will be a whole lot cleaner. The point is that it will lead to increased quality. Thanks GerardM (talk) 22:00, 10 July 2015 (UTC)

Blogged again

Zolo provided new magnificent statistics. This allowed me to present my case in a new way. Set theory easily explains how to subdivide known items to a Wikipedia and those known to Wikidata. Given that Wikidata knows about most if not all Wikipedia articles, functionality based on the items themselves is easier than providing information in templates. The best part is that it enhances Wikipedia and does not take any of its shine away. Thanks, GerardM (talk) 06:50, 13 July 2015 (UTC)

All things Wikidata at Wikimania

Hey folks :)

Here are the sessions related to Wikidata that are happening at Wikimania this week: https://wikimania2015.wikimedia.org/wiki/Category:Wikidata Hope to see many of you there  :) --Lydia Pintscher (WMDE) (talk) 00:58, 13 July 2015 (UTC)

Enjoy the event. My office is still more appealing at that moment of the year. Snipre (talk) 02:51, 13 July 2015 (UTC)
Impossible! ;-) --Lydia Pintscher (WMDE) (talk) 02:57, 13 July 2015 (UTC)

Bibliographic items

Are there any examples of items about bibliographic data (eg. book chapters, articles) which can be used in references? --AS (talk) 11:00, 12 July 2015 (UTC)

@AS: Look at help:sources and Wikidata:WikiProject_Books. Snipre (talk) 22:08, 12 July 2015 (UTC)
@Snipre: Thank you, help:sources was helpful. --AS (talk) 07:55, 13 July 2015 (UTC)

Behaviour with the sources tool enabled

When I look at a Wikidata item, I typically am in search for something. Award received is often my target. When an item has source tool data, I find the result but immediately get yanked away because of the screen not being complete yet. I do care about this information but I do resent it that it shows as a second class citizen.

To me it is obvious that data that is from a serious source is probably of a higher quality than the statement we add by hand. Consequently the notion that someone has to "approve" them by hand may be a diplomatic necessity but at the same time it is bonkers. Thanks, GerardM (talk) 08:12, 13 July 2015 (UTC)

Suggestion: transfer redirections to wiki data

Hello.I suggest transfer redirections to wiki data Instead of repetition in all the projects , by making interwiki links and names Common conversions, and for scientific names, be in English --ديفيد عادل وهبة خليل 2 (talk) 11:20, 13 July 2015 (UTC)

This isn't international. What wiki A wants a redirect to point to is not the same as wiki B. There are some limited cases (people). --Izno (talk) 14:40, 13 July 2015 (UTC)
Izno, We can exclude some items (example) using a page in MediaWiki namespace --ديفيد عادل وهبة خليل 2 (talk) 15:13, 13 July 2015 (UTC)

"List of properties" needs some TLC

Is everyone struggling as much as me with finding a useful list of all properties? There seems to be an abundance of different ways of listing all properties that are either incomplete, inconvenient or too sparse to be useful.

Right now i found these ones:

  • Wikidata:List of properties/all - Should be the 'definite' overview, but lots of properties aren't listed because of Lua errors.
  • Wikidata:List of properties/Summary table - Should be sparser than the previous one, but suffers from the same problems: lots of Lua errors and, hence, no properties
  • Special:AllPages/Property: - More or less useful, but has paging so you need to click through. Also, not very descriptive.
  • Special:ListProperties - The only page on this list that actually has *all* properties on a single list, but virtually useless because it doesn't list property IDs or anything else but the label.

I don't know, but four different ways of listing all properties seems like overkill :)

Maybe i haven't found the 'definite' list, but if it is available, i think we should try to update documentation so people aren't guided to all these different, mostly non-functioning, pages.

Husky (talk) 16:23, 10 July 2015 (UTC)

It's at Wikidata:Database reports/Constraint violations/All properties. --- Jura 16:38, 10 July 2015 (UTC)
I see two options to fix those property lists:
  1. @Beta16: Make the bot that updates "List of properties/Summary table" enter the labels of the properties itself.
  2. File a bug report to add the property id's to Special:ListProperties.
Wikidata:List of properties/all is unsalvageable in my opinion and should be deleted.--Snaevar (talk) 13:19, 11 July 2015 (UTC)
Using items like Wikidata property for authority control for people (Q19595382), it should be possible to build lists for particular domains ("properties for people", "...for buildings", "...for works", etc), and with time, at greater granularity (".. for athletes", "...for musicians"). Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 14:24, 11 July 2015 (UTC)
One trick I have found very useful: You can search for properties by putting P: at the start of your search. Filceolaire (talk) 18:02, 12 July 2015 (UTC)
Right. I am thinking of changing the whole summary table of properties to simple wikilinks. That will get rid of these Lua errors. Any objections? --Snaevar (talk) 12:28, 14 July 2015 (UTC)

redundancy and categories, and properties

Just discovered some Wikidata property about Wikimedia categories (Q18667213) View with Reasonator View with SQID like category of people buried here (P1791) View with SQID I don't like this because it seems useless! we can just create a class people died in X, and add claims

< people died in X > subclass of (P279) View with SQID < human >
< people died in X > died in search < X >
...
< people died in X > category search < the cat >
...

I don't like the idea to multiply the categories properties. This links to complicated constraints with not much value added, isn't it ? it also installs a parallel class system to link with the old one, and overall it seems messy. See the classification on the property also. It's not really economic compared to the class solution for the same result. Plus the class will map to a query who will be able to list all the person who died there … is it worth to have this kind of properties ?

TomT0m (talk) 19:48, 13 July 2015 (UTC)

Don't know who to ping so I @Pigsonthewing, Jura1: some editors of the propertier page. TomT0m (talk) 19:58, 13 July 2015 (UTC)
Pick whatever makes it easiest for you to convert all those categories to properties. Once you are done, we can delete them.
Just bear in mind that the categories at Wikipedia are not necessarily named "people who died in X" for any "X" in every language. --- Jura 20:10, 13 July 2015 (UTC)
@Jura1: Sorry I don't have a clue of what you mean. TomT0m (talk) 20:24, 13 July 2015 (UTC)
Neither idea is satisfactory, frankly. --Izno (talk) 22:14, 13 July 2015 (UTC)
The correct way it should be done is category combines topics (P971) on the categories; in this case combining "death" or similar and "year". --Izno (talk) 22:17, 13 July 2015 (UTC)
@Izno: I don't think so. These categories in Wikidata have equivalents, very better defined : classes and the corresponding properties instance of (P31) and subclass of (P279). Eberything you want to express on the category label, ie. person who died in X is sayable into statement. Person by subclass of human, died in X by the statement you did. This should imho be done with the Help:Classification and Help:BMP solutions. Instances could be implicit by queries. Of course this could be done using the category item, but there is a concensus than category should not be used in the data model, but be mapped with the category entity. In the end maybe if you add subclass of (P279) to the category item we can use this as a class item. But this would add to the confusion beetween lists, classes and categories :) TomT0m (talk) 09:31, 14 July 2015 (UTC)
My point is essentially that the category property is enough and that I think the category for person who died in X should be deleted, as they can be decomposed into simpler one. TomT0m (talk) 09:31, 14 July 2015 (UTC)
@TomTom, maybe your usecase is different. What are you trying to do with these categories? --- Jura 05:14, 14 July 2015 (UTC)
@Jura1: I don't really think in term of usecase, the usecase here is clear : managing the persons who died in some place. I'm interesting into the best way to do this in Wikidata. In that sense, of course the old Mediawiki categories are to disapear, the mapping to them is useful of course, but I'm interested into how to go from a category based solution and a Wikidata one. Of course we will have to inform Wikipedians that they should better add a statement and a class on Wikidata than adding a Category on a Wikipedia page. TomT0m (talk) 09:31, 14 July 2015 (UTC)
If you don't have a usecase we can't really help you. --- Jura 11:46, 14 July 2015 (UTC)
@TomT0m: i don't quite get the idea of your proposal. Currently we are using arbitrary access to get from person to birthplace and to birthplace category (category for people born here (P1464) and category for people who died here (P1465)) and add it to result page wikicode. It is pretty simple and straightforward. Same can be used with property-in-subject category of people buried here (P1791). What is your idea how to include person into local wikipedia category? -- Vlsergey (talk) 10:12, 14 July 2015 (UTC)
@Vlsergey: The global plan : I would want to harmonize way to express stuff, as much as possible, around the class and instance concepts, because they are simple and well understood, both theoricaly and with the available toolset of languages outside of Wikidata. And for Wikidatans to talk a common language.
The specifics here : Imagine:
  • our category item corresponds to a class item on Wikidata. Those are linked with
    < class item > category search < cat item >
    .
  • The class item is linked to the location item with
    < class item > died in search < location item >
    .
With arbitrary access, or queries, we will be able to find the category item from the location item in a template. So we don't really need these properties.
Why should we do that ? There is a lot of similar categories, who essentially have an intended use similar to those of a class, or a query, or for which we can automatically list all the members with a query on Wikidata. So ... if we can harmonize the model for all those usecases this will help everybody. For example, with some class item having statements:
it becomes easy, just like reasonator does with list of statements, to automatically generates the query (CLAIM[31:(TREE[5][][279279])] AND CLAIM[:2005] AND CLAIM[20:New York]) ... except this model will work without any change for any of this types of usecases. We will be able to map queries to classes, populate categories or add statements, the work that autolists2 does, with a minimal effort of model harmonization, on a number of usecases.
So, with generic and powerful concepts, we gain actually a lot, either in terms on community communications if everybody understand that (which is not that hard), and the work on person will benefit as is to other usecases of classes mapped to categories. Instead of hiding knowledge into templates or in complex property definition, we make it explicit into class definitions. TomT0m (talk) 10:40, 14 July 2015 (UTC)
@TomT0m: With arbitrary access... we will be able to find the category item from the location item in a template. So we don't really need these properties. --would you mind to provide Wikidata entity and LUA code example how to get Category:Died in New York City (Q8365583) from New York City (Q60) without help of category for people who died here (P1465) (no queries, but with arbitrary access). -- Vlsergey (talk) 11:34, 14 July 2015 (UTC)
@Vlsergey: Of course with only arbitrary access we have to take care of the direction of the properties, and the lack of inverse properties/care not to introduce too much of them could be a blocker. That's the reason why I asked the devteam to introduce the simple query feature a while ago, but that seem to be abandoned :( With that limitation in mind, I would still prefer the introduction of a class of people who died here, while we don't have the ability to deal with inverse properties computation or a working query engine. In the wikipedia category we might find articles about an accident, for example. TomT0m (talk) 11:53, 14 July 2015 (UTC)
Well, personally I don't like classes thing, so weak oppose to this. Because I prefer direct properties instead of indirect queries. You plan to actually hide the link them from entity page (i.e. New York City (Q60)) don't you? Anyway, once wikidata will have fast LUA way to get the local category name, I don't care much how exactly it's stored on Wikidata. But functionality is in the first place for me. -- Vlsergey (talk) 13:21, 14 July 2015 (UTC)
@Vlsergey: It's interesting, what don't you like in classes ? It's very similar to categories. TomT0m (talk) 13:29, 14 July 2015 (UTC)
I prefer single object to be instance of single class. Replacing categories with classes will mean single object would be instance of multiple classes, often partially overlapping. This would be a mess. Also such classes will duplicate object properties themselfs. I.e. stating that person dead in New York we don't need to duplicate it in "class" or "instance of" field. -- Vlsergey (talk) 13:35, 14 July 2015 (UTC)
@Vlsergey: No we don't. It's possible to have an object member of a class implicitely if we want to, it's not implied that by using solid foundation for our classification system we have to explicitely add all the instance of (P31) statements. See Help:Classification#Classes_definition. We can imagine, for example, a gadget that, considering the item is in the results of a query, adds a section in the item page about all the classes he's a member of, maybe hidden and expansible (I don't know how, it's just hypothetical for now.) And no, sorting things is not a mess, it's the lack of sorting and powerful enough tools to sort correctly who create a mess :) That's why it's important to think about this tools in a global manner. Turns out these concepts and tools are studied for a very long time ... TomT0m (talk) 13:44, 14 July 2015 (UTC)

Outputting templates

Hi, a little bit of bikeshedding : I started a discussion about template formating in talk and project pages. There is several ways actually in use in Wikidata. To discuss their strengh and weakness, and to try to find a concensus, I started a discussion and wrote down a few examples on Template_talk:Statement. Please feel free to join. TomT0m (talk) 16:07, 14 July 2015 (UTC)

Commons category and Commons gallery

There was talk earlier about having Commons category and Commons gallery in the main boxes on the right side of the entry along with Wikipedia and Wikiquote links, as opposed to property fields. Is that still going to be done? You can add in "commons" in the "other" box to get to the gallery page, but not the category page. It isn't obvious to a newbie that this is the thing to do, especially since "commons" is the only entry allowed in that box. I almost never create a gallery page at Commons, I usually just create a category for that person. Some Commons editors delete gallery pages unless they have a minimum number of entries or delete if that person does not have a corresponding Wikipedia page. I add about 20 Bain collection images from the Library of Congress each week. --Richard Arthur Norton (1958- ) (talk) 17:34, 14 July 2015 (UTC)

Profession vs. occupation

Profession (Q28640) and occupation (Q13516667) appear to overlap. Any suggestions, or better way to distinguish which to use? Profession is a subcat for occupation but Category:Occupations appears under profession. It seems that profession requires specialized training and we define occupation as "hobby, work, pastime, professional sport..." When confronted with dual choices is it best policy to fill in both fields? Or should they be merged? --Richard Arthur Norton (1958- ) (talk) 20:22, 13 July 2015 (UTC)

They are two different things: occupation is a much broader concept. For instance, philanthropist (Q13472585) is an occupation because it does not need any training and a person cannot making a living out of philanthropy (Q185733). Andreasm háblame / just talk to me 02:35, 15 July 2015 (UTC)

Human degrees of separation for notability

(Also posted to notability discussion) It seems we have agreed that we can create entries for the spouse, parents, siblings, and children of notable people. Are we limiting that to one degree of separation? Once the spouse has an entry, if we apply the rule again, her parents and siblings are eligible for entries. I think the goal is to have one degree of separation from someone with a Wikipedia, Wikimedia Commons, Wikiquote, etc. entry. --Richard Arthur Norton (1958- ) (talk) 17:28, 14 July 2015 (UTC)

WD:N probably has your answer from a broad point of view; for your specific case yes, you can connect them as far as you want so long as there is some notable concept relying on the presence of those items. --Izno (talk) 20:36, 14 July 2015 (UTC)
The purposed solution is:
  1. Any people with a 3rd party source which describes their non-trival information (thus white pages are excluded) is notable.
  2. If a source describes the relation of Item1->Item2->Item3->...->Itemn, and Item1 and Itemn are notable because it contains a sitelink or meet #1, Item2, Item3, ... are notable.
  3. Any items linked from a item with a sitelink or meet #1/#2 are notable.

--GZWDer (talk) 06:38, 15 July 2015 (UTC)

Sorting of choices to populate a field

When I type in "New York" I see all the possible choices, the city, the county, the state. Are they sorted numerically by Q-number or sorted by the one with the most links to it? – The preceding unsigned comment was added by Richard Arthur Norton (1958- ) (talk • contribs).

@Richard Arthur Norton (1958- ): They are sorted by the larger number of the number of sitelinks and the number of labels. See phab:T94404.--GZWDer (talk) 06:41, 15 July 2015 (UTC)

Field attributes

How do I see field attributes? Some expect numbers, some expect a valid Q-value. --Richard Arthur Norton (1958- ) (talk) 03:54, 15 July 2015 (UTC)

Typically you go to the relevant page connected to the property in question. For example: IMDb ID (P345) says that it is a "string" it expects (under "Data type"). number of children (P1971) says "Quantity". director (P57) says "Item". And so on. Is this what you were looking for? Gabbe (talk) 05:52, 15 July 2015 (UTC)

Change order

Hello. Is there a way to change the order of datas in a statement? For example, in Q4243782 I want to change to move "2001-02 Cypriot Men's Volleyball Cup" to last place of statement "has part" with out delete it and write it again. Xaris333 (talk) 05:34, 15 July 2015 (UTC)

Links to constraint report for item

Have we considered having a local link on each page for the item's corresponding constraint report, or even a gadget that can enable that in situ? I think that there would be value in such being available.  — billinghurst sDrewth 00:23, 13 July 2015 (UTC)

+1. We should have a link under the tools menu in the left part of the frame, something like the "What links here" tool to avoid the need to move to a special page each time we want to acces to these information. Snipre (talk) 00:28, 13 July 2015 (UTC)
Short-term I think that's ok but I'd like it to be a temporary thing. Version 2 of the work the students have done already contains the code to show the violations on the item itself right next to the problematic statement. So there shouldn't be much need to go to the special page. We just need to get the code polished in the next weeks. --Lydia Pintscher (WMDE) (talk) 00:56, 13 July 2015 (UTC)
Hot. --Izno (talk) 02:12, 13 July 2015 (UTC)
Until version two of our work will go live, you might want to use https://www.wikidata.org/wiki/User:Jonas.keutel/ConstraintReport.js you can include in you commons.js. It displays a link to the constraint report on top of an entity's page. It also shows a link to the cross-check we perform against the GND. This extension is still in the code review process, so the link leads to our instance on labs. Jonas.keutel (talk) 09:58, 15 July 2015 (UTC)
@billinghurst: I'll also add a link to {{Item documentation}}. TomT0m (talk) 09:24, 13 July 2015 (UTC)
Perfect Lydia Pintscher (WMDE).  — billinghurst sDrewth 10:21, 13 July 2015 (UTC)

New Constraint ?

I was thinking about whether a blacklist constraint would be possible desirable. Specifically it would be something like "Property:PX cannot be (list of Q's or filenames or something)". I encountered this in coat of arms image (P94) where it would be nice to blacklist various placeholder images which otherwise tend to sneak in. /Lokal Profil (talk) 19:29, 14 July 2015 (UTC)

For filenames, you should be able to include them in the regex. For other item values, to some extent Template:Constraint:Conflicts with can do it. --- Jura 19:38, 14 July 2015 (UTC)
By definition this is a constraint (almost; distinction between a constraint as a whitelist and as a blacklist too minor to note). Any item which does not pass the regex/type/whatnot is "blacklisted". If you actually want to stop certain edits, I might suggest you seek the addition of an edit filter on WD:AN Wikidata talk:Abuse filter. --Izno (talk) 20:41, 14 July 2015 (UTC)
There isn't an actual "blacklist" constraint. --- Jura 08:36, 15 July 2015 (UTC)
@Lokal Profil: Template:Constraint:Conflicts with can provide such a blacklist, see Property talk:P31. Matěj Suchánek (talk) 09:27, 15 July 2015 (UTC)
Thanks. Hadn't thought about using {{Constraint:Conflicts with}} with the property itself. For filenames I guess {{Constraint:Format}} is then the solution. Can you use two separate {{Constraint:Format}} in the same talk page? That would make it cleaner to separate the (already existing) whitelist from the blacklist. /Lokal Profil (talk) 14:33, 15 July 2015 (UTC)
Of course, {{Constraint:Conflicts with}} is usually used separately for separate issues. Matěj Suchánek (talk) 15:03, 15 July 2015 (UTC)

Some geocoordinates from Freebase

I have uploaded to the Primary Sources tool a small dataset of 9,000 new geocoordinates. It is called freebase-coordinates and contains the Freebase coordinates for items that were not already geolocated. I have also created a list of the around 2,000 items which have a location that has a difference of more than 2 degrees from the location provided in Freebase. This list is available here. I believe it would be nice to have a look at them as sometimes it highlight an error in Wikidata. Tpt (talk) 20:57, 14 July 2015 (UTC)

Thanks :) However the page really needs splitting up, there's too many items so it just creates a lot of Lua errors.
What should we do with ones which have been fixed or where the error is in Freebase? Remove them from the page? If so, more sections would be nice (grouping by country or latitude or something), finding the thing I'm editing in a list of nearly 2000 items is not easy.
Would you be able to make a separate list of ones which would be less than 2 degrees if you ignore the direction? I've seen a few like that already.
- Nikki (talk) 22:23, 14 July 2015 (UTC)
I believe that as soon as you have checked the item you could remove it from the list. I have created a separated list of possible sign errors and I have sorted the two resulting list by Wikidata latitude. I let you split the list into subpages or create other subsection if you want. Thank you for your interest in this list, Tpt (talk) 23:33, 14 July 2015 (UTC)
@Tpt: Thanks! I've moved the possible sign errors to a subpage and started to go through them. I decided to remove the lines where I've fixed them in Wikidata, but strike out the lines where Freebase is wrong, that way there is a list of known exceptions if you (or anyone else) ever want to generate a new list. - Nikki (talk) 15:01, 15 July 2015 (UTC)
Great! Thank you very much. Tpt (talk) 17:01, 15 July 2015 (UTC)

Viola vs Viola di gamba

IHMO, Q40125 (viola) is superclass of Q4111981 (viola di gamba). Some wikipedia articles in viol (Q40125) seems more suitable in no label (Q4111981), check and move it, please. -- Sergey kudryavtsev (talk) 10:52, 15 July 2015 (UTC)

Automated reciprocal links

Will it be possible to have automated reciprocal links for family relationships in human entries? For instance, if I add someone in as the spouse of someone, can we have it reciprocate automatically. I add in the wife for a husband a bot adds in the husband to the wife entry, and do it for all relationship fields. Findagrave does this to reduce errors. --Richard Arthur Norton (1958- ) (talk) 19:20, 14 July 2015 (UTC)

This is something that we have long requested, the answer to which is apparently "No". See phab:T51165. --Izno (talk) 20:44, 14 July 2015 (UTC)
I fear bringing robots into this could lead to further errors. If someone adds a spouse (P26) that is incorrect, then automatic reciprocation would lead to that falsehood also being reciprocated. In addition, I've previously chosen not to add a P26 statement to the corresponding husband or wife because that person had been married several times and I wanted to add the complete list of spouses at once at some later time.
The problem doesn't appear to be very significant at the moment. Of over 31,000 P26 statements only 656 are referring to items that lack a P26 statement. That is few enough to be feasibly double-checked manually. For parent-child relationships, the vast majority of child (P40) statements also have a corresponding father (P22) and mother (P25) statement (although child (P40) is used more infrequently than the parent properties). Väsk (talk) 21:37, 14 July 2015 (UTC)
"The problem doesn't appear to be very significant" .. as mysteriously missing links get added? --- Jura 06:56, 15 July 2015 (UTC)
Yes, I realise that adding reciprocal statements adds somewhat to the workload (I've added a few myself). Although if you've made the effort to determine a husband-wife relationship, then the relative added effort of inserting that relationship in two places is trivial, even with Wikidata's interface.
Even if the relationship isn't mentioned directly on both items, it is still present in the database and can be inferred through tools like Reasonator. One of the advantages of redundant reciprocity is to make the actual data more solid by aiding in error detection and forcing double-checking. That gets lost if it is reciprocated automatically by a robot. Väsk (talk) 12:05, 15 July 2015 (UTC)
My take on this is that we should have a "what statements link here" option so we can see all the statements that have the current item as their object. Then we don't need reciprocal properties. Filceolaire (talk) 15:50, 15 July 2015 (UTC)
  • Most people leave the relationship fields empty, that is why most are complete, if you want to scale up imagine the work. If a couple have 15 children, then you have to add all 15 to each parent, then add 14 siblings to each child. That would require 210 transactions just to add in siblings. Look at Findagrave here for Max Freudenberg and see how many relationships were calculated automatically and are marked with an asterisk. The trick is to just name a person's parents and spouse, and generate the sibling links automatically. I could set it up. --Richard Arthur Norton (1958- ) (talk) 04:00, 16 July 2015 (UTC)

Restricting choices

Is there any interest in restricting choices for fields to a pull down menu. Manner of death should only have a few choices, just like the online entry form for a death certificate when issued by a state. It should be: accident, natural causes, suicide, homicide or killed-in-action or war death, or whatever the official title is, I will check later to see what is used in New Jersey for a war related death. Gender should be limited to the proper choices. This way we can ban outliers in the data set. --Richard Arthur Norton (1958- ) (talk) 02:22, 15 July 2015 (UTC)

There is something called "Wikidata:Database reports/Constraint violations". It lists what are (arguably) improper choices of values for various properties, which can then be checked manually and fixed. I for one don't think it's self-evident what would count as the "proper" set of choices for something like "gender" (see Property talk:P21 for part of this discussion). Gabbe (talk) 05:48, 15 July 2015 (UTC)
There are lots of thoughts on hinting and suggesting appropriate values. There are also, as Gabbe said, constraints so that bots can highlight suspicious values but there was a deliberate design decision made by the developers some years ago that the wikidata software itself would not prevent you from entering any value. The thinking behind this is that there are strange edge cases and the way to deal with these is by discussion here rather than by laws built into the code. Filceolaire (talk) 15:46, 15 July 2015 (UTC)
In fiction, there can be an endless range of possible statements. I read Star Trek novels and Death by war in New Jersey is nothing I would be suprised to find here. In the last book I read, two main characters died several decades before they were born. Their spaceship, Armageddons arrow crashed into itself, before it was constructed. -- Innocent bystander (talk) 06:32, 16 July 2015 (UTC)

Calculated fields

Are there any calculated fields like "age at death"? --Richard Arthur Norton (1958- ) (talk) 02:30, 15 July 2015 (UTC)

No. For Wikidata that isn't needed, Wikidata is meant to store information. Age of death is derived information, it can be generated everytime you need it. Someone's date of birth can't be generated, so needs to be stored. If you want to display it on a Wikipedia, you could use LUA to calculate the age of death bases on date of birth and date of death. Mbch331 (talk) 06:22, 15 July 2015 (UTC)
You could make a feature request for it. --- Jura 06:57, 15 July 2015 (UTC)
If it's well known how long a person live, but both their dates are not known, a property with an external source would be the only way how to provide the age at death. (The property should have "conflicts with: date of birth (P569); date of death (P570)" to prevent data duplication. I'm not familiar with any such case though.) Note that similar property number of children (P1971) has recently been created.
By the way, some infoboxes in cswiki already either current age, or age at death, depending on the dates. Matěj Suchánek (talk) 09:50, 15 July 2015 (UTC)
Well, we know that Methuselah (Q156290) claims to had been 969 years old when he died, but with only vague details about the time of birth and death. -- Innocent bystander (talk) 10:00, 15 July 2015 (UTC)
P1971 is not calculated by Wikibase --- Jura 10:31, 15 July 2015 (UTC)
No, it isn't, but some could claim it to be redundant to the family-statements. A new "age at death"-property could be created as a suplement to such statements if we like. -- Innocent bystander (talk) 11:42, 15 July 2015 (UTC)
Actually, I requested it because for most items where it's currently used, Wikibase couldn't calculate and display it automatically when some conditions are met. --- Jura 11:49, 15 July 2015 (UTC)
@Jura1: Out of curiosity, which conditions? -- Innocent bystander (talk) 13:00, 15 July 2015 (UTC)
For age, it could work with single P569/P570 of precision 8 or higher. --- Jura 13:13, 15 July 2015 (UTC)
Age at death, as popularly used in everyday speech, newspaper obituaries, etc., handles time in a different way that Wikidata. Wikidata purports to represent birth and death as points on the UT timescale (even though the user interface does not support correct entry of such points). But in everyday speech, a person's date of birth is regarded as just a name for the date in the local time where the person was born. Then the date of death is the name of the date in the local time where the person died. Finally the calendar is used to compute the age, ignoring time zones and time of day of the vital events. To handle this kind of calculation the way it is done in everyday speech, WikiData needs a new data type to represent a calendar date instead of a point on a timeline.
Another complication is that the birth and death dates may not be stated in the same calendar, so the calculation code must be able to convert between Julian and Gregorian calendars. Jc3s5h (talk) 12:25, 15 July 2015 (UTC)
More or less, it works at Recent deaths. --- Jura 12:51, 15 July 2015 (UTC)
Jc3s5h I think the new data type we need is a time interval i.e. a number with dimension seconds/minutes/hours/days/years.
Innocent bystander James Ussher (Q333481) worked out the dates of birth and death for Methuselah (Q156290) (and everyone else in the bible) and these were considered reliable enough that they were included in the margins of some editions of the bible. As soon as we sort out how to represent BCE dates we can add all these birth and death dates to wikidata. Filceolaire (talk) 15:40, 15 July 2015 (UTC)
There are subtleties with calendars. For most everyday purposes, a calendar day is considered a basic unit and people (and the law) are reluctant to look inside the day for the time of day; the bartender deciding if you're 21 doesn't want to know where you were born or what time you were born. But calendar months or calendar years are often viewed as approximations that can be improved if better records are found, especially for birth and death dates. Jc3s5h (talk) 16:27, 15 July 2015 (UTC)
When I got my drivers licence, the authorities did not ask if I was born in the morning or in the evening. The day I became 18, I had the right to get a license, no matter when the earth had made 18 full circles around the sun. Since we have leap years, it could have been the day after or the day before. -- Innocent bystander (talk) 17:17, 15 July 2015 (UTC)
We would have to make the calculation restricted to humans and exclude fictional characters, and quasi historical figures. --Richard Arthur Norton (1958- ) (talk) 17:34, 15 July 2015 (UTC)
Yes, but that can better be solved in the WP-templates. In the last Star Trek novel I read, some characters died several decades before they were born, since they participated in time travel. I doubt that we have any item about these characters, but I guess when we are connected to Memory beta at Wikia, they will have items of their own. -- Innocent bystander (talk) 06:37, 16 July 2015 (UTC)

P577 vs P1191

What should be the right way to add the premiere date of a film/TV-Movie? first performance (P1191) or publication date (P577)? --Escudero (talk) 21:02, 15 July 2015 (UTC)

I would say P577. P1191 looks more like it's meant for live performances like a theatre performance. Mbch331 (talk) 21:05, 15 July 2015 (UTC)

Editing labels in more than one language

I need to edit the labels for World Geodetic System (Q215848) in all the languages for which there is already an entry. If I use the user interface and am not logged in, I can see all the different labels in various languages when I click "In more languages". But if I am logged in, I can only see the English label. How can I make the user interface show me, and allow me to edit, all the languages?

In this case, the label is a proper noun, so is the same in all languages. Also, whoever created the label in non-English languages decided to use the initials instead of the spelled out name (WGS instead of World Geodetic System). This is reasonable, since the system is usually known by its initials. Jc3s5h (talk) 13:41, 15 July 2015 (UTC)

Enable the label lister gadget. Sjoerd de Bruin (talk) 13:49, 15 July 2015 (UTC)
Thanks Sjoerddebruin, but that doesn't seem to make any difference. Jc3s5h (talk) 13:56, 15 July 2015 (UTC)
The LabelLister gadget does not influence what is displayed, it has to be opened manually using the "Labels list" link on the top of the page (usually between "Read" and "View history". To influence the languages that are displayed by default, you can add more Babel boxes on your user page (where there currently only is the "en-N" one). However, that only displays that very languages, not all, as the LabelLister does. --YMS (talk) 15:09, 15 July 2015 (UTC)
As far as I am aware it no longer works. Thanks, GerardM (talk) 15:23, 15 July 2015 (UTC)
Weird, it's still working here. Have you tried the beta version? Sjoerd de Bruin (talk) 12:28, 16 July 2015 (UTC)

P447 as qualifier or statement?

I'm adding tv premiere dates for some movies, tv series and episodes. Until now I was adding original network (P449) as qualifier of publication date (P577), along with place of publication (P291) (see an example in Arabian Nights (Q623864)). I'm also adding premiere dates in other countries, and there is a wide range of cases: TV movies that premiere in theatres, pay television and "open" television premieres, direct-to-video releases...

I try to express different cases by qualifiers of publication date (P577). However, I see in the property talk that data should only be stored as statements. What I should to do? If we add original network (P449) as statement, how we can manage other cases? --Escudero (talk) 11:09, 16 July 2015 (UTC)

I suppose the topic of this thread should be "original network" (P449), not score by (P447).
Personally, I had been using P449 as a statement, to indicate the network that original broadcast the series. I don't think there is a general "network" property, but if you are really keen on having a qualifier on the publication date, maybe "publisher" (P123) can do. --- Jura 12:53, 16 July 2015 (UTC)

Display of old data in field

User:Joern/altLabels.js broken, again....

Hello,

I remember there is a special place to signal the broken tools, but couldn't find it in archives :/

Could someone please direct me to the right place ? Thanks --Hsarrazin (talk) 18:05, 16 July 2015 (UTC)

Phabricator or WD:Contact the development team. Mbch331 (talk) 20:24, 16 July 2015 (UTC)

Fusionproblems =

Can someone fix de:Kategorie:Neokeynesianismus (Q9504993) with english category en:Category:New Keynesian economics (Q9038619) ? Huztio (talk) 01:14, 17 July 2015 (UTC)

Got it. --Izno (talk) 03:48, 17 July 2015 (UTC)
Actually, this is wrong. Neo Keynesian and New Keynesian are not the same economics. --Izno (talk) 03:52, 17 July 2015 (UTC)

Redirects

The following example shows the problem. In the french Wikipedia the article 'Thrombose' is a redirect to 'Thrombus'. Hence because of the used system, 'Thrombose' cannot be linked to the similar articles in other languages. In my opinion a redirect should not automatically be linked to the article it directs to. So, the french article 'Thrombose' should not link to 'Thrombus', but to the cluster with the english title 'Thrombosis'.Madyno (talk) 10:43, 12 July 2015 (UTC)

bug T54564 deals with links to redirects. There is a page on Wikidata looking at using links from random pages to fix this but I can't find it just now. Filceolaire (talk) 17:58, 12 July 2015 (UTC)
There is a dirty trick as a workaround. You must transform the redirect to something else by removing the link and putting in some random text. Then you can add the link on wikidata. After that you can reset to the previous version with the redirect. The link remains unchanged in WD. The same applies if you link an article and move it afterwards. The link on WD remains on the redirect in this case, just make sure the new lemma is also linked, so a bot can not alter the link to the new lemma.--Giftzwerg 88 (talk) 13:20, 17 July 2015 (UTC)

National identification number

I am sure this has been discussed before, is there an interest in a national identification number for dead people? In the US it would be the Social Security Number (SSN) from the Social Security Death Index, Sweden assigns a number for historical figures based on their order of birth for the day they were born. Personal Code Number 18690118-780 for Frideborg Winblad. We also have "military number (MN)" for US servicemen. --Richard Arthur Norton (1958- ) (talk) 17:44, 15 July 2015 (UTC)

I would worry about making things too easy for identity thieves but limiting this property to dead people should address that. These we tend to have specific properties for each index we link to so we might need a separate property for each index (i.e. separate properties for SSN, MN, etc.) Best to propose one on Wikidata:Property proposal/Person and see what people say. Message me if you need help drafting the proposal. Filceolaire (talk) 22:09, 15 July 2015 (UTC)
I would say that Personal identity number (Q5453848) is a far to sensitive information to be allowed to add here. Winblad died in the 1960s, so there is maybe not much to worry about. But the temptation to add it to living or recently deceased people would be to strong. I have earlier oversighted such information when it was inserted to pages on Wikipedia. -- Innocent bystander (talk) 16:34, 16 July 2015 (UTC)
Then we make the field conditional to having a valid death date. That should be very easy to program in. The reason why we have a social security death index is to prevent fraud, not instigate fraud. It is so someone cannot takeover the identity of a dead person and get a credit card in their name. --Richard Arthur Norton (1958- ) (talk) 01:31, 17 July 2015 (UTC)
How does a software know that a death date is "valid" and not a fraud? And observe that the BLP-policy in some interpretations of it, protects the integrity of recently deceased people. And that somebody is dead, does not mean that they cannot own bank accounts, real estates etc. I have seen cases with a person who died in the 1910s, who just recently sold their real estate and suddenly become millionair. -- Innocent bystander (talk) 07:19, 17 July 2015 (UTC)

Unhappy about "primary sources tool" and import from Freebase

When I add data, my tools do not know if there is data waiting to be "curated". This word is actually wrong, all you can do is approve or reject. As a consequence of this booboo I do work that does not need to be done. I find that my time and the time of the original people at Freebase is wasted.

You will not see what I mean if you have not added the "primary sources tool" as a gadget. When you do, you will find that the UI jerks you around. You will find that "Freebase" is not given as the source... The implementation of this whole process does not pass the whiffy test. Thanks, GerardM (talk) 07:39, 16 July 2015 (UTC)

Sorry, but your message is very vague. Sjoerd de Bruin (talk) 10:37, 16 July 2015 (UTC)
  • Why don't you express those concerns on Wikidata_talk:Primary_sources_tool where they're more likely to be seen by the people working on the tool? In fact, one of your concerns has already been brought up as a response to one of your own posts there… --Mineo (talk) 10:36, 16 July 2015 (UTC)

Challenge

Please let me know how to automatically add the winners of the Hannah Arendt Prize. NB I only care about the award on the persons themselves.. Thanks, GerardM (talk) 10:21, 16 July 2015 (UTC)

My experience of Freebase beyond viewing specific items is somewhat limited, but let's give it a try (I got a crash course the other day).
The list is at https://www.freebase.com/m/0j427x3#/award/award_category/winners but not really exportable.
Given that their structure for awards is somewhat complicated, you can't just extract it from a standard view, sample at:
https://www.freebase.com/award/award_category/winners?instances= (click on the right of "Object/Value" to see the details).
It might work by querying freebase directly and then converting the result
At https://www.freebase.com/query enter
{
  "type": "/award/award_category",
  "name": "Hannah Arendt Prize",
  "mid": null,
  "winners": [{
    "year": null,
    "award_winner": [{
      "name": null,
      "mid": null
    }]
  }]
}


This gives you an oddly formatted result. You should be able to convert this into the list like:

 Navid Kermani   /m/0b6fw_d     2011
 François Jullien        /m/09rx5m7     2010
 Kurt Flasch     /m/05w26l7     2009
 Victor Zaslavsky        /m/0gysxnt     2008
 Tony Judt       /m/0cl84m      2007
 Julia Kristeva  /m/046gs       2006
 Vaira Vīķe-Freiberga    /m/01vcbj      2005
 Ernst-Wolfgang Böckenförde      /m/0574v4b     2004
 Michael Ignatieff       /m/0g3l5v      2003
 Gianni Vattimo  /m/07k8bc      2002
 Ernst Vollrath  /m/0n9gz8d     2001
 Daniel Cohn-Bendit      /m/01p1ft      2001
 Yelena Bonner   /m/04mr7_      2000
 Massimo Cacciari        /m/0ct_72      1999
 Antje Vollmer   /m/02bym1      1998
 Claude Lefort   /m/07xsdp      1998
 Freimut Duve    /m/0c34nss     1997
 Joachim Gauck   /m/0b6g1f1     1998
 François Furet  /m/05dw45      1996
 Ágnes Heller    /m/06s96n      1995

Ideally, one could use https://tools.wmflabs.org/wikidata-todo/multibeacon.php to convert the MID (second column) to Wikidata's QID:

Here this gives you:


P646    Q
/m/0b6fw_d
/m/09rx5m7      Q929177
/m/05w26l7
/m/0gysxnt      Q4011670
/m/0cl84m
/m/046gs        Q159876
/m/01vcbj       Q151301
/m/0574v4b      Q69183
/m/0g3l5v
/m/07k8bc       Q159648
/m/0n9gz8d
/m/01p1ft       Q76837
/m/04mr7_
/m/0ct_72       Q369756
/m/02bym1
/m/07xsdp       Q373036
/m/0c34nss
/m/0b6g1f1      Q2538
/m/05dw45
/m/06s96n       Q235125


Doing the same with the file found at http://tools.wmflabs.org/wikidata-todo/beacon.php gives you the a slightly better result:

MID     QID     Year
/m/0b6fw_d              2011
/m/09rx5m7      Q929177 2010
/m/05w26l7              2009
/m/0gysxnt      Q4011670        2008
/m/0cl84m       Q312279 2007
/m/046gs        Q159876 2006
/m/01vcbj       Q151301 2005
/m/0574v4b      Q69183  2004
/m/0g3l5v       Q311684 2003
/m/07k8bc       Q159648 2002
/m/0n9gz8d              2001
/m/01p1ft       Q76837  2001
/m/04mr7_       Q446808 2000
/m/0ct_72       Q369756 1999
/m/02bym1               1998
/m/07xsdp       Q373036 1998
/m/0c34nss              1997
/m/0b6g1f1      Q2538   1998
/m/05dw45       Q318667 1996
/m/06s96n       Q235125 1995

The ones without any QID either don't have an item at Wikidata yet (or more likely, Wikidata lacks the mapping).

Now you can convert the result for upload with https://tools.wmflabs.org/wikidata-todo/quick_statements.php

        P166    Q1575780        P585    +00000002011-00-00T00:00:00Z/9  S143    Q1453477
Q929177 P166    Q1575780        P585    +00000002010-00-00T00:00:00Z/9  S143    Q1453477
        P166    Q1575780        P585    +00000002009-00-00T00:00:00Z/9  S143    Q1453477
Q4011670        P166    Q1575780        P585    +00000002008-00-00T00:00:00Z/9  S143    Q1453477
Q312279 P166    Q1575780        P585    +00000002007-00-00T00:00:00Z/9  S143    Q1453477
Q159876 P166    Q1575780        P585    +00000002006-00-00T00:00:00Z/9  S143    Q1453477
Q151301 P166    Q1575780        P585    +00000002005-00-00T00:00:00Z/9  S143    Q1453477
Q69183  P166    Q1575780        P585    +00000002004-00-00T00:00:00Z/9  S143    Q1453477
Q311684 P166    Q1575780        P585    +00000002003-00-00T00:00:00Z/9  S143    Q1453477
Q159648 P166    Q1575780        P585    +00000002002-00-00T00:00:00Z/9  S143    Q1453477
        P166    Q1575780        P585    +00000002001-00-00T00:00:00Z/9  S143    Q1453477
Q76837  P166    Q1575780        P585    +00000002001-00-00T00:00:00Z/9  S143    Q1453477
Q446808 P166    Q1575780        P585    +00000002000-00-00T00:00:00Z/9  S143    Q1453477
Q369756 P166    Q1575780        P585    +00000001999-00-00T00:00:00Z/9  S143    Q1453477
        P166    Q1575780        P585    +00000001998-00-00T00:00:00Z/9  S143    Q1453477
Q373036 P166    Q1575780        P585    +00000001998-00-00T00:00:00Z/9  S143    Q1453477
        P166    Q1575780        P585    +00000001997-00-00T00:00:00Z/9  S143    Q1453477
Q2538   P166    Q1575780        P585    +00000001998-00-00T00:00:00Z/9  S143    Q1453477
Q318667 P166    Q1575780        P585    +00000001996-00-00T00:00:00Z/9  S143    Q1453477
Q235125 P166    Q1575780        P585    +00000001995-00-00T00:00:00Z/9  S143    Q1453477

The lines without a QID wont work, obviously. If there ways to simplify any of the steps, I'd be glad to know. --- Jura 11:42, 16 July 2015 (UTC)

Doing the query slightly different, I found that all awardees but Q1360461 have an enwiki article listed in Freebase. So it should be possible to find QIDs for the entire list. --- Jura 12:58, 16 July 2015 (UTC)
What I am seeking is to use the data that shows when the gadget is activated. As it is I feel enormouslyt frustrated because their work or my work is redundant. The assumption that I want to do this by hand is  !@#$%^ GerardM (talk) 13:28, 16 July 2015 (UTC)
I think the primary use of the tools is to get a feel of the quality of the data and potential issues. What I have seen for awards is that the date is sometimes incorrect, but maybe it's my interpretation (on July 1, 2015 one could get the 2014 or 2015 or 2016 award, depending on how they label it). The other issue is that the mapping of people is incomplete (see above). Obviously, it's easier to check award data per award than per awardee. Maybe, this just means that the property should is the wrong way round.
Maybe it's possible to interrogate the data through http://quarry.wmflabs.org/ ? --- Jura 14:59, 16 July 2015 (UTC)
Strong +1 to what Jura1 (talkcontribslogs) said. The current state of Primary Sources allows to see the quality of mapped data and to have feedbacks about it. It is really valuable to me in order to improve the mapping work. What I would like to do in the future is to build an API and a user interface allowing to do mass actions on Primary Sources content (retrieve/accept and add to Wikidata/reject the statements matching a given query...). I hope it'll be a good compromise between efficiency and curation needs. Tpt (talk) 19:00, 16 July 2015 (UTC)
I've just added a new feature to the Primary Sources backend API in order to do simple searches into its content. Here is a sample with the winners of Hannah Arendt Prize: https://tools.wmflabs.org/wikidata-primary-sources/statements/all?property=P166&value=Q1575780&limit=100 A small doc is here. Tpt (talk) 17:09, 17 July 2015 (UTC)

Proposal

I agree that the "Primary sources tool" is useful to work with mappings. That is imho its role and place. When we are to import data from sources, either through the "Primary sources tool" or directly, we should have additional functionality. In my blog I wrote a proposal that will have imho significant benefits. It will improve transparency, it will improve quality both at Wikidata and the source. Have a read and let us know what you think. Thanks, GerardM (talk)

What P to use for a CEO?

I tried to add chief executive officer (Q484876) to Philippe Fortunato (Q20670264), however I can't use occupation (P106), because that gives a constraint error for value type. So checked another (former) CEO and there it was added to position held (P39). When I did that for Philippe Fortunato (Q20670264) I still got a constraint error on value type. So what is the correct P to use? Mbch331 (talk) 14:42, 16 July 2015 (UTC)

P169 on Q759210? --- Jura 16:55, 16 July 2015 (UTC)
According to P169, there has to be an inverse as well and thats position held (P39). Added that and according to Special:ConstraintReport/Q20670264 there is a violation on ValueType. I checked and a few levels up chief executive officer (Q484876) has subclass of (P279) = position (Q4164871). Mbch331 (talk) 20:23, 16 July 2015 (UTC)
The requirement for P39 was recently added by Matěj Suchánek. Personally, I don't think function within a company should be used in P39. --- Jura 10:50, 17 July 2015 (UTC)

P2000

I have just created CPDL ID (P2000) - another milestone. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 09:18, 17 July 2015 (UTC)

Locations

The Studio Museum in Harlem > Harlem > Manhattan > New York City > New York > United States > North America > Earth ...
When I am describing The Studio Museum in Harlem would I enter into the field "located in the administrative territorial entity" the next entity in the location chain only? Some entries add just the next one in the chain, some add the next two, some add the next three. What is it best policy? --Richard Arthur Norton (1958- ) (talk) 18:16, 14 July 2015 (UTC)

The next one of up the chain only. What you might be seeing in some items is either: a) legacy imports that are not best practice; b) "deprecated" values which nonetheless were true at some point; or c) places where the "next one up the chain" is not obvious. (Innocent bystander (talkcontribslogs) likes to moan and grown about Sweden. ;)) --Izno (talk) 20:39, 14 July 2015 (UTC)
The problem I have here in Sweden, is that there is no simple hierarchy. It's like if Harlem would partly would have been in Brooklyn, Manhattan partly in New Jersey, and New York state partly in Canada. When I follow such a hierarchy, The Studio Museum would easily be located in Brooklyn, Canada.
I have no solution for it yet. The solution I have used this far, is to add the lowest levels which have a hierarchy. That means I add the municipality, but not the county, since there is a hierarchy beetween them. But I also add the province, since there is no simple relation between a province and a municipality. There are also smaller entities here, like civil parishes, parishes and districts. They are smaller, and they neither have in a simple hierarchy. A parish can be located in several municipalities, same thing with civil parishes and districts. Some say there is a hierarchy between the civil parishes and the provinces, but I am not sure about that. -- Innocent bystander (talk) 05:27, 15 July 2015 (UTC)
One very strange example here is that Särna (Q10688119) used to be both in Sweden and Denmark/Norway at the same time. Ok, that was 1645-1751, but it's still an example when real life is much more complicated than the simple structures we tend to create guidelines based on. -- Innocent bystander (talk) 06:52, 15 July 2015 (UTC)
What generally works is to add a stable level. Most of the time this is the municipality or the state. It helps locate the item. The other layers can be calculated from the coordinates. --- Jura 08:33, 15 July 2015 (UTC)
There is no stable level. 80.134.81.115 22:12, 17 July 2015 (UTC)

date of birth (P569) and date of death (P570) accuracy

Hello! Is it possible to get a list of items which have date of birth (P569) and date of death (P570), but only with a year, not the exact date? I would especially be interested in items which also have Database of Classical Scholars ID (P1935). Because the DBCS has exact dates in almost all cases. Jonathan Groß (talk) 08:59, 18 July 2015 (UTC)

You could try TAB, but it doesn't show the accuracy. Usually the ones ending by "01-01" or "00-00" are year only. --- Jura 09:48, 18 July 2015 (UTC)

Thanks for the help. There were some false positives in this, but not too many. Jonathan Groß (talk) 11:05, 18 July 2015 (UTC)

Badge updates?

After I have seen some updates of featured articles without any change of badges here, I have had some thoughts about how we should detect such missing badges. sv:Module:Utvald now helps to identify articles with featured article-templates with missing badges. Such articles can now be found in sv:Kategori:Wikipedia:Utvalda artiklar utan korrekt badge på Wikidata. But I have no idea of how we should detect badges that has been added by mistake or when removal of status as featured articles hasn't been confirmed by the the removal of the corresponding badge. -- Innocent bystander (talk) 09:22, 18 July 2015 (UTC)

Connection to other genealogy databases

We connect to "genealogics.org personID" which seems like a person's database, not a national database. Should we connect to the Wikia project Familypedia? or Geni.com (all trees merged into a forest)? or the LDS Familysearch-Familytree (requires free registration to view the records). Familysearch-Familytree connects a person to census records, birth, marriage and death records, parents and children and spouses. Unlike Ancestry.com, all the individual trees are merged into one forest. --Richard Arthur Norton (1958- ) (talk) 18:11, 15 July 2015 (UTC)

In general wikidata has been open to lots of what the librarians call 'authority control' indexes (the author is John Smith but which John Smith?). I these would probably be accepted. Wikidata does allow properties that link to indexes that are not reliable sources (it makes it easy for them to take info from us) so that shouldn't be an issue. Propose a new property on the Wikidata:Property proposal/Authority control or the Wikidata:Property proposal/Person page. Filceolaire (talk) 22:02, 15 July 2015 (UTC)
If there's a solid single-tree project out there which is willing to make minimal data open to view, this would probably be worthwhile. I believe we picked up genealogics mainly because it was open and scrapable, rather than selecting one site over another. Andrew Gray (talk) 19:13, 18 July 2015 (UTC)

Primary sources 404s?

Trying to verify a couple of primary sources links (i.e. imported from Freebase), I've discovered that some reference URLs produce 404. I wonder if it'd be possible to scan through the database and automatically remove URLs that are no longer alive from proposed references? --Laboramus (talk) 08:20, 16 July 2015 (UTC)

remove them is a little drastic as sometime the document still is there but the uri scheme on the website just changed, this happen more often than it sould (cool uris don't change ...) TomT0m (talk) 09:56, 16 July 2015 (UTC)
Imho already broken URLs have good chances to be broken again in the future if we fix them. So, my opinion would be to just remove them. I have opened a bug on Primary Sources repo about it. Tpt (talk) 19:02, 16 July 2015 (UTC)
You're assuming that a bad hoster is correlated to bad source quality ? That remains to be proven :) I corrected a URL about an authority in chemistry just yesterday: http://goldbook.iupac.org/ there was broken uris on their site. I say this happens even to the best. TomT0m (talk) 19:17, 16 July 2015 (UTC)
What I am saying is that an already broken URL has more chances to be broken again. It could be rephrased: a bad hoster as good chances to keep being bad. There are no relations with data quality in my statement. Tpt (talk) 19:59, 16 July 2015 (UTC)
Not sure I agree. Anyway a url is a hint and if you're interested enough you'll correct it if possible, I did it many times (a title, even encoding in a url, can be enough). If it has been deleted, it's possible you'll miss a good source, and we all want good sources. If it's bad or really uninformative, then it can be manually deleted later, no problem. TomT0m (talk) 20:08, 16 July 2015 (UTC)
Sure, if we have a way to automatically fix URLs it would be better. But I don't see any simple way to do it. Do you have one? Tpt (talk) 20:18, 16 July 2015 (UTC)
No I don't, but here we're in semi automated data curation mode ... so curating and verifying urls is a part of the game imho. If they are deleted automatically before someone has a chance to check, it's potential data loss. Moreother this is a problem who is not specific to freebase imported urls, the same problem will appear for pure Wikidata url, and already happens a lot on Wikipedia so ... I don't think deleting all 404 uris is really a viable option in the long term. Maybe we can check if there are archived in some internet archive, or elsewhere (with frwiki archiving system for example). TomT0m (talk) 20:52, 16 July 2015 (UTC)

Cross ref doi project

@Tpt: Just read a word about url peremtion problem on today's raw on frwiki : the lifetime of a url seems to be 6 years and it promise to be a challenge into Wikipedia, so also here. We better have to think about this :) They mention the crossref project Crossref (Q5188229) View with Reasonator View with SQID whose goal is to give dois to every uris to have a stable id. This reminded me of a solution we talked of with the devteam on the beginning of the project : use item instead of urls for sources, which in the case of a broken url used a lot as a source, would ease a lot the correction as the other metadatas about it would help to find it again, with the non stability of uris in practice problem in mind. I still think its a good idea, but it would be a community decision as it adds to the workload of course. With the wikipedias doing the same ... Wikidata could be an alternative solution, more opened to everybody than dois. author  TomT0m / talk page 10:22, 18 July 2015 (UTC)

I left a comment on the wikimedia blogpost about this project : https://blog.wikimedia.org/2015/06/01/preserving-wikipedia-citations/#comment-24411 Maybe @Lea Lacroix (WMDE): could contact them ? author  TomT0m / talk page 10:22, 18 July 2015 (UTC)

Hey :) There is already a WikiProject here on Wikidata for this and afaik some people related to crossref are there as far as I know. So there is probably the best starting place. Sorry I don't have a link at the moment. (Bad internets...) --Lydia Pintscher (WMDE) (talk) 20:04, 18 July 2015 (UTC)

Dokonjō Gaeru (Q4388674)

item need to be divided someone who speaks Japanese is needed --Rippitippi (talk) 19:52, 18 July 2015 (UTC)

Schema.org metadata in Wikipedia?

Hi Folks,

Sorry if this info is posted in another place but I've been searching rigorously and cannot find a definite answer! So here it goes, I want to add Schema.org metadata to the Boca Raton, FL Wiki page. Is this possible? I tried to add http://schema.org/GeoCoordinates and http://schema.org/Place (https://www.wikidata.org/wiki/Property:P625) as a statement in the coordinate locations data type for the page. But it doesn't seem to work, I can only extract the coordinates by using "geo". Schema.org uses different attributes in the HTML tags ie: itemscope, itemtype, itemprop. So just as geo is being extracted I would like to use "latitude", "longitude", and "elevation". This is what I am wanting to do for Boca:


<div itemscope itemtype="http://schema.org/Place">

     <h1>What is the latitude and longitude of the <span itemprop="name">Empire State Building</span>?</h1>
     Answer:
     <div itemprop="geo" itemscope itemtype="http://schema.org/GeoCoordinates">
       Latitude: 40 deg 44 min 54.36 sec N
       Longitude: 73 deg 59 min 8.5 dec W
       <meta itemprop="latitude" content="40.75" />
       <meta itemprop="longitude" content="73.98" />
     </div>
   </div>

This way the Boca coordinates could be extracted using a Schema.org standard format.

Thanks, meMelster

Why are you discussing adding information to a Wikipedia page (I guess you mean en:Boca Raton, Florida at WikiData? Why wouldn't you discuss it at Wikipedia? Jc3s5h (talk) 23:39, 18 July 2015 (UTC)
Currently microdata and rdfa support in wikitext is not enable in Wikimedia wikis. So I think it is something that should be fixed before talking about adding schema.org annotations in articles. Tpt (talk) 01:17, 19 July 2015 (UTC)
The Infobox on en:Boca Raton, Florida already emits hCard and Geo microformats. See Wikipedia:WikiProject Microformats. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 09:02, 19 July 2015 (UTC)

NLI from VIAF

Hello! Is there an (half) automated way to check all Wikidata items with VIAF identifiers against the VIAF database, and if there is a NLI identifier, add that to the Wikidata items? Jonathan Groß (talk) 13:37, 19 July 2015 (UTC)

This used to be part of Magnus Manske's AuthorityControl gadget (User:Magnus_Manske/authority_control.js), but has been disabled. Mbch331 (talk) 14:01, 19 July 2015 (UTC)

This may be due to that the NLI doesn't use stable identifiers. They don't seem to mind, though, as they assured me that alterations of the NLI identifiers are rare. @Magnus Manske, Kolja21:: Thoughts? Jonathan Groß (talk) 14:11, 19 July 2015 (UTC)

Add Commons as a sister site?

I'm a big fan of using the "Other sites" box to link to Commons categories to enable backlinks to Wikipedia articles. I've always thought it to be a bit odd that Wikibooks/Wikivoyage/et al. have their own boxes to add such a link, though, while Commons is relegated to the "Other" box. Why is this the case? Could Commons have its own box, or could the other boxes be merged together so that Commons isn't the odd one out? Thanks. Mike Peel (talk) 21:54, 4 July 2015 (UTC)

Hey, it's in the same box as Wikidata!
I think the reason is that there is just one link to each of them. --- Jura 21:58, 4 July 2015 (UTC)
See phab:T102417. - Nikki (talk) 10:20, 5 July 2015 (UTC)
@Nikki: thanks for the pointer! Mike Peel (talk) 16:47, 6 July 2015 (UTC)
  • Remember there are two types of links to Wikimedia Commons: A category link and a gallery link, I think we need both in the "Other sites" box. Some entries have one or the other or both. In the past, galleries for non-notable people have been deleted, even if the category stays. I don't know if the inclusion policy for a gallery is now more liberal. "Category:Foo" vs just "Foo". --Richard Arthur Norton (1958- ) (talk) 00:22, 13 July 2015 (UTC)
"Remember there are two types of links to Wikimedia Commons" - as for all Wikipedias. Nothing special. Deletion of items in WD if there is a category in Commons is evil.80.134.89.98 22:57, 19 July 2015 (UTC)

Additional badges

Many Wikipedias (especially Swedish, Cebuano and Waray Wikipedia) have a high amount of bot created articles. The numbers are too high, that the bot-created articles can't be queried by CatScan anymore. The creator of an article is language specific (see #Properties for local wikipedia needs) and can't be done with a Wikidata statement.

Therefore I would like to introduce a badge "Bot generated". With badge qualifiers it could also be stated, which bot (e.g. Lsjbot) has created an article. Later, when badges can be queried, it would be easy to query who (bot or human) has created a specific article.--Kopiersperre (talk) 09:43, 18 July 2015 (UTC)

As I have said in other places, I prefer to not see Wikimedia:Ownership of articles (Q4663994) installed into the framework of Wikidata. -- Innocent bystander (talk) 10:30, 18 July 2015 (UTC)
Bots can't own articles. It's only for statistics. When you take a look at Statistics/Wikipedia you will notice the high amount of taxons in some languae versions.--13:00, 18 July 2015 (UTC) – The preceding unsigned comment was added by Kopiersperre (talk • contribs).
And neither should carbon based users own articles. That is my main point.
I am fully aware of that some projects have a large amount of bot-created articles. I have run such projects myself. Should we also list who has reviewed and removed the bot-created-templates in these pages? You see, these templates/categories are regarded as temporary, even if many of them will stay for years. You also have to be aware of that many botcreated pages has been created by users who never have had a botflag. And many of them have never had any bot-created-template or category. For example my bot-created articles have never had such a template or category. -- Innocent bystander (talk) 14:15, 18 July 2015 (UTC)
@Innocent bystander: I guess, that German Wikipedia has highest requirements on taxon articles. There are only 42,796 on roughly 100m speakers. Swedish Wikipedia has 1,326,962 taxons and 10 million speakers. Do you really think, that all these articles will be improved in sufficient time (2050)?
It would be no problem for me, if some bot articles get missed. I just want to have statistics about the top article-creating bots.--Kopiersperre (talk) 15:08, 18 July 2015 (UTC)
A badge would not change anything. Especially not for wikidata with it's nearly 2,000,000 taxa. --Succu (talk) 15:28, 18 July 2015 (UTC)
This is a specific problem of one WP. I don't think WD should deal with that problem. A bot and a specific tag on the considered aticles can handle that without creating a new feature on WD. Statistic of WP should be managed by the corresponding WP. Snipre (talk) 13:23, 19 July 2015 (UTC)
This is not wiki-specific. Nearly all languages have created articles with bots. But onyl sv, war, ceb, nl and vi have done this on a big scale.--Kopiersperre (talk) 14:17, 20 July 2015 (UTC)

Querying badges

I'm trying to get a list of articles in cawiki with quality article badge, but I can't find any way to get it, since https://wdq.wmflabs.org/api_documentation.html doesn't mention badges, nor other tools do. Is there a simple way to get a list of articles (or items) selected by badge?--Pere prlpz (talk) 10:26, 18 July 2015 (UTC)

It's not possible at the moment. A solution is being worked out at phabricator:T72209. --Lydia Pintscher (WMDE) (talk) 19:53, 18 July 2015 (UTC)
I suppose that retrieving badges from a single element every time with bot could be enough for me, but I see that the API doesn't give it - as far as I can see. Is there a simpler way than parsing the html item page?--Pere prlpz (talk) 11:27, 19 July 2015 (UTC)
I think it looks like it does. I am not using API in sv:Modul:Utvald, but the data retrieved by mw.wikibase.getEntityObject() in Scribunto looks almost like a API-question. -- Innocent bystander (talk) 12:11, 19 July 2015 (UTC)
It can be done using lua and templates - it was done in cawiki today by Joancreus using ca:Plantilla:Article de qualitat and ca:Mòdul:Article de qualitat. Now it categorizes quality articles without the badge. Now we still need to check articles with the badge but without the template, but I suppose we should wait for badge queries. Thank you.--Pere prlpz (talk) 17:59, 20 July 2015 (UTC)

Wikidata weekly summary #167

Comments

  • The presentation about the future of templates gave me a braingasm. This is so right ! good bye issues about whether or not we have to solve property aliases in template calls @Lea Lacroix (WMDE): any comment ? The references questions seems also promising and in line with the other thread about primary source tool here with Tpt. author  TomT0m / talk page 16:33, 19 July 2015 (UTC)
    • Sorry can you please rephrase the question? I don't understand it. (It's pretty late right now after the Wikimania party...) --Lydia Pintscher (WMDE) (talk) 06:23, 20 July 2015 (UTC)
      • Sorry, That was indeed confused :) But as I understand they plan to be able to use Visual editor to edit the visualisation part of the templates, and to have first class infoboxes. This would mean we would get rid of the ugly parser functions and dilemna such as whether or not we should be able to lookup properties with aliases on wikitext. So actually it's totally related to the 100 mail thread on the mailing list. I thought if would be nice to have a comment about this from you ;) author  TomT0m / talk page 11:03, 20 July 2015 (UTC)
        • Yes at some point in the future that should be possible but we're still quite a bit away from that and I don't think it completely solves the issue unfortunately. --Lydia Pintscher (WMDE) (talk) 16:26, 20 July 2015 (UTC)

Q about a property proposal

Disclaimer: This is not my q, but a q about this q: Wikidata:Property proposal (Q17586656). Why am I taken there when I type Wikidata:Property proposal in the search field and not to Wikidata:Property proposal? Jane023 (talk) 10:28, 20 July 2015 (UTC)

I do not have a answer to your q, but I have a workaround: Type "WD:Property proposal" instead. -- Innocent bystander (talk) 10:37, 20 July 2015 (UTC)
Somehow search defaults to Q and P only and not to internal pages. Mbch331 (talk) 10:46, 20 July 2015 (UTC)
I think it's just Q. Would be good to include P as well. --- Jura 10:51, 20 July 2015 (UTC)
Well, it looks like I find the P-namespace when I search. -- Innocent bystander (talk) 10:57, 20 July 2015 (UTC)
Search yes, but not the suggestions. --- Jura 10:58, 20 July 2015 (UTC)

What is that Q anyway? Nothing links to it. I was actually looking for the original discussion about a property and it seems to me that when properties are created, they should link to the archive spot for such discussions, no? It's really hard to find otherwise. Jane023 (talk) 11:04, 20 July 2015 (UTC)

Ah - answering my own q here for future reference: Go to the property page, click "What links here", then select namespace "Wikidata" and you should see the proper Property proposal archive page. Jane023 (talk) 11:28, 20 July 2015 (UTC)
Easy! ;) -- I suppose we could add it to the top of this page as well. --- Jura 11:30, 20 July 2015 (UTC)
The archived proposal is linked from the property's talk page too, e.g. where it says "Proposal discussion" on Property talk:P1938. - Nikki (talk) 12:01, 20 July 2015 (UTC)
Thanks - I didn't see that, and that helps a lot. Jane023 (talk) 13:12, 20 July 2015 (UTC)

Official website URL - http://www.wikidata.org or http://www.wikidata.org/ or https://www.wikidata.org/?

We have various forms for official website (P856) - e.g. for Oneworld (Q8787) we have http://www.oneworld.com without the ending slash, but for many others we have ending slashes. Also, some sites may redirect to https by default, while some may be http. This may lead to some confusion - e.g. for Oneworld (Q8787) primary sources tool proposes http://www.oneworld.com/ not realizing it's the same URL. Should we adopt some guidelines about how we represent official website URLs? --Laboramus (talk) 20:59, 13 July 2015 (UTC)

Symbol strong support vote.svg Strong support. We should have a guideline about it. It would be very useful for Primary Sources. Tpt (talk) 22:05, 13 July 2015 (UTC)
Regardless we should prefer secure HTTP. But anything else really doesn't matter. I don't see a need for guidelines. --Izno (talk) 22:15, 13 July 2015 (UTC)
@Izno: The main use case I see for a guideline is for comparison and search. Currently if I want to get the entity that has as official website "http://foo.com", I should look for "http://foo.com", "http://foo.com/" and maybe even "https://foo.com" or "http://foo.com/index.html"... If we have a convention, I would have just to look for the normalized URL according to this convention. But, yes, what the convention is doesn't really matter. Tpt (talk) 16:23, 14 July 2015 (UTC)
From what I can see, the linksearch finds them all, no matter how you write them. -- Innocent bystander (talk) 16:47, 14 July 2015 (UTC)
@Innocent bystander: Try looking for google.com - I don't see Google (Q95) in the results. And if you use wildcards, you get so many results it becomes useless. --Laboramus (talk) 08:15, 16 July 2015 (UTC)
That finds all partial matches too, e.g. see [6], so that doesn't solve the problem of determining whether or not two URLs are actually the same. - Nikki (talk) 18:00, 14 July 2015 (UTC)
If we could avoid listing http://www.wikidata.org/index.html or https://www.wikidata.org/wiki/ and https://www.wikidata.org/wiki/Wikidata:Main_Page ;) --- Jura 05:17, 14 July 2015 (UTC)
Symbol strong support vote.svg Strong support, but (parts of) the guideline should apply to any URL, not just official website (P856) – for instance, many reference URL (P854) are also available in both HTTP and HTTPS, and we should have a guideline on which one is preferred. —DSGalaktos (talk) 11:05, 21 July 2015 (UTC)

Property for Elevantion/Height (e.g. mountains) still missing!

One of the most obvious Properties of geo-objects is still missing (but was often discussed last year). Why there is still no elevation Property (wikidata is >1 year old now) ? One artificial reason (from my point of view) for discussion was the possible unit. But even in english wikipedia the height is given in Meters (and feet in brackets). Meter is the international unit (SI = Système international d’unités). Even in US (the only country worldwide not officially using it) big organisations using it as their master system (e.g. Boing, Nasa). If still someone wants to use feet, he could scale it easy by himself. As it is an international project, we should use meters (which is understood and used worldwide). I think its time to add this property soon. The complicated details of different units could even be added later, if someone sees a later need for it. Thanks a lot.  – The preceding unsigned comment was added by Aleks-ger (talk • contribs).

  • I've been working on stages in multi-staged cycle events (like the Tour de France, and many more). And altough we don't miss height, we do miss distance as a key property for a race or a stage. Same story, and same solution. The SI-system is the international standard and should anyhow be the international standard on this project, to my idea. Is there anyone that has an update on this? Edoderoo (talk) 07:22, 20 July 2015 (UTC)
I believe there is a Phabricator ticket for that, but I think @Lea Lacroix (WMDE): will know more about this. Mbch331 (talk) 07:34, 20 July 2015 (UTC)
There was some design work left to do, but no one was available to do that. Since a few weeks, extra people are working on Wikidata again. I thought I've heard September or something as ETA. Sjoerd de Bruin (talk) 08:11, 20 July 2015 (UTC)
  • Several actions were taken to solve the numeric datatype with unit. See here for the whole list of tasks related to this problem.
Main problems are the user interface to select units and the link between the user interface and the items representing the units in order to get the label in the appropriate language and if possible the symbol of the units. If you take the time to read the discussion about that topics, you will see that is not simple to built a tool which can work from the start.
No intermediate solution will be applied because all intermediate solutions imply some work to convert the data once the system is completely implemented and most of the time nobody is available at that moment to do that job.
There is no emergency after all the time we were waiting so no need to implement a temporary solution especially when the dev team is on the track to develop the feature. According to the progress done until now, something should be available in the next months. And there is enough work to be done in WD so don't shout for more things now: everybody wants to have a special feature for his activity in WD. Breath deeply and wait your turn in the queue. Snipre (talk) 09:44, 20 July 2015 (UTC)
  • If you need it for a short term project and you don't mind converting the property later, I'd suggest to use the current quantity datatype. --- Jura 09:48, 20 July 2015 (UTC)
NO, this an error: this will mean value conversion with lost of precision and discrepancies with original data. We have to provide the data like they were given in the sources and no conversion with approximated conversion factor should be done.
WD is not your toy which can be modified according to your own desires. This is a collaborative project so if you want something please discuss it and get a community agreement. Snipre (talk) 09:59, 20 July 2015 (UTC)
I think you are just repeating what you just wrote above. Did you know that the word "collaborative" includes the word "work" --- Jura 10:04, 20 July 2015 (UTC)
We have already discussed this a number of times. The most recent time I can remember, Andy started a thread here. We agreed then, to wait for the developers, and I cannot see that anything has happend here in the short time since that discussion ended. -- Innocent bystander (talk) 11:15, 20 July 2015 (UTC)
The suggested usecases seem a bit easier to tackle and if there are additional persons interested in it, all the better. Besides, we learned that they wont fix the quantity datatype anytime soon (and the unit quantity datatype will have the same issues). Having some data already available that could be used for the new datatype is actually an advantage. Obviously, some always want to wait for perfection. --- Jura 11:26, 20 July 2015 (UTC)
This example ("Height above sea level in meters") definitly have some advantages that such properties like "area in km2" doesn't have. The number of places where "Height above sea level in feet/apples etc" are in use, are very rare. The only exception I can think about is US. If we can avoid that specific area, the transfer to the new property can be done without loss of precision and sources. -- Innocent bystander (talk) 13:20, 20 July 2015 (UTC)
I agreed no such thing. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:17, 20 July 2015 (UTC)
No, but we failed to find a consensus to approve such properties then. -- Innocent bystander (talk) 13:20, 20 July 2015 (UTC)
@Jura1: "and the unit quantity datatype will have the same issues". Please no opinion but facts: give us the reference where the dev team said that they won't fix the problem of the units soon ? They hired a new guy for working on the interface problem begining of summer and one of the tasks of that guy is to work on the unit interface. Read that comment from the dev team and tried to say again that the dev team won't solve that problem soon again. That's fact. Snipre (talk) 15:50, 20 July 2015 (UTC)
I think it says that there, timestamped "13:52, 22 June 2015 (UTC)", but it might refer to something else. --- Jura 16:33, 20 July 2015 (UTC)
@Jura1: By the way give us please the number of obsolete properties which are still in use. All these properties where declared obsolete after a common decision but why are they still there ?
no label (P357), no label (P438), no label (P513) and no label (P387) were created before the multilingual datatype and now are obsolete. Who is taking care about these properties now and is working to cure the items using them ? Is there any project ? I am not looking for the perfection I just want to avoid to repeat a well known scenario. Snipre (talk) 16:16, 20 July 2015 (UTC)
That's actually a good thing you bring this up. These made us realize that monolingual string is still incomplete and needed some post-implementation fine tuning (which hasn't been done). If we had waited for everything to be done before, we still wouldn't have any of these properties. Besides, having both can make people realize that there may be several properties they have to check in a defined order of priority. In any case, we know you wont be collaborating, as you wont be around to do the job ("most of the time nobody is available at that moment to do that job"). --- Jura 16:33, 20 July 2015 (UTC)

@Snipre:, is there an overview somewhere of what this new quantity with unit feature is supposed to look like. There are a number of aspects that might or might not be headed in the right direction. In the case of elevation, one might want to indicate that the tolerance on the quantity (upper and lower bound) are unknown (at least to the editor who added the data) which is a different concept from the quantity being exact. Also in the case of elevation, there are a number of different systems in which one might specify the elevation, such as height with respect to a particular geoid (,, etc.) or the height with respect to an ellipsoid such as WGS84 (which is the system required in Wikipedia location-related templates and several other widely used web standards). Elevations in the various systems can differ by 100 m. And our unresolved confusion with horizontal position suggests we should specify what the upper and lower bound mean; are they the upper and lower bound of the measurement of a particular point of interest, or are they the upper and lower bound of the object being described? Jc3s5h (talk) 13:51, 20 July 2015 (UTC)

@Jc3s5h:, all the discussions about the feature numeric values with unit are available on Phabricator. Please just take the time once to read all the tasks related to this datatype described on this page. You will have an idea about the problem related to the implementation of the units. There are currently some design problems to solve so no beta version exists now. Snipre (talk) 16:00, 20 July 2015 (UTC)

An Internet RFC which seems to be the latest web-centric specification for describing a location is interesting; it differs in many respects from WikiData. It also says " However, as defined above, altitudes are relative to the WGS-84 reference geoid rather than Earth's surface. Hence, an <altitude> value of 0 MUST NOT be mistaken to refer to "ground elevation" . [Boldface added.] Jc3s5h (talk) 14:24, 20 July 2015 (UTC)

We already have properties like determination method (P459) to solve some of those issues.
I live in a part of the world where the surface level of the sea is becoming lower every year. But that does not effect our measurements very often. A nearby lake is located 0 meters above sea level according the maps, but I have never seen it being flooded by salt water. And I guess that hasn't happend for the last 2-300 years. But it affects how people name things here. The lake is named "fjärd", something you normally only name sea waters. And the nearby church is located in what people name "the island", even if it was several hundreds of years since it was surrounded by waters. -- Innocent bystander (talk) 14:59, 20 July 2015 (UTC)
I live in a part of the world where all hieghts "above sea-level" are actually above a en:Ordnance datum agreed in the 1800's when the ordnance survey was carried out. Changes in sealevel will not affect the datum although we may need to update the correction factor to convert AOD hieghts to hieght above sea-level.

Facts

For the people who don't know what the dev team is doing on the topic of the numeric datatype with units, please follow this bug reports:

  • since 5.5.2015 the definition of the user interface is in discussions, see here
  • since 30.6.2015 the interface problem is under construction, see here

So please respect the work of the dev team and before judging their work take a moment to learn what they are doing. The dev team was working the last months on the arbitrary access which has a higher priority and now they put their effort on the units problem. The numeric datatype is quite complex with the units problem and the problem of the uncertainty. Only a global vision can solve all these problems so this needs time and a lot of discussions. Snipre (talk) 16:53, 20 July 2015 (UTC)

@Snipre:, you mention the units problem and uncertainty in the same sentence, and the Phabricator tasks mentioned make no mention of uncertainty. Is the vision to regard the uncertainty as another quantity, with its own unit, or is the vision to regard the statement about a particular property as a nominal value, upper bound, and lower bound, each with its own unit. In the case of locations, treating it as nominal value, upper bound, and lower bound means that for practical purposes, they must all be given in degrees. It eliminates the possibility of giving the nominal value in degrees and the uncertainty in meters, as is the practice of RFC 5870 and the National Geodetic Survey datasheets.
In terms of widespread discussion, no mention of this change has been made at mw:Wikibase/DataModel. Is that because all the changes will be specific to WikiData, and not to any other instance of Wikibase? If so, do we need a document comparable to mw:Wikibase/DataModel which will explain WikiData specific requirements for those who enter data, and allow those who read information retrieved from WikiData to understand it? Jc3s5h (talk) 18:08, 20 July 2015 (UTC)
Uncertainty is far beyond the current problems of the numeric datatype. No model is discussed now so no answer is available. There are some bug reports about that topic. The best is to follow them and take part to the discussions. But be aware that the main discussions is about code implementation so quite technical and very generalized from application point of view.
For uncertainty description we should provide the value , upper and lower bound, if the uncertainty is percentage or absolute value and the statistical method used to describe the uncertainty. Snipre (talk) 10:10, 21 July 2015 (UTC)

In conclusion, we currently still have only the datatype for quantity. The development team continues work on a datatype for quantity with units. A role-out date hasn't been fixed. Not all problems of these datatypes are on the plan to be resolved in the medium term. The current development plan doesn't implement the "ideal" solution for elevation/altitude data. A compromise with the options available now or in the medium term might need to be found. Discussion about one of the other shouldn't be a judgment of the work of any of the participants. Discussion of the possibilities of the current options is encouraged. If there are specific new properties that should be considered, it might be worth creating specific property proposals. --- Jura 14:49, 21 July 2015 (UTC)

new items without statements

The number of items without a statement is going up. see the stats. I have a hard time understanding what the reason may be. It is counter intuitive... Who knows what is going on? Thanks, GerardM (talk) 07:52, 21 July 2015 (UTC)

You mean: https://tools.wmflabs.org/wikidata-todo/stats.php#t2_statements
Someone creating items, but not adding statements? --- Jura 07:58, 21 July 2015 (UTC)
I know a few days ago a bot was running (without botflag) and creating a lot of items with only a sitelink. No labels, no statements. Mbch331 (talk) 08:05, 21 July 2015 (UTC)
Well, it did add labels, just not in English. Anyways, it stopped before having any impact on the chart.
Generally, it's User:‎GZWDer/User:GZWDer (flood) or User:Danrok who rely on other users adding statement. --- Jura 08:13, 21 July 2015 (UTC)
It didn't stop completely voluntarily. I blocked the account because of the lack of botflag. Mbch331 (talk) 08:24, 21 July 2015 (UTC)
So there are still many articles without items... OK then it is ok GerardM (talk) 09:23, 21 July 2015 (UTC)
Personally I don't think it's a good idea to add statements to new items created from pages in a category. There're many false positive. It's better to add them later by adding statements to items from pages in a smaller category (and old items will also be covered).--GZWDer (talk) 09:26, 21 July 2015 (UTC)
The most important statement to new items are the ones that define them for what they are.. Categories are quite powerful for more statements once you know what they are.. Thanks, GerardM (talk) 15:29, 21 July 2015 (UTC)

Creating a Cern Encyclopedia out of existing resources using MediaWiki

Hello all,

We are thinking about creating a Cern encyclopedia which should include all the Cern related people, the accelerators, experiments, accidents, institutes and so on. The data to cover those classes is scattered all around different resources and databases, such as the Cern document server(https://cds.cern.ch/) or the Cern Greybook Database (https://greybook.cern.ch/greybook/), but we would like to get use of Wikipedia and Wikidata as well. What we are aiming for is a central repository where we can store and structure the extracted data from the different resources. Then, we would like to display the data in our encyclopedia, so that there is the fulltext Wikipedia article of the item (if existing) on the top, followed by the statements of the structured data such as in Wikidata. The datasets should be published as Linked Open Data.

To give you an impression of how we think it should look like, here is a link to a quick mockup: http://6jhrh3.axshare.com The are two examples of items (People->Francis Farley and Accelerators->LHC)to look at, be aware that it's a rudimentary mockup with the only purpose to show how the items should be displayed.

We already did a short research and thought about using MediaWiki to build the encyclopedia and Wikibase, if that is a suitable approach for a central repository for our project? Or would there be other alternatives to consider such as Semantic MediaWiki especially for the aspect of Linked Open Data? Is there a way to transclude Wikipedia pages and display them in a third-party Wiki? We are in general pretty new to this topic and we would be really pleased to get some impressions and ideas for a possible approach for our project.

With our best thanks,

Filip Martinu MaFi (talk) 13:45, 22 July 2015 (UTC)

Hi Filip. I'm currently traveling but if you don't get answers I am happy to have a chat in August when I am back. --Lydia Pintscher (WMDE) (talk) 19:28, 16 July 2015 (UTC)
I'm certainly interested in helping with that. --Daniel Mietchen (talk) 20:35, 16 July 2015 (UTC)
I'm trying to set up a Wikidata instance for stuff related to companies (non notable ones) and jobs.
I'd very much like to benefit from the experience of setting up custom instances of Wikidata, along with an import system, a bot system and custom instances of the tools around Wikidata. I've followed the steps and have a somewhat running instance (both for this project and for experimenting for Open Food Facts), but that's a process that should be documented further.
Anyhow, I'd like to be kept in the loop if possible. --Teolemon (talk) 20:06, 17 July 2015 (UTC)
@Lydia Pintscher (WMDE): I'm very much looking forward to this!
@Daniel Mietchen: Thanks a lot, we will talk soon!
@Teolemon: Thanks for replying! What do you mean by you have a somewhat running instance, could you state that more precisely? MaFi (talk) 14:14, 22 July 2015 (UTC)
@Mafi9: I've installed MediaWiki, Wikibase on a server, I've created some job related properties, and I realize setting up an instance of Wikibase is not just about installing it, but setting up a complex ecosystem (bots, mass import, constraints, links to matching wikidata items…) . I'd be interested in shadowing the CERN as they go through the process and that we get most of the untold steps documented (or automated).--Teolemon (talk) 15:40, 22 July 2015 (UTC)

former country nonsense

An "empire" is an instance of a former country?

More such nonsense at: https://www.wikidata.org/w/index.php?title=Special:WhatLinksHere/Q3024240&from=33691&back=0

80.134.89.98 22:51, 19 July 2015 (UTC)

Country can mean a lot of things. An empire is an instance of a 'sovereign state'. Pretty much the definition of a component 'country' of an empire is that they give up some sovereignty and they can't get it back without the agreement of the empire. Hope that helps. Filceolaire (talk) 20:24, 20 July 2015 (UTC)

Another former country nonsense I just spotted only due to a bot adding the reverse property - capital (P36) set to administrative entities created long after the former country stopped to exist [7]. Was imported by a bot, but nevertheless nonsense. Ahoerstemeier (talk) 09:10, 21 July 2015 (UTC)
Just put an enddate to it, and there is no problem. See Q25287#P1376. -- Innocent bystander (talk) 10:36, 22 July 2015 (UTC)

Search oddity (Unicode - emotion)

If I search for the Unicode character "☺" on en.Wikipedia, I am successfully redirected to en:Smiley, from which I can see that smiley (Q11241) has that character as an alias.

However, if I perform the same search in Wikidata, I get no results.

Why not? Is this a bug? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:13, 21 July 2015 (UTC)

Works for me. --Succu (talk) 16:38, 21 July 2015 (UTC)
Same here, no results to "". And aliases should show up in search results? --Stryn (talk) 19:54, 21 July 2015 (UTC)
What I noticed is that when I paste the smiley in the searchbar and do nothing, smiley (Q11241) shows up as a suggestion. If I press enter or click on the search icon, the search results are empty. And yes aliasses should give results (I typed in Bruce Jenner and got Caitlyn Jenner as a suggestion. In all fallback languages for me he/she is labelled as Caitlyn Jenner). Mbch331 (talk) 20:01, 21 July 2015 (UTC)
Yep. But this sounds to me like the return of very old bug. Can not remember where and when I encounted this behavior first. --Succu (talk) 20:20, 21 July 2015 (UTC)

Fusion problems

Can someone fusion en:Category:Deutsche Bank people (Q8375612) and deutsche de:Kategorie:Person (Deutsche Bank AG) (Q9070052) ? 88.71.61.245 21:50, 21 July 2015 (UTC)

Can someone fusion en:Category:German Progress Party politicians and deutsche de:Kategorie:DFP-Mitglied (Q8923441) ? 88.71.61.245 22:16, 21 July 2015 (UTC)

Both ✓ Done. Please could you next time link the QID's? That's a lot easier when merging. (They can be linked the same way as articles). Mbch331 (talk) 07:07, 22 July 2015 (UTC)

Wikidata:Requests for permissions/Oversight/Sjoerddebruin

I've accepted a nomination for oversight access, here is a neutral notification as requested on that page. You can find the request here. Sjoerd de Bruin (talk) 15:50, 22 July 2015 (UTC)

Plz delete duplicate of "National Book Store"

Plz delete this duplicate Q17081424 of this original Q6971094 ("National Book Store"). Thanks. SomeRandomPasserby (talk) 17:56, 22 July 2015 (UTC)

I've merged them, we don't delete (anymore). Sjoerd de Bruin (talk) 18:00, 22 July 2015 (UTC)

News about Freebase migration

I have just push out today a lot of new things related to the Freebase to Wikidata migration project:

  • A new dump have been uploaded in the freebase-testing dataset with 3.4M new statements. It contains 1.4M statements about unemployment rate (P1198) and 0.75M about population (P1082) in the U.S. The data previously in freebase-testing have been move to freebase. I am not sure that all the data in this dataset are good. Feel free to ask me to remove too bad data.
  • A new dump containing 0.8M new ids (Discogs artist ID (P1953), IMDb ID (P345)...) have been updated in the freebase-id dataset.
  • The number of statement in Primary Sources per property list have been updated.
  • I have created A page with the list of all not-mapped Freebase topic with more than 200 incoming links from mapped Freebase topic. It may be interesting to map them in order to be able to improve our mapping of some properties like occupation (P106).
  • I have created an other page to map Freebase types to Wikidata classes in order to create the instance of (P31) statements that may be missing in Wikidata. As the Freebase type system is very different to Wikidata class hierarchy I think that most Freebase types will remain unmapped but this mapping effort may help our classification.
  • I have created a beginning of a “new mode” for Primary Sources, allowing to display list of Primary Sources statement matching a given query. It only supports simple statements currently (i.e. without qualifiers and source) and don’t allow to do mass actions yet (but it is planned). It is accessible in the toolbar using the “Primary Sources list” link.

Tpt (talk) 00:09, 23 July 2015 (UTC)

The freebase-id dataset doesn't seem to be available in the UI.--Mineo (talk) 09:16, 23 July 2015 (UTC)
Sorry, I have by mistake uploaded the ids in the freebase-testing dataset. It is now fixed. Tpt (talk) 18:00, 23 July 2015 (UTC)

Semi-automatically extracting data from templates

A while back, a number of bots would go through some of the Wikipedias and import IMDb ID (P345) from various templates on them, such as {{IMDb title}} on the English Wikipedia. Bots doing this (at least for enwiki) seems to have been idle for quite some time. The task produced a number of inconsistencies that had to be weeded out manually, which is why I guess they were stopped. Anyhow, I am not suggesting restarting them, but is it possible to request doing so for a listed subset of pages? If I wanted the IMDb identifier imported from, say, all the pages in en:Category:Films directed by George Archainbaud, how would I do this if not by hand? Gabbe (talk) 06:52, 23 July 2015 (UTC)

Ask a bot and indicate the way to ship you the data. The best is to work with a bot and to analyze data before the importation. So a discussion about the filters to apply and the automatic corrections to perform should be done before any importation in WD. 141.6.11.24 07:12, 23 July 2015 (UTC)

Properties for local wikipedia needs

Since it is not yet possible to have local repository, is it possible to create properties for local needs of some wikipedia community? Say, if I want to link an item with a topic project in my wikipedia, could I create a property Property:WikiProject in XXwiki? I'll appreciate any other suggestions how to store pages metadata so they are available via MediaWiki API. --AS (talk) 11:05, 17 July 2015 (UTC)

@AS: You can already create an item for a WikiProject here. You can link it to the project page on your Wiki, so if the item about a WikiProject has a page in your language, it covers your usecase. If the project is specific to your wiki, then it will have only this language link, otherwise the properties will also benefits similar wikiprojects on other wikis. But yes, it's possible to ask for properties to describe wikiprojects, I guess, but there is already properties to describe the topics of some items :) We for example have is a list of (P360) View with SQID to describe lists items. author  TomT0m / talkpage 11:13, 17 July 2015 (UTC)
@TomT0m: I mean exactly cases when properties are wiki-specific. I want to store 1) state of an article in my wiki (stub, complete etc.). 2) priority of an article within multiple projects (hign priority for Biology, low for LocalProject). --AS (talk) 12:05, 17 July 2015 (UTC)
@AS: then you don't need properties but badges, you're at the wrong door. They are made for stuffs like article status, like feature and so on. author  TomT0m / talk page 12:10, 17 July 2015 (UTC)
@TomT0m: Ok, I've read about badges. They seem to have too limited usage (only as per-wiki boolean values). I'd like to have per-wiki values, but with power of properties (ability to add multiple values and set qualifiers). As I understand it is not possible with badges to set importance of an item for specific wikiproject. --AS (talk) 18:44, 17 July 2015 (UTC)
Like Ceres (Q596) is considered of low importance for wikiproject Astronomy@xywiki, but high importance for wikiproject Asteroid belt@xywiki. No, that is not what badges were intended for. But your ideas looks interesting. My proposal: Set up a projectpage here at Wikidata, summon interested users and developers and see if you can find a solution! -- Innocent bystander (talk) 06:22, 18 July 2015 (UTC)
„Like Ceres (Q596) is considered of low importance for wikiproject Astronomy@xywiki, but high importance for wikiproject Asteroid belt@xywiki“ — exactly. --AS (talk) 10:07, 18 July 2015 (UTC)
@AS: You have categories on your WP to do that work so why you want to do it on WD ? I think this is waste time to find a solution for a problem which doesn't exist and will bother our time and the time of the dev team for nothing. Snipre (talk) 09:37, 18 July 2015 (UTC)
Categories would be suitable for now (even without planned WD features for lists/reports), but a page info seem to be not available in Modules yet. --AS (talk) 10:07, 18 July 2015 (UTC)
@AS: I don't understnd your position: what you ask already exist on WP. See this talk page about methanol where two projects evaluated differently the priority [8] and here for a summary table (scroll down and look at the right side of the page). You just need to have a bot on your WP to handle that kind of system. No need of WD. Snipre (talk) 13:30, 19 July 2015 (UTC)
No. Statements on wikidata are (generally) about concepts, not about wikipedia articles.
Having said that there are still things that can probably be done. Identifying stubs can certainly be done with badges.
Identifying the importance of a concept is a property of the concept (or the associated wikidata item). Having different importance levels in different projects can be accommodated - a statement can have multiple values and a qualifier can identify which project that applies to. The problem is getting a consensus on importance between different languages. We have meta:Vital Articles - a list of 1000 articles every wikipedia should have - but beyond that I don't know if there is consensus on the rest of the topics in Wikidata. If there existed even one cross-language wikiproject working on a consensus for the importance of articles then I am sure you could get a property approved to support that cross-language project. but without that I don't think it would get approved - see my proposal from last year. Filceolaire (talk) 02:51, 21 July 2015 (UTC)

Ok, I'll wait for support for local repositories. --AS (talk) 20:41, 23 July 2015 (UTC)

Why is Mount Everest not a Property:P610 (highest point) of anything?

See title. --Ysangkok (talk) 17:32, 22 July 2015 (UTC)

Now it is! --Izno (talk) 17:49, 22 July 2015 (UTC)
The highest point (P610) of the Mount Everest (Q513) is Earth (Q2)? That seems wrong to me. Sjoerd de Bruin (talk) 17:50, 22 July 2015 (UTC)
Now it's not! --Izno (talk) 17:53, 22 July 2015 (UTC)
That's an incorrect use of the property, Ysangkok. You are supposed to put Property:P610 on the place item with the highest point, not on the point itself. Right now Earth (Q2) highest point (P610) Mount Everest (Q513). --Izno (talk) 17:53, 22 July 2015 (UTC)
@Izno: But what is "the point itself"? Mount Kilimanjaro (Q7296) has the highest point Kibo (Q1394606). The Q for Africa references this latter Q. If that is how it is supposed to be, then a lot of high points must be made to use the property correctly... Or can I use the mountain itself also? --Ysangkok (talk) 22:24, 22 July 2015 (UTC)
There is of course always doubts if an item should be regarded as a "point" or a larger "area". Most projects have an article about the massif of Kilimanjaro, but none of Kibo. "Kebnekaise southern peak" is the highest point of Sweden (Q34) but it has no item here, only for the whole mountain of Kebnekaise, who also have a second "nothern peak". -- Innocent bystander (talk) 06:11, 23 July 2015 (UTC)
The answer to the initial question is at https://tools.wmflabs.org/reasonator/?q=Q513&lang=en --- Jura 06:15, 23 July 2015 (UTC)
Yes Innocent bystander Some things need to be "instance of:summit (Q207326)" rather than "mountain". Joe Filceolaire (talk) 22:07, 23 July 2015 (UTC)
Apsolut! @Filceolaire You can feel free to add a new item about Kebnekaise southern peak if you like.
BTW here we have a typical case when queries does not work. Yding Skovhøj (Q529888) is here higher than Møllehøj (Q689106). But since Yding Skovhøj is partly man made, it is not considered as the "highest point" in Denmark. -- Innocent bystander (talk) 06:37, 24 July 2015 (UTC)

Žalec

Hello, anyone can please fix interwiki for Q15931? There are two different articles en:Žalec and en:Municipality of Žalec with some inccorect interwiki links. Thank you --Elm (talk) 04:33, 23 July 2015 (UTC)

@Elm: I would move this to WD:IWC. Could you please give an example which links are wrong? Matěj Suchánek (talk) 18:45, 23 July 2015 (UTC)
That would be probably better, thanks. Sure, Q631387: pl Gmina Žalec should be just pl Žalec and Q15931 which is a instance of municipality has got some interwiki like fr Žalec which is about the city and not about the municipality. --Elm (talk) 19:13, 23 July 2015 (UTC)

Proposal to create PNG thumbnails of static GIF images

The thumbnail of this gif is of really bad quality.
How a PNG thumb of this GIF would look like

There is a proposal at the Commons Village Pump requesting feedback about the thumbnails of static GIF images: It states that static GIF files should have their thumbnails created in PNG. The advantages of PNG over GIF would be visible especially with GIF images using an alpha channel. (compare the thumbnails on the side)

This change would affect all wikis, so if you support/oppose or want to give general feedback/concerns, please post them to the proposal page. Thank you. --McZusatz (talk) & MediaWiki message delivery (talk) 05:08, 24 July 2015 (UTC)

Automated help with adding a statement

I'd like your opinion on an idea. I have proposed a new property for the entries in the Catalogus Professorum Halensis (Q20680681). I'd like to link the biographical entries with Wikidata items. I already have a list with all the identifiers (row A) and the names of the persons (row B), see here.

I wanted to check the names from row B against Wikidata, but this will take a lot of time with 1105 entries. Hence I thought we could try to put a script to the task. I don't speak any programming languages, but I admire the concept ;) But seriously, this is what I do. Maybe some of you have an idea how to turn it into a script an run it.

Or if that's not a good idea, I'd like your opinion on what to do instead.

Here's what I want to do:

  1. Open http://www.catalogus-professorum-halensis.de/[VALUE ROW 1,COLUMN A].html
  2. Open https://www.wikidata.org/w/index.php?search=[VALUE ROW 1,COLUMN B]&go=Go
    1. IF number of results = 1, open and check:
      1. IF instance of (P31) human (Q5) is true, add qualifier invalid ID (P20xx) with statement "[VALUE ROW 1,COLUMN A]".
    2. IF number of results > 1, open all and check:
    3. IF the string "Halle" is somewhere on the page, add qualifier invalid ID (P20xx) with statement "[VALUE ROW 1,COLUMN A]".
  3. Close all tabs, repeat:
  4. Open http://www.catalogus-professorum-halensis.de/[VALUE ROW 2,COLUMN A].html
  5. Open https://www.wikidata.org/w/index.php?search=[VALUE ROW 2,COLUMN B]&go=Go
    1. IF number of results = 1, open and check:
      1. IF instance of (P31) human (Q5) is true, add qualifier invalid ID (P20xx) with statement "[VALUE ROW 2,COLUMN A]".
    2. IF number of results > 1, open all and check:
      1. IF the string "Halle" is somewhere on the page, add qualifier invalid ID (P20xx) with statement "[VALUE ROW 2,COLUMN A]".
  6. Repeat 1103 times.

invalid ID (P20xx) is just a placeholder for the property, which I proposed here: [9].

What do you think? Jonathan Groß (talk) 18:00, 19 July 2015 (UTC)

to me, this looks like a job for Mix'Match, by @Magnus Manske: :) --Hsarrazin (talk) 22:51, 19 July 2015 (UTC)
Ah, I see. But how does this work? Jonathan Groß (talk) 13:01, 20 July 2015 (UTC)
you should ask @Magnus Manske: - he will explain you, I can't, I only play mix'n'match afterwards ;) --Hsarrazin (talk) 20:16, 20 July 2015 (UTC)

Added here. Try "game mode" if you do this for the first time :-) Note that there won't be any change to Wikidata unless we have a property for this, or I do some one-off hack once all entries are matched. --Magnus Manske (talk) 21:38, 20 July 2015 (UTC)

Great tool :) I'll play around with it. Just one question @Magnus Manske: What's the difference between "NoWD" and "N/A"? Jonathan Groß (talk) 09:43, 22 July 2015 (UTC)
Also, the identifier "Arnoneidhartkarl" needs to be changed to "neidhartkarl". Jonathan Groß (talk) 09:45, 22 July 2015 (UTC)
The tooltip over "NoWD" should be changed to "Mark this entry as not (yet) available on Wikidata", because as of now the tooltip for "NoWD" and "N/A" is the same. Jonathan Groß (talk) 11:46, 22 July 2015 (UTC)
That seems too long. Sjoerd de Bruin (talk) 21:07, 22 July 2015 (UTC)
In any case, the tooltip for "NoWD" and "N/A" should not be the same. That's confusing. Jonathan Groß (talk) 11:03, 23 July 2015 (UTC)
The identifier "Richardmatrinybenno" should be changed into "matrinybenno". Jonathan Groß (talk) 20:17, 22 July 2015 (UTC)
The identifier "Wernernagelarno" should be changed to "nagelarno". Jonathan Groß (talk) 21:05, 22 July 2015 (UTC)

Matching's done, big thanks to Magnus Manske for this great tool! It would be great if some of you voted on the property as well. Jonathan Groß (talk) 09:48, 24 July 2015 (UTC)

If somebody ....

... could write this page > [10] Thanks and have a good day all ! Mike Coppolano (talk) 07:45, 21 July 2015 (UTC)

@Mike Coppolano: What specifically are your requesting?--Jasper Deng (talk) 07:46, 21 July 2015 (UTC)
@Mike Coppolano: ✓ Done Et le résultat dans Reasonator : http://tools.wmflabs.org/reasonator/?q=Q2550070&lang=fr J'ai juste rajouté quelques déclarations, pas de sources. C'est pas encore possible de donner les dimensions. Next time, you'll do this yourself :) It seems jura1 did also help :) author  TomT0m / talk page 08:13, 21 July 2015 (UTC)
Thank you all. Best regards. Mike Coppolano (talk) 08:17, 24 July 2015 (UTC)

Two or one statements?

Does Q3471364#P31 make sense? Or should I merge the two statements into one single, with four qualifiers? -- Innocent bystander (talk) 08:01, 23 July 2015 (UTC)

Makes sense to me. I think a merge will make it more confusing. Deryck Chan (talk) 09:21, 23 July 2015 (UTC)
A merge doesn't seem correct to me. Not sure how other systems will handle that. Sjoerd de Bruin (talk) 09:27, 23 July 2015 (UTC)
When I made this, I thought of if somebody would ask: "what was 'Q3471364' 'an instance of' in '2003'". With a merge, I am not sure what the result would be. Now it's obvious that it wasn't a 'Q14839548'. -- Innocent bystander (talk) 10:14, 23 July 2015 (UTC)
Much better as two statements each with start and finish. It does leave the question as to what happened before 1995 and between 2001 and 2005 Maybe add some "part of" statements (I assume it was part of something else at those times?) or maybe a "replaced" statement. Joe Filceolaire (talk) 22:01, 23 July 2015 (UTC)
@Filceolaire: These kinds of entities are statistical. I am not sure "part of" makes any sense here, except for when they are amalgamated by the neighbour entity, but that was not the reason here. The most likely reason here was that the population was below 50 persons 1990 and 2000. When they no longer fulfill the requests Statistics Sweden demands of them, they cease to exist as a statistical entity. They of course survives as an human settlement (Q486972), when the population drops below 50 persons. Therefor, it could be a good idea to add P31:Q486972 to these entities. -- Innocent bystander (talk) 08:41, 24 July 2015 (UTC)

Label lister

Regularly I find problematic labels in "other" languages, languages that I do not want to always see. Label lister is a gadget for this. It no longer works. It is quite an important tool and I find it sad that it has been broken for so long.. Given that there is no alternative for its function... what can be done, what is planned to remedy the current status quo? Thanks, GerardM (talk) 08:53, 23 July 2015 (UTC)

What doesn't work with the tool? I just tested Labellister (release version) and have no problems with it. I must say I prefer the Beta version, which works fine for me as well. Mbch331 (talk) 11:56, 23 July 2015 (UTC)
It's working here in Safari, Firefox and Chrome. Sjoerd de Bruin (talk) 12:09, 23 July 2015 (UTC)
I've actually had some problems upon trying to save changes in the beta version, but in the stable release-version, I only very, very rarely have a problem, which usually goes as soon as I've refreshed/purged the item. Jared Preston (talk) 08:36, 24 July 2015 (UTC)
Could you describe those problems? We're trying to fix the bugs left in the beta version and put that one live. Sjoerd de Bruin (talk) 16:49, 24 July 2015 (UTC)
@GerardM: could you please describe this problem? Thanks. Sjoerd de Bruin (talk) 16:49, 24 July 2015 (UTC)

Creating a profile

Google knowledge graph is drawing data from wikidata profiles.

So how do I create a profile, please?

I have set up the preferences page, but where to from there?

thank you.  – The preceding unsigned comment was added by Geoff Duffell (talk • contribs).

What makes you notable for a item on Wikidata? Just because you want a Google Knowledge graph isn't enough. Mbch331 (talk) 09:16, 24 July 2015 (UTC)
See our Freebase FAQ. - Nikki (talk) 09:38, 24 July 2015 (UTC)

P442

This property is clearly an external identifier. It does not link to an external website but does that really matter ? Could we please make it that codes like this one show as an external identifier and not as a regular statement ? Thanks, GerardM (talk) 16:40, 24 July 2015 (UTC)

Your comment would be easier to read if you would wikify "show" and "China administrative division code (P442)". --- Jura 18:29, 24 July 2015 (UTC)
There is no difference between external identifiers and string values except that the gadget AuthorityControl is adding a hyperlink if an external website is available. --Pasleim (talk) 21:02, 24 July 2015 (UTC)
I was wondering if with "show" he didn't mean Reasonator. --- Jura 05:20, 25 July 2015 (UTC)

Somebody reading Arabic, please ?

On the arwiki, Aline Lahoud (Q258700) is indicated as deceased on 23-7-2015, which seems quite suspicious - in fact, it seems to be her aunt, whose name is the same.

If someone could correct it on arwiki, would be nice, because she appears on Wikidata:Database_reports/Recently_deceased_at_Wikipedia and if someone transfers the info, it will probably be a mess — Thanks :) --Hsarrazin (talk) 20:09, 24 July 2015 (UTC)

done --Pasleim (talk) 20:52, 24 July 2015 (UTC)
Thanks @Pasleim: :) --Hsarrazin (talk) 21:36, 24 July 2015 (UTC)

Coaches

The people, not the buses.

See for example José Mourinho. It seems wrong to me to say that he was "member of sports team" when he was coaching them (All of them, except the first three ones). Wouldn't it be better to have a property "coach of" or something like that ?

Pleclown (talk) 12:05, 24 July 2015 (UTC)

Being a coach is a job right? employer (P108) could be used, with position held (P39) as qualifier. Sjoerd de Bruin (talk) 12:23, 24 July 2015 (UTC)
We do have a "coach of" property, but it is used on the item for the entity coached, not the person item of the coach him/herself. See head coach (P286). Jane023 (talk) 13:05, 24 July 2015 (UTC)
@Jane023: I know the P286 property, but, as you say, it's used on the entity coached, and thus not reported on the person.
@Sjoerddebruin: This could work for a professionnal coach. But if the person is a volunteer, should we say he is "employed" ?
Pleclown (talk) 14:33, 24 July 2015 (UTC)
Yes employed works, even though one may be employed as a coach at zero salary, that person is still responsible on behalf of the organisation (the team or the team's owner). Jane023 (talk) 15:50, 24 July 2015 (UTC)
Pleclown; Or <position held (P39):coach (Q41583) (of (P642):'team')> Joe Filceolaire (talk) 08:46, 25 July 2015 (UTC)
@Filceolaire:, coach (Q41583) is a profession (Q28640) not a position (Q4164871), which would lead to a constraint violation by using it as a value for position held (P39). To avoid that, occupation should remain coach (Q41583) (or better yet the appropriate subclass, like association football manager (Q628099)). Then, position held (P39) could be a qualifier used together with employer (P108), but only with values such as head coach (Q3246315), assistant coach (Q11703711) or caretaker manager (Q1050607), which are hierarchical positions in the coaching team (as in Tito Vilanova (Q222924)). The only disadvantage would be that employer (P108) only allows three qualifiers (so replaces (P1365) and replaced by (P1366) are out), whereas position held (P39) does allow them. So, another option would be to use position held (P39):head coach (Q3246315) with the qualifiers of (P642):'team', replaces (P1365) and replaced by (P1366). Andreasm háblame / just talk to me 21:10, 25 July 2015 (UTC)
@Andreasmperu: Yes. <position:head coach (of:team)(start time:?/?/20??)(end time:?/?/20??)> is better. It doesn't need to be a qualifier to the <occupation:coach> statement - these two statements can both exist side by side. Joe Filceolaire (talk) 21:47, 25 July 2015 (UTC)
The coach of the England national mens' football team (for example) is, presumably, employed by the Football Association, not the team. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:49, 25 July 2015 (UTC)

Is there a place to save Wiktionary links?

In Wikidata, is there a place to save links to Wiktionary pages? I know there are place to save links to Commons, Wikisources, Wikinews. But how about Wiktionary? (I want to know this because in EnWP anatomy project, an interesting idea proposed. The idea that Latin or Greek anatomy terms linking to Wiktionary pages[11].) --Was a bee (talk) 06:09, 25 July 2015 (UTC)

Not right now, no. Since Wiktionary is based around words rather than concepts, it's more complicated to do. The most recent proposal is at Wikidata:Wiktionary/Development/Proposals/2015-05. - Nikki (talk) 06:24, 25 July 2015 (UTC)
Thank you @Nikki:. I understand, difference between words and concepts making that difficult. Very helpful advise. Thank you very much. --Was a bee (talk) 07:43, 25 July 2015 (UTC)

User page

When to Wikimedian's User page link to Wikidata. --Nakare✝ (talk) 09:39, 25 July 2015 (UTC)

What is the use of that? What knowledge can be learned from a userpage? Mbch331 (talk) 09:53, 25 July 2015 (UTC)
If you need interwiki, see Wikidata:Wiktionary/Development/Proposals/2015-05#Task 1: Wiktionary interwiki links. -- Innocent bystander (talk) 10:11, 25 July 2015 (UTC)
@นคเรศ:If the Wikimedian is notable, and has a Wikidata item, use website account on (P553) with a value of (say) Wikipedia (Q52) and a website username (P554) qualifier of the user name. See, for example, Jimmy Wales (Q181). However, note that there is currently a proposal to replace this method with a specific property. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:43, 25 July 2015 (UTC)
Andy Mbch331 Adding user pages to wikidata seems (to me) to be the quickest and easiest way to create universal user pages that are internationalised and can easily be localised - basically by replacing most of the user templates with wikidata statements so the user page becomes a special Reasonator page and user talk pages become a Flow page. (yes this will need some new properties so people can list different types of achievements as well plus a property for some free text.) Joe Filceolaire (talk) 22:02, 25 July 2015 (UTC)
This is not a good idea: we are mixing too many things inside WD. At some point we will have to split the database into general data and wikipedia data. If WD becomes only a substructure of WP, we will be less attractive for external users of WD. Please at least create a new space for that kind of data. User data is not common knowledge. Snipre (talk) 22:45, 25 July 2015 (UTC)
All user pages? I think a lot of people would not accept generated user pages if it restricts what they can do with their page. It is already possible to have a global user page by making a user page on metawiki. I think it would be pretty cool if templates could be stored in a central place (e.g. also on metawiki) and reused by any project (in the same way that images on Commons can be used by any project without needing a local upload). I think the translate extension would make it possible to translate the templates. Something like that could make it possible to have a global (or local user page if you want) with translated templates that don't need to be duplicated across every project. I won't get my hopes up though, there are still thousands of user language templates still being used despite the introduction of the Babel extension years ago. Even Wikidata has some... - Nikki (talk) 22:57, 25 July 2015 (UTC)

Interwikis preferential treatment for English

I would need to know why when creating a new article in the spanish Wikipedia, which still does not exist in the English version, for example this, the link to the non existent english version appears in red. With this approach they would not have to appear in red non existent versions in all languages? (obviously illogical on the other side). Greetings Antur (talk) 02:05, 26 July 2015 (UTC)

@Antur: Go to your settings and see if "Traducción de contenidos" is activated. -- Innocent bystander (talk) 05:39, 26 July 2015 (UTC)

ːːYes, it is. And this is that causes te red label?. --Antur (talk) 06:38, 26 July 2015 (UTC) ːːːThanks for your answer. It remains unclear for me the reason for this. Why not a red label for Tagalog?. --Antur (talk) 06:45, 26 July 2015 (UTC)

Your interface language? --- Jura 06:51, 26 July 2015 (UTC)

Linking from Wikia to Wikidata

How do I link from Wikia projects? The d:Q### does not work, but w:Foo works for Wikipedia. --Richard Arthur Norton (1958- ) (talk) 03:15, 26 July 2015 (UTC)

Your Wikia project of interest has to add it to their mediawikiwiki:Interwiki map. Alternatively, you might request that the company as a whole add it to their global interwiki map, which I believe is in use on Wikia. --Izno (talk) 04:32, 26 July 2015 (UTC)

Bot operators needed

We seem to have quite a backlog at Wikidata:Bot requests. How an we recruit more bot operators?

Can we make a point of highlighting these requests, at hackathons? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:38, 25 July 2015 (UTC)

Reducing requests

Propositions to reduce the number of pending requests:

  1. Create a template for request with all data needed for the bot action. People have to provide the details from the beginning in order to avoid too long discussions between the bot operators and the request person.
  2. All requests should be linked to a discussion on the wikiprojects pages or the project chat in order to have clear consensus about the need of request. If people take more time to discuss about their ideas, perhaps can we merge several requests into one.
  3. Too few persons know autolist: we should redirect more people to this tool when the request is simple. The template cited above should allow a clear identification of requests which can be performed by autolist. Snipre (talk) 22:59, 25 July 2015 (UTC)
[I've numbered your suggestions for ease of reference] Regarding #2, surely Botreq is the place where such consensus is determined? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 15:03, 26 July 2015 (UTC)
Well, my bot can handle "ISBN conversions" or "Twitter and Instagram property migration". If one admin guy could create new accounts with bot flags for me for those approved tasks I can handle those. But my last experience of RfP/b is more than enough for me, don't want to visit this page ever again. -- Vlsergey (talk) 15:43, 26 July 2015 (UTC)

Wikidata weekly summary #168

Feature request

I just created a new feature request on Phabricator calling for a new section to be added to the "statements". This new section would show all the statements on other items, which use the current item as their object.

I also called for a new function in the wikibase-client to let the client access statements which use the current items as their object as well as the current functions which let the client access statements which have the current item as their subject.

This would be like having an automatic "inverse property" for every property with "item" datatype.

Anyone who thinks this is a good idea can log in to phabricator and support this support or follow this task. Joe Filceolaire (talk) 21:53, 23 July 2015 (UTC)

@Filceolaire: It's a good idea especially if it is also doable with a lua call :) It's what simple query was meant for, I don't know where the devteam is on this. author  TomT0m / talk page 10:35, 28 July 2015 (UTC)

Number of images per item

Hi all, I thought the consensus was one image only in the image field of an item, but I don't know why. Pls see this discussion here: Talk:Q2550070#Number_of_images. Does this have to do with consumer tech (like accessing the image field from an infobox?) or is this just Wikidata convention, or is there something more behind it? Thx, Jane023 (talk) 07:53, 25 July 2015 (UTC)

Without the possibility to add a caption/description to the file, I see very little value of P18 at all. -- Innocent bystander (talk) 08:14, 25 July 2015 (UTC)
I didn't expect that answer at all - what do you mean? I often add qualifiers to the P18 statement, such as "painting by so and so" or even collection where it came from etc. Jane023 (talk) 08:26, 25 July 2015 (UTC)
Then I have missed your statements. All P18-statements I have seen, have been without such descriptions. Also the linked item missed such descriptions.
My personal opinion is that such things as pictures should be choosen by those who edits the article. Should you choose a picture of somebody dressed for football, when the articles infobox mainly is about the subjects career as a politician for example? You can add pictures of Arnold Schwarzenegger (Q2685) both as a Guvernor of California, and as an Actor and add descriptions that would make it possible for the software behind the infoboxes to choose the correct file, but it looks like a to farfetched solution to me. -- Innocent bystander (talk) 08:55, 25 July 2015 (UTC)
OK I think I see where you are going with this. The P18 is useful for Wikidatans who don't know who those people are and a quick check of the image helps. Also it is a quick shortcut to commons where the image can be checked for global use. I would definitely not advocate getting rid of P18, but I am just confused about whether there was any consensus about using it once. I know there is not constraint, but maybe there should be? Jane023 (talk) 10:16, 25 July 2015 (UTC)
I am not proposing that we should "get rid of P18" either, but rather that it should be used with some thoughts, just like it sounds you already do. -- Innocent bystander (talk) 15:21, 25 July 2015 (UTC)
The caption is the description field on Wikimedia Commons. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:46, 25 July 2015 (UTC)
I think the idea is that you have one image and you have one Commons category. That makes it easy to make infoboxes in 200 different languages with minimal human intervention but you still have the option to include other images if one language want to be more hands on. Joe Filceolaire (talk) 08:51, 25 July 2015 (UTC)
This was my understanding, but is it mentioned anywhere? Jane023 (talk) 10:52, 25 July 2015 (UTC)
I suppose this will be different depending on the topic, but in most cases selection of an image here at Wikidata may well be crucial. If the writer of an article knows better, he can override this choice. A second image makes sense only if there is a property "second image". - Brya (talk) 09:44, 25 July 2015 (UTC)
@Brya, Pigsonthewing, Filceolaire, Jane023: I think we're other engeniering rules and trying to solve a nonexisting problem. There is nothing crutial here. The first image second image stuff is solvable using qualifiers. It's to data clients (Reasonator, Infoboxes, domain specific clients ...) to choose what they do with the datas. Reasonator handles pretty well several images for example. If a topic has only three of four images, like picture scan of oldbooks, we can be exhaustive on images with the statement, and I don't see how this could harm anything. I think we just should rely on common sense here, unless someone find a real problem. author  TomT0m / talk page 10:42, 28 July 2015 (UTC)
The whole purpose of the project is to be selective. Adding qualifiers may work, but it may not work. - Brya (talk) 10:49, 28 July 2015 (UTC)
@Brya: I don't get you, the whole purpose of the project is to build a repository of structured datas. The question is "why would we have to set rules in this case". Beeing selective to be selective does not lead us anywhere. author  TomT0m / talk page 13:35, 28 July 2015 (UTC)
Sounds like a good idea for an RfC. An ideal outcome seems to me like a differentiation by topics. We do not need )and do not want) to add all Commons images here, but for some subjects one image is clearly enough (and who chooses it? who can replace it?), and for others like buildings we may want to have several depicting different aspects.--Ymblanter (talk) 11:26, 25 July 2015 (UTC)
Yes probably. Jane023 (talk) 13:22, 25 July 2015 (UTC)
I think there is no consensus to have just one image at P18, but there is clear consensus to not use P18 as alternative to Commons gallery or Commons category. In some cases even six images may be appropriate (Třeboň Altarpiece (Q2428392) for example). At the moment for external use there is use only for one picture. --Jklamo (talk) 11:35, 25 July 2015 (UTC)
I disagree. In this specific case, the image should be the entire altarpiece, and it should be an instance of painting series and link with " has part" to each sub item that has the appropriate image. So total of 7 items, each with its own image. Jane023 (talk) 13:22, 25 July 2015 (UTC)
If there are several images, usually a qualifier should be added. For people, this would generally be point in time (P585). Generally, I suppose 1 picture per every couple of years should do. For 3D objects, it could make sense to have images from different perspectives. For 2D objects, 1 image should do. --- Jura 12:06, 25 July 2015 (UTC)
To strict rules could harm the usefulness the this property, I would think. But demanding useful qualifiers when more than one file is used could be a good rule of thumb. -- Innocent bystander (talk) 15:21, 25 July 2015 (UTC)

This discussion copied to Property_talk:P18#Number_of_images_per_item Suggest we continue the discussion there. Joe Filceolaire (talk) 22:10, 25 July 2015 (UTC)

It's rather confusing. One could think that people actually commented there. Besides, now we have twice the same text. Why don't you want to continue this discussion here? --- Jura 04:14, 26 July 2015 (UTC)
I've deleted the copy of this discussion from that page. If anything, a pointer to this discussion is what should be posted there. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 14:52, 28 July 2015 (UTC)

Wikidata discourages interwiki links

Looking at Q1074987 Zhuanhgzi, I see that this data item is only for the work as it exists in Chinese. The work as translated into English is a separate data item, and the French Wikisource translation is another data item.

Effectively, this means that Wikidata discourages interwiki links to and between Wikisource projects, because they will never be part of the same data item. Further, anyone seeking a translation of a work into another language must first come to Wikidata and surf the links even to find out if translations of a work in another language exist on Wikisource; it is not possible to do that from any Wikisource directly.

I thought the whole point of moving the links to Wikidata was to promote connections between projects, not to eliminate them. But perhaps I am wrong. --EncycloPetey (talk) 02:19, 26 July 2015 (UTC)

@EncycloPetey: There are other ways to create interwiki by the help of Wikidata. See s:sv:Bibeln 1917 where I have made some tests with the Bible (Q1845). The interwiki is created by the help of a Lua-module that follows edition(s) (P747) and edition or translation of (P629). The big advantage is that it makes possible to create intewiki to more than one page in every project. For example, that page have 13 links to enwikisource. -- Innocent bystander (talk) 05:49, 26 July 2015 (UTC)
@Innocent bystander:
This is very interesting, what template (and Lua-module) do you use ? it should be done for all wikisources, that often have a lot of translated texts :) --Hsarrazin (talk) 09:43, 26 July 2015 (UTC)
@Hsarrazin: It's s:sv:Modul:Sandlåda/Innocent bystander 3 who today is included in s:sv:Mall:Titel, a template that can be found in almost every page on svsource. Observe that the module is not secured against loops in the P747/P629-hierarcy. It also needs support by the "interwiki-extra"-class in s:sv:MediaWiki:Common.js, otherwise you cannot have more than one interwiki in each page. -- Innocent bystander (talk) 10:11, 26 July 2015 (UTC)
BTW, another thing the code does: It makes it possible to have interwiki to the Text-namespaces in als/bar/frr/pflwikis. -- Innocent bystander (talk) 10:18, 26 July 2015 (UTC)
@Hsarrazin: You'll also be interested into WD:XLINK, a project on how to solve and list such issues and techniques. author  TomT0m / talk page 10:30, 28 July 2015 (UTC)

Quality of Freebase and quality of Wikidata

It is news that yet again a lot of new data has been uploaded in the "primary sources tool". Tpt then suggests that he does not know about the quality of the Freebase data. To decide if this is of concern, it has to be seen in context. The context is Wikidata itself. I seriously doubt that Wikida quality is better where there is data. When there is no data, One-eye is king in the land of the blind anyway.

When data is available in the "primary sources tool", it is important to accomplish a few things.

  • match the data to existing Wikidata data
  • find what data fits the Wikidata structure and make it fit in the way we do things in Wikidata
  • compare the data and find the differences and keep what is good
  • add all missing data and attribute properly (this is currently NOT done)

When we compared data and decide on what is correct, we should be able to signal this. For all the combined data in Wikidata, we should look for corroboration by checking other sources that have data on the same subject.

Thanks, GerardM (talk) 06:55, 26 July 2015 (UTC)
For coordinates (Wikidata:WikiProject Freebase/Coordinates), Wikidata and Freebase have approximatively the same proportions of errors. --Nouill (talk) 13:29, 26 July 2015 (UTC)
Exactly my point... Will the missing data in Freebase be imported as requested? Thanks GerardM (talk) 09:09, 27 July 2015 (UTC)
I have been mainly working in the past weeks on getting more data from Freebase. I agree that it would be nice to add good data automatically using a bot. Feel free to create a page listing "good" datasets currently in Primary Sources and to start a community discussion about adding them using a bot. If there is a consensus in favor of it I'm ready to write and run a bot to do such importation (in fact as Primary Sources have a nice API, anybody could do it). I plan also to extract quality annotations form Freebase (i.e. mostly facts that have been reviewed/not) in order to be able to highlight reviewed facts in Primary Sources and allow a bot to add them automatically if the community agrees. Tpt (talk) 20:17, 27 July 2015 (UTC)

Honorary academic degrees

Does this make sense or is it preferred that honorary academic degrees be listed with award received (P166)? Jonathan Groß (talk) 09:51, 27 July 2015 (UTC)

Many people are already noted by making it an award.. Please keep it that way. Thanks, GerardM (talk) 11:17, 27 July 2015 (UTC)
@GerardM: What do you mean, "keep it that way"? I'm not going to take away anybody's award. Jonathan Groß (talk) 07:14, 28 July 2015 (UTC)
It is an either or situation.. I do not think both are good. GerardM (talk) 07:54, 28 July 2015 (UTC)
I still don't understand what you're saying. I was asking if an honorary academic degree should be included using academic degree (P512) or rather award received (P166). Are you telling me I shouldn't use both? Because that is self-evident. Jonathan Groß (talk) 17:09, 28 July 2015 (UTC)
I would prefer it to be "award received" ... In the end it can be converted from one to the other ... Thanks, GerardM (talk) 17:39, 28 July 2015 (UTC)
Okay, got it. Thanks! I think it's still a matter of opinion though ... Jonathan Groß (talk) 17:45, 28 July 2015 (UTC)

Wikidata Pseudo-Bacon Numbers

@GerardM: (and anyone else interested) I have made a live version of the program for calculating WDPBNs, which can be downloaded here. Enjoy! Popcorndude (talk) 23:56, 27 July 2015 (UTC)

It is code and I do not know how to use it.. Can you package it in a way that makes it usable to me? Thanks, GerardM (talk) 07:58, 28 July 2015 (UTC)
Sorry, I should have been more clear, it's an HTML file, so you just have to down load it and open it with a webbrowser. Popcorndude (talk) 13:46, 28 July 2015 (UTC)

Batch imports: don't truncate the label at the comma if what is after the comma is important

It appears various bots/users in the last few months have done a whole lot of imports from Wikisource. One project in particular I've noticed is thousands of biographical articles from the s:en:Dictionary_of_National_Biography,_1885-1900. Whenever these were imported, something (the tool used? some bit of code?) stripped anything after the first comma when creating the label, which means that now there are 125 new Wikidata items named simply "Jones" and 94 "Hamiltons" and 84 "Campbells"... etc.

Since the item is really for individual biographical article entries called things like "Jones, Christopher" and "Jones, David" and "Jones, Harry David," the information after the comma is quite important and should remain in the label as part of the work's title. But I don't know of a way to fix these thousands of items now, without manually typing the first names back in. Does anyone have any ideas?

I recognize that the information after a comma in a given title is often extraneous and not appropriate to include in a Wikidata label, but in cases like this it's vital to include. Certainly this feature (of stopping labels at the comma) can be tweaked when doing imports of this nature, yes? Could those of you doing these huge batch imports be more conscious of the type of data you're importing and extract better labels, so as to avoid significant work later? Sweet kate (talk) 18:46, 28 July 2015 (UTC)

Pywikibot

Hello. Can anyone help me with pywikibot? I can't log in! Xaris333 (talk) 17:43, 28 July 2015 (UTC)

Hi Xaris333 I'm sorry to hear that. Best way to get help is probably to go to #pywikibot on Freenode (webchat). Can you post the output of pwb.py login.py and pwb.py version.py? Multichill (talk) 18:19, 28 July 2015 (UTC)
Multichill with pwb.py login.py the output is: ImportError: No module named requests Python module request is required. Try running 'pip install request'. Xaris333 (talk) 18:47, 28 July 2015 (UTC)
Xaris333, recent versions of pywikibot use the library Requests instead of httplib2. You need to have that installed. You can read how to do that here. Multichill (talk) 18:56, 28 July 2015 (UTC)
FYI, I created phab:T107193. Multichill (talk) 19:04, 28 July 2015 (UTC)

Fusion problems

Can someone fusion en:Category:Erwin Schrödinger (Q15813277) with Germen de:Kategorie:Erwin Schrödinger (Q8949213) ? Mouztre (talk) 20:48, 28 July 2015 (UTC)

✓ Done. Jared Preston (talk) 20:59, 28 July 2015 (UTC)
The problem now is that we don't quite know whether the category state is merged or unmerged. :-D --Teolemon (talk) 22:40, 28 July 2015 (UTC)

Freebase migration: what should we do with unemployment rate (P1198) and population (P1082)

I have been able to map 1.4 million statements about unemployment rate (P1198) and 0.75M about population (P1082), mainly for US cities and upload them into Primary Sources. For unemployment rate (P1198) there is often the rate for each month since something like 20 years. See, for example, San Francisco (Q62). Do we want these data into Wikidata? If not, I can remove them (or hide them) in the Primary Sources tool. Tpt (talk) 22:03, 27 July 2015 (UTC)

Assuming we actually want to include such data, the question is if we want to import it from the primary source (the US Census, and go through mapping it) or from Freebase (and use the mapping done by FB/Primary Sources tool/Tpt).
I'd import them as long as Freebase's source indication is imported as well (looking at Q62, this seems ok). --- Jura 07:57, 28 July 2015 (UTC)
I'm a bit afraid we'll run into a scaling problem here (items getting too large). Maybe just put these on hold for a while and first focus on other sources? That way we can easily come back later to it. Multichill (talk) 18:23, 28 July 2015 (UTC)
I have hidden these statements (their state is now "skipped"). They are still in the Primary Sources backend but are not displayed anymore. Tpt (talk) 16:04, 29 July 2015 (UTC)

Wikidata linking

Hello! Could this be technically possible with some script or bot? The problem in short: we (Latvian Wikipedia) have quite much templates, categories and other non-article stuff, that isn't connected to Wikidata. As adding one by one is too slow and too boring, I was thinking of some kind of automatisation. Could I write such file:

"title in Latvian Wikipedia" (some kind of separator) "title in English Wikipedia"
"title in Latvian Wikipedia" (some kind of separator) "title in English Wikipedia"
...
"title in Latvian Wikipedia" (some kind of separator) "title in English Wikipedia"

And give to some bot or some script and it does the job (linking to Wikidata). OK, if needed I can make sure, that title in English Wikipedia really exists and is not a redirect and also title in Latvian Wikipedia isn't already connected to Wikidata. Or... We (lvwiki) put iw link in pages itself (in the old pre-Wikidata style: [[en:title]]) and some bot does the job. Yes, for me/us it would be faster (I think with AWB it could be done automatically). --Edgars2007 (talk) 11:03, 28 July 2015 (UTC)

In Ladino Wikipedia, I have many templates like this, and some are really obvious matches (Babel templates, for example). I'd love some help with this. StevenJ81 (talk) 12:28, 28 July 2015 (UTC)
To both, this is pretty routine. You should each make a (separate) request at WD:Bot requests. --Izno (talk) 16:21, 28 July 2015 (UTC)
@Izno: Thank you. But there's no format over there. Do I just say, I have a bunch of entries of such-and-such a type; please connect them for me? StevenJ81 (talk) 20:37, 28 July 2015 (UTC)
Yup! You might want to start a subpage to capture the links you want to make. You may also want to state other requests you might have at the bot request page. --Izno (talk) 05:49, 29 July 2015 (UTC)
@Edgars2007, StevenJ81, Izno: You may use this tool. The syntax is (e.g.):
en-title TAB Slvwiki TAB lv-title
...

Note it only adds sitelinks to existing items. It does neither create any new item nor merge any existing item. So you should create items for all en pages, and merge existing lv item to en items (probably via "Item merging" function).--GZWDer (talk) 06:55, 29 July 2015 (UTC)

@GZWDer: Thank you! Will try to play with the wikidata-todo tool :) --Edgars2007 (talk) 09:04, 29 July 2015 (UTC)
Thanks to all. I'm going to try to make a request on the bot request page first. I have a ton of such links, but a lot of them meet a pattern. So if I can polish them off with a brief description over there first, I'd like to go that way. I'm sure I'll have to go back and to others by means of the tool, and that's ok, too. I appreciate the help from everyone. StevenJ81 (talk) 13:22, 29 July 2015 (UTC)

Page title -> Q

Hope here are some Lua guys, too. Problem basically is such. Is it possible (probably with Lua) to get Wikidata item (Q...) if I pass as a parameter page title? I have seen the vice versa solution. Yes, I'm passing other page title, not the one, in which page I am. Expensive functions are OK. I just need to convert these <10'000 wikilinks to Wikidata items to work with them. If somebody has some other solution to problem, that I described (about there links), then please - you're welcome. --Edgars2007 (talk) 10:15, 29 July 2015 (UTC)

Special:ItemByTitle? Wiki code + title → Q#. —DSGalaktos (talk) 10:27, 29 July 2015 (UTC)
That is too manual work. What I wanted, is something like this:
{{some template/module|Page title1}}
{{some template/module|Page title2}}
{{some template/module|Page title3}} etc.
And I get Q# for each of them. --Edgars2007 (talk) 10:50, 29 July 2015 (UTC)
{{Item}}DSGalaktos (talk) 10:58, 29 July 2015 (UTC)
It was so close... I already started to be sooo happy. But no. I can't get Q# directly with it. Copy {{Item|Really Don't Care|enwiki}} into Special:ExpandTemplates. I would need in the output Q# (just the number or part of the line, it doesn't matter), that I could work with them. --Edgars2007 (talk) 11:12, 29 July 2015 (UTC)
Currently you need "some javascript guys", since you today need to look into the API to get this information. -- Innocent bystander (talk) 11:41, 29 July 2015 (UTC)
Yes, probably that would be better. Noticed, that quick statements does the job (converting Wikipedia links to Q#), maybe Magnus Manske (talkcontribslogs) can give some input? Such solution, that is in quick statements (for converting) is totally OK :) --Edgars2007 (talk) 12:02, 29 July 2015 (UTC)
Slightly work-aroundish, but will do the trick: Generate wikitext with wiki links to the pages you want, then paste that here. One wiki at a time only. If you already have these links on a wiki page on that wiki, just use the page name in the form on the top. --Magnus Manske (talk) 12:40, 29 July 2015 (UTC)
Has somebody already said you today, that you are a cool guy? If not, then I will be the first one today :) So fast and cool. Thank you! --Edgars2007 (talk) 13:07, 29 July 2015 (UTC)
@Edgars2007: You may want phab:T74815.--GZWDer (talk) 13:44, 29 July 2015 (UTC)
Thanks for the note, added to subscibers. --Edgars2007 (talk) 13:52, 29 July 2015 (UTC)

Fusion problems

Can someone merge en:Category:Fridtjof Nansen (Q8474831) with German de:Kategorie:Fridtjof Nansen (Q8959621) ? 92.76.111.8 10:50, 29 July 2015 (UTC)

✓ Done - Mbch331 (talk) 11:04, 29 July 2015 (UTC)

no label (Q20018918)

Could someone help me figure out what this is? I suspect it needs to be merged with one of the other oceans, but I don't know which one. Popcorndude (talk) 13:36, 29 July 2015 (UTC)

Ask the creator. ;) --Succu (talk) 13:43, 29 July 2015 (UTC)
My best guess is something like Panthalassa (Q208378), but unless the original contributor speaks up, I'm not sure we'll know for sure. (Check at Wikidata:Bistro, since s/he seems to have been a Francophone.) StevenJ81 (talk) 13:44, 29 July 2015 (UTC)
Hi, it's not an ocean, it's a type of ocean. So it can't be merged with an ocean. It's the class of all earth oceans that do not exists anymore. Indeed Panthalassa (Q208378) is an instance of it, as historic superocean (Q20018921) is a subclass of no label (Q20018918). author  TomT0m / talk page 13:54, 29 July 2015 (UTC)

Module:Linguistic

Why I see content model of Module:Linguistic as wikitext instead of Scribunto. Paweł Ziemian (talk) 20:29, 29 July 2015 (UTC)

many templates display "Lua error in package.lua at line 80: module 'Module:Arguments' not found." instead of normal behaviour. I guess the dev' team is working on it :/ --Hsarrazin (talk) 20:47, 29 July 2015 (UTC)
Wow! It is already fixed. Why it was broken? Paweł Ziemian (talk) 20:53, 29 July 2015 (UTC)
Stuff broke and was mostly fixed, see phab:T91170 for details. Multichill (talk) 21:07, 29 July 2015 (UTC)

Fusion problems

Can someone fusion en:Category:Peter Paul Rubens(Q19953949) with German de:Kategorie:Peter Paul Rubens (Q9074053) ? Miolertz (talk) 21:11, 29 July 2015 (UTC)

✓ Done by Silverfish Pamputt (talk) 21:57, 29 July 2015 (UTC)

Website user names

I'm unclear as to why we record things like Twitter user names as a sub-property of website account on (P553), rather than giving them a dedicated property. This, for instance, stops us using formatter URL (P1630). What's the reason for this? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:12, 17 July 2015 (UTC)

I don't know what the original reasons were, but in my opinion (as I expressed before) we should have on property for every (relevant) website… that's just easier to work with and it is consistent with other identifier properties. - Hoo man (talk) 14:38, 17 July 2015 (UTC)
Thanks for bringing this up! I agree with Hoo that it should be split. The current property has quite a few issues, it's harder to use (both to enter the data and to use the data), we can't have formatter URLs or useful constraints and it's not really clear what counts as a username or even what counts as a website account (I recently saw someone add a link to a Wikipedia article by using "Wikipedia" as the website and the article name as the "username"...). I've spent some time recently looking into how the property is currently used since I'd been planning to propose splitting it up and well over 80% of the existing usages are for Twitter, so I think that would be a good place to start. - Nikki (talk) 15:51, 17 July 2015 (UTC)
Symbol support vote.svg Support, as long we have someone to do the job. Would make it possible to use better constraints. Sjoerd de Bruin (talk) 16:12, 17 July 2015 (UTC)
Symbol support vote.svg Support If we indeed split up and start with the biggest ones (I would say the big social media sites: Twitter, Facebook, Instagram), we can indeed try on making constraints and formatter URL's. Mbch331 (talk) 16:14, 17 July 2015 (UTC)
Symbol support vote.svg Support GerardM (talk) 05:45, 18 July 2015 (UTC)
Since multiple people support this and nobody has anything against it so far, I went ahead and proposed a few new properties: Twitter username, Instagram username and Wikimedia username. If everything goes well with those, I'll write some more proposals for other sites (I would need more time to look into Facebook, Google+ and YouTube anyway because they have multiple identifers/names/URL formats that might be better represented as multiple properties). - Nikki (talk) 12:35, 18 July 2015 (UTC)
Thank you. I'd add Flickr, LinkedIn & GitHub to the mix. I suggest we need to keep website account on (P553), in the long term, for new sites and edge cases. Does anyone have a list of the sites currently included, and their numbers? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:38, 18 July 2015 (UTC)
I did a SPARQL query which you can see here, here, or here (hopefully one will be working :)). I don't know how old the data is though. - Nikki (talk) 21:53, 18 July 2015 (UTC)
Thank you. I dumped the data from one of them at User:Pigsonthewing/websites. As might be expected, there;s a long tail. We have a mix of WMF accounts (Wikipedia, Wikidata, Wikimedia Incubator, Wikiversity, Wikisource, Wikispecies, Finnish Wikipedia, etc.) totalling at least 82. We also have duplicates due to typos or different spellings/ capitalisation (Daum/ Daum Communications; Diaspora/ diaspora* (the former a mistake?); WordPress.com/ WordPress; and, inexplicably, Google scholar/ Google Scholar). Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 00:19, 19 July 2015 (UTC)
"the former a mistake?" →‎ No. The current english Label of diaspora* (Q1973097) is "diaspora*" (sic). Visite fortuitement prolongée (talk) 19:18, 22 July 2015 (UTC)
But "former" refers to "Diaspora" not "diaspora*". Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:39, 23 July 2015 (UTC)
Pictogram voting question.svg Question Since website account on (P553) allow only items, how could it be duplicates and mispellings??? Visite fortuitement prolongée (talk) 19:18, 22 July 2015 (UTC)
Added proposals for Facebook iD and YouTube user name - the second- and third- most used values. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 18:25, 22 July 2015 (UTC)

Twitter username (P2002) and Instagram username (P2003) have been created. I've requested a bot to migrate the data. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:38, 23 July 2015 (UTC)

Great work! I was actually recently at a hackathon and someone requested exactly this! ·addshore· talk to me! 12:50, 23 July 2015 (UTC)
On Property talk:P553 it says the Russian Wikisource is using website account on (P553) and we should let them know before making any big changes. Are there any Russian speakers around who could do that? - Nikki (talk) 18:16, 23 July 2015 (UTC)
@Ymblanter, Ivan A. Krestinin, Vlsergey: ^ Matěj Suchánek (talk) 18:37, 23 July 2015 (UTC)
No, not me. The only russian-speaking project I edit is Wikivoyage.--Ymblanter (talk) 19:43, 23 July 2015 (UTC)

Also creataed: Facebook profile ID (P2013). Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:15, 30 July 2015 (UTC)

(on behalf of ruwiki) Waiting for bot work to be done before updating JS and LUA / templates. -- Vlsergey (talk) 12:51, 30 July 2015 (UTC)

Adding property colour(s) for flowers

Is it possible to add a property colour(s) for flowers? Timboliu (talk) 14:58, 26 July 2015 (UTC)

@Timboliu: Can you give some examples, please? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 15:00, 26 July 2015 (UTC)
For instance en:Ranunculus. I would like to add the colour yellow to this flower, so it can show up when somebody is looking for yellow flowers. Timboliu (talk) 15:33, 26 July 2015 (UTC)
You mean like this? Mbch331 (talk) 16:26, 26 July 2015 (UTC)
Not good; because Ranunculus flowers are yellow, but the plants are green. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 23:36, 26 July 2015 (UTC)
@Timboliu: Thank you. In that case, we'd need a proposal for a new property, at Wikidata:Property proposal/Natural science. Ping me if you want to proceed, and need help or advice. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 23:36, 26 July 2015 (UTC)
I remember this discussion was held before, and the conclusion was, that colour was not a generic qualification for flowers or plants in general. So why asking again for the same thing? Edoderoo (talk) 20:38, 26 July 2015 (UTC)
Possibly because the OP doesn't have knowledge of everything we've ever discussed; and possibly because that's a dumb decision that needs to be rethought. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 23:36, 26 July 2015 (UTC)
This is going to get complicated very quickly. To begin with, describing the flower colour of Ranunculus as "yellow" is inaccurate. - Brya (talk) 18:25, 27 July 2015 (UTC)
Since when did we let "complicated" stop us ;-) ? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 22:31, 28 July 2015 (UTC)
Yes, but look at the messes this attitude has gotten us into! - Brya (talk) 10:42, 30 July 2015 (UTC)

Proposal posted; see Wikidata:Property proposal/Natural science#flower colour. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 22:54, 29 July 2015 (UTC)

Sister city mess

It seems we are having a big mess of bogus data for sister city (P190). I stumbled on that because someone added sister relation between La Paz (Q1491) and Bangkok (Q1861) - and I am sure Bangkok has no such relation. Apparently a lot of the entries for La Paz are bogus, not found on any Wikipedia nor have any reference, and (coincidentally?) added April 1. Checking the other relations added, maybe half of the relations of Milan (Q490) are not referenced nor found in English Wikipedia. The half-automatic add of the inverse link with Widar makes the problem worse, spreading the bogus data even more. Ahoerstemeier (talk) 10:24, 29 July 2015 (UTC)

Fully agree! I have in many cases looked into the homepages of the city and failed to find any source to the majority of the statements. In many other cases, the linking have been done between a city and a populated place without administration. And without a local political administration, nobody can sign a sister-city agreement. -- Innocent bystander (talk) 11:38, 29 July 2015 (UTC)
Yamaha5 recently ran a script to resolve thousands of inverse property constraint violations by simply adding the inverse statement to the corresponding item. Unfortunately, that also ended up introducing or repeating a large amount of low-quality or false data where the original statement was wrong or the properties weren't really inverse (example), which is apparently what has happened here. Väsk (talk) 17:44, 30 July 2015 (UTC)

Phase II support for Wikinews and Wikibooks

When we can deploy both? --Liuxinyu970226 (talk) 10:44, 29 July 2015 (UTC)

I am back in the office next week. I'll look at finding a date then. If I didn't get around to announcing the dates by the end of next week please poke me again. --Lydia Pintscher (WMDE) (talk) 18:54, 30 July 2015 (UTC)

Quickstatements

In Wikimania I heard that there is a "QuickStatements generator". Any address? -geraki talk 10:49, 29 July 2015 (UTC)

@geraki: Hm, I haven't seen it either. Do you know at least where it should generate data from? Matěj Suchánek (talk) 10:52, 30 July 2015 (UTC)

Translating WikiProject pages

On translators' noticeboard TomT0m suggested that more WikiProject pages (eg. WD:WikiProject Taxonomy) should be translatable. I don't oppose it but my practise is only to mark for translation pages of those projects where it was requested (on the noticeboard or by markup) by their participants/founders. So I'm asking whether we should try to have as many WikiProject pages translatable as possible or whether this possibility should only be limited to requests. Matěj Suchánek (talk) 10:48, 30 July 2015 (UTC)

I think all Projects main pages could benefit of a translation, since some have been originally written in non-English language (Names, for ex.), and not everybody can read English. It's really not attractive to find a Main Page about a subject that could interest you, and... you can't understand it easily… :/ --Hsarrazin (talk) 15:25, 30 July 2015 (UTC)

Pywikibot script

Hello. I need help for a script. I use pwb.py claimit -namespace:0 -lang:el -family:wikipedia -file:volley.txt P710 Q12069029 -exists:ptq to add property P710 the item Q12069029 to a list of arcicles that were written in a txt file named volley. It works. I use -exist:ptq so if the property already exist in the wikidata page, I can to put another item in the same property. But, if the item is already in the property, it adding again and then it shows twice. Any ideas? Xaris333 (talk) 15:57, 30 July 2015 (UTC)

You can search in Phabricator to see if it's known problem, or create a task. Matěj Suchánek (talk) 17:58, 30 July 2015 (UTC)

Adding statements - "Save" greyed out

I want to add some statements, but the "save" is greyed out. Why is this? All the best: Rich Farmbrough17:42, 30 July 2015 (UTC).

To what item? What statements? Matěj Suchánek (talk) 17:51, 30 July 2015 (UTC)
Could be some monolingual properties. Sjoerd de Bruin (talk) 17:57, 30 July 2015 (UTC)
Spouse and Notable Works to Q933253. All the best: Rich Farmbrough17:58, 30 July 2015 (UTC).
OK I understand it needs to be an object. All the best: Rich Farmbrough18:01, 30 July 2015 (UTC).

MoMA artwork id (P2014)

I have applied the "ignore all rules" principle and speedily created this straightforward property, in order that work on adding data from MOMaA's recent release of metadata using User:Magnus Manske's "Mix'n'match" tool can begin immediately and we can capitalise on publicity surrounding this development. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 14:06, 30 July 2015 (UTC)

whohooo... more fun on mix'n'match soon :D --Hsarrazin (talk) 15:55, 30 July 2015 (UTC)
It's already there! Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 18:57, 30 July 2015 (UTC)

Those of you with social media accounts may like to retweet this; or to make a similar post in your own words. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 18:58, 30 July 2015 (UTC)

Andy, we're not on Wikipedia and Wikidata:Property proposal/Authority control clearly states that you should wait at least a week before creating a property. With this action you deny people the ability the comment on this. I strongly oppose the creation of this property and I strongly appose the process you took. This property is completely redundant to inventory number (P217) which is unique in the collection and described at URL (P973). And come on Magnus, this is not productive at all we use that in several places. Please restore the described at URL (P973). Multichill (talk) 20:01, 30 July 2015 (UTC)
I don't know what I've done to upset you, but that's the fourth thing in recent weeks you've criticised me for; yesterday you falsely used me of spamming. Such hectoring is becoming tiresome, and demotivating. You now, bizarrely, object to the replacement of described at URL (P973) with a new, more specific and far from redundant property, even though the former is only "to be used to provide links to external resources that are not the item's official website, when no relevant "authority control" property exists (for instance, because the website is too messy)". And the item in your example, Woman on a High Stool (Q8030744), already has a inventory number (P217) (which "represents no label (Q4672511)) from MOMA, in a different series - the MOMA inventory (= accession) number is 506.1964; whereas the MoMA artwork id is 7986 and is not an accession number. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:49, 30 July 2015 (UTC)
Multichill, do you have any actual point against this property? As far as I can tell, this is how we do things on Wikidata - create specific properties for specific external catalogs. If we decided to do this another way, please point me to that decision. And described at URL (P973) is a we-don't-have-anything-more-specific hack, as clearly stated on Property talk:P973, and is now redundant for MoMA artworks. If this screwed up one of your bots, I'm sorry; I had similar things happen before. But Wikidata is not there for your bot, your bot is there for Wikidata. As to Pigsonthewing's "speedy creation" - yes, that was against the rules. Again, the rules are there for Wikidata, not vice versa. MoMA is just like all the other catalogs (some of them significantly smaller) that have been approved in the past; even more, it is a poster child for openness, putting its catalog on github, which makes the argument for creation all the stronger. If you have any actual arguments against MoMA artwork id (P2014), please let us know. --Magnus Manske (talk) 22:48, 30 July 2015 (UTC)
I Symbol support vote.svg Support the creation of this property. Joe Filceolaire (talk) 01:42, 31 July 2015 (UTC)
  • Symbol support vote.svg Support b/c of 125,000 number which includes *everything* they have in depot and of course also to reward them for the CC0 (let's hope many follow). And for the record, I also support the use of described at url for the item's web address at the list holder, iff that info is significantly greater than the usual pic+dimensions+provenance. Jane023 (talk) 07:22, 31 July 2015 (UTC)
  • Pictogram voting comment.svg Comment Devil's advocate: if we create a specific property for ID numbers of the MoMA, does that not mean that we should do the same for all other art collections - for the sake of consistency and ease of querying? I neither approve nor object to that, BTW. I do think we should strive to a system that is 1) consistent, 2) logically correct, and when these criteria are met: 3) as simple as possible. Spinster (talk) 09:02, 31 July 2015 (UTC)
    This is how we have handled things so far. The technical alternatives would be switching to bare URLs in described at URL (P973) (which would be a mess IMO), or something like inventory number (P217) with qualifiers (example: The Starry Night (Q45585)). Such qualifier constructs would be harder to create, query, auto-link, etc. but possible. The only downside I can see to the current process is the slow creation buerocracy, which thankfully was short-circuited in this rather obvious case. --Magnus Manske (talk) 09:19, 31 July 2015 (UTC)

Is "writer" the only gender-related occupation (P106) ?

OK, I'm the first one to ask for the feminisation of worknames (in French and other gender-determined languages), but on Giuliana Fiorentino Tedeschi (Q18811970) I found woman of letters (Q381353) instead of writer (Q36180) as occupation (P106). There are [106%3A381353 lots of other cases].

As a result, men and women who do the same job, "write", do not share the same statement, which seems really ridiculous.

AFAIK, this is the only occupation (P106) which is gender-related. Is this really necessary ?

So far, on fr, the Infobox with wd values, takes the gender of the person into account to automatically display the right form, so why use different items ?

I thought that on wd, the occupation (P106) were not genre-related ? --Hsarrazin (talk) 15:21, 30 July 2015 (UTC)

Do you mean gender? Sjoerd de Bruin (talk) 15:36, 30 July 2015 (UTC)
yes, sorry, corrected. I translated from French, and just had a blank ;) --Hsarrazin (talk) 15:52, 30 July 2015 (UTC)

Hi, we still need to solve the problem "Stephen King is a writer". "Is a" is ambiguous in natural language. If writer is a profession, then it's a type of activity who is made for money. BUT we say "is a writer" like we would say "is a dog" or "is a human", in which case we refer to a class of animal or person. When we use "letter woman" in french, this refers clearly to the class of all women who do write for a living (or a close activity), and this should be expressed by instance of (P31). So in items it should be clear if they are about a type of activity or a class of persons who practices some activity ... author  TomT0m / talk page 17:50, 30 July 2015 (UTC)

err... @TomT0m:, this mention was used as occupation (P106), not instance of (P31) in the items I found it :/ - AFAIK, the only instance of (P31) that should be used there is human (Q5) :) --Hsarrazin (talk) 22:20, 30 July 2015 (UTC)
woman of letters (Q381353) exists because there are wikipedia articles in french and italian about this topic. My feeling is that this should not be used with occupation (P106) unless there is a very specific reason why this is appropriate to a particular writer - and I mean more of a reason than that they happen to be female. There is a general problem in some languages in that the occupation names are gendered. In English 'actor' is now acceptable for both sexes but until a short time ago women were always referred to as 'actresses' and in some languages this distinction is still important. There has been some discussion as to how to manage this but no resolution as yet. Joe Filceolaire (talk) 01:39, 31 July 2015 (UTC)
@Filceolaire: I totally agree with you — this item refers to a very specific period in time, where women were not considered to be able to do the same things as men.
the problem about the gendered occupation names is generally solved (in French at least), by using the male/female form, or by putting the female form as alias. woman of letters (Q381353) is the first one I find that was not solved that way. Hence my question here... --Hsarrazin (talk) 09:22, 31 July 2015 (UTC)
Also Swedish have this problem. But the "female" and the "male" versions of some occupation (P106) is not always exactly the same occupation. "Sjuksköterska" (nurse) is a female word but the occupation can have both males and females. Therefor, also male "sjuksköterskor" uses the female word. The male version of this word: "Sjukskötare" can also have both males and females. But the occupation and education is here not the same. The female nurse is an academic degree, while the male is mainly pre-academic. The female have the right to deliver medicine, the male version have not. The male works mainly in psychiatry (Q7867), the female version can be in all kinds of medicine, including psychiatry.
That's funny, it's exactly the same in France, where "infirmières" (women) work in all kind of medical places, and "infirmiers" (male) were mainly, until recently, working in psychiatry ;)
There are even more occupations in Swedish that can be translated to "nurse" in English. Most of them are female, even if the occupation can have also males. The differences are what they work with and their education.
There is also historical differences in some cases: "Lärarinna" (teacher) was female, and used to be another type of occupation than male "Lärare". They also used to have different educations. -- Innocent bystander (talk) 04:27, 31 July 2015 (UTC)
I guess it's the same with all traditional female occupations and jobs that were forbiden to women, and it's probably the case in all languages that use gendered forms or declinaisons — English is probably an exception on this point ;) --Hsarrazin (talk) 09:22, 31 July 2015 (UTC)
@Hsarrazin:
  • I'm explaining the ambiguity, and this part of my message is independant from Wikidata conventions, so please don't reject the explanation because of that other part of that message :)
Second I said many times I was did not agree with that convention :) instance of (P31) is about classification of items, and to classify humans then ... putting them all in the "human" class is not much of a classification :). The letter woman concept is a beautiful illustration of that. author  TomT0m / talk page 07:58, 31 July 2015 (UTC)
@TomT0m:
I understand the ambiguity you mean ; I'm not sure it exists in all languages, though… I never used is a to attribute instance of (P31) because of this ambiguity, because is a can mean so many things (including is an idiot, which can't be a classification). It's much too subjective, and items should have only 1 instance of (P31) : I use nature de (instance of), which is more objective.
instance of (P31) -> human (Q5) : After all, all humans are equal, aren't they ?  ;)
well, all human (Q5) have very distinctive characteristics (other than obious physical ones) that make them the same class : they have a gender, a birth/death date, a birth/deathplace, they do things, whether this occupation allowes them to earn money or not is another matter, they live somewhere, they have one ore more names, a language, opinions, family ; they exist (in real life, not in fiction), they have a bio, and they are probably the only instance of (P31) on which almost noone has difficulty to tell : yes X is a human (Q5), once the definition is given.
I would certainly not say the same about geographical entities, works, events and other top-classes where it is so difficult to give a instance of (P31). ;)
As for occupation (P106), I disagree with you, it is not only, nor should be, a work to earn your living, it's also main or marginal occupation for which people are known.
  • If only painters who earned their living with painting could be indicated here, Picasso would be alright, but not Modigliani, nor Van Gogh, who almost never sold any painting.
  • For writers, Courteline would not be counted as one (he was civil servant), but blockbuster authors would be ?
  • In politicians, big cities mayors or deputies would be considered professionals (full-time work and high indemnities), but not small villages mayors, that continue to be medicine or teacher ?
  • In sport, professional sportpersons would be categorized there, but not amateurs ? - remember, not so long ago, Olympic Games were only for amateurs ? where would you put all Olympic champions in history ? Or rugby players before they became professionals ?
and for people from early centuries, how would you know if it was their job ? that would be completely illogical ! --Hsarrazin (talk) 09:22, 31 July 2015 (UTC)
@Hsarrazin: It's just a matter of property definition. And that's precisely where classification can be very very useful. For example, if "letter woman" refers to a specific period of time, then we can define the "letter woman", almost literally, as "woman who authored letter artwork in the XIXth century". This is a class defined by intention, and it gives a lot of freedom to define classes. Where properties are to be well defined, this is an opportunity to regain a lot of freedom. Bonus : it has a name in standard knowledge representation languages in which we can translate such expressions : OWL name these as "class expressions". Where if you have a "job" property, you can define a class expression such as "a literature human is someone who is an editor, or a writer, or ...". We can define classes wrt. other properties of the item, that's what makes them interesting. author  TomT0m / talk page 09:33, 31 July 2015 (UTC)
@TomT0m:
I have the feeling you are trying to reproduce with classes the mess that is Commons categories :/
I mean, the principle of wikidata is to structure data in concept that are reducable to 1 characteristic (i.e. : gender, time, place, etc.) : this would be mixing concepts, just like "Canadian hockeists" or "Swedish writers from the 16th century" do in Commons... and the result is... a mess :/ --Hsarrazin (talk) 09:47, 31 July 2015 (UTC)
@Hsarrazin: I don't think it's a realistic feeling : Wikidata has way more ways to structure informations. First, instance of (P31) and subclass of (P279) is are already two ways to represent the kind of relationships that was handled only by categorisation before, which by itself solves a part of the mess. Second, there will be ... queries. And the way they will work on Wikidata is something like "one item per query", which is not different than "one item per class". So if you have a query "Swedish writers from the 16th century", and I don't see a good reason why there will not be, then ... how is it less of a mess ? As I like to say, the problem is not the mess itself as reality is a mess, it's how we handle the mess. And Wikidata is way more stuffed to handle the mess. I like to see the query <-> class definition (expression) analogy and imagine ways to handle the mess. If I know that someone is a "Swedish writers from the 16th century" instance, but I don't really have the informations to make precise statement, then I could set the
< the mysterious item > instance of (P31) View with SQID < Swedish writer from the 16th century >
. The fact that it would not appear in the corresponding query could be a hint that there is missing informations in the item ... And the class definitions would be explicit, and not only contained in the name of the category. A little more advanced stuff I'd like to feature request in the future is to have in an item page the set of queries he's a result of, which could serve as implicit instance of (P31) statements. author  TomT0m / talk page 12:51, 31 July 2015 (UTC)
It's even more complicated in Slavic languages. So problem should not be solved on level o properties. --EugeneZelenko (talk) 13:54, 31 July 2015 (UTC)

Property to indicate something NE is mentioned on a certain page?

Hi, Is there a Property to indicate something that will probably be rejected as an article on most WP is mentioned on a certain page? I have a chapel on Openstreetmap, I created a WD item for. It is mentioned on the fr WP page of Florenville, but I'm pretty sure the chapel will be considered NE, as I don't know more to tell about it. Still I'd like to link it somehow to that page. --Polyglot (talk) 20:34, 30 July 2015 (UTC)

Just add the item and provide as much details in statements.. Do not worry too much about if there ever will be an article... Maybe there will be.  – The preceding unsigned comment was added by GerardM (talk • contribs) at 20:39, 30 July 2015‎ (UTC).
That's indeed what I'd like, not to have to worry about creating articles on WP anymore for items I want to add to WD, because I want to be able to reference them on Openstreetmap. I had hoped there would be a way to show that I did find it mentioned on WP though.--Polyglot (talk) 20:46, 30 July 2015 (UTC)
If the information is on OSM then why the need to add it to wikidata also? What does this add? Wikidata has it's own WD:Notability rules. If the chapel doesn't meet those then it is unlikely that there is more info about that chapel than can be recorded in an OSM entry. Joe Filceolaire (<span class="signature-talk"talk) 01:30, 31 July 2015 (UTC)
There is a street named after the chapel on OSM. I wanted to add a tag name:etymology:wikidata to that street. To do that the chapel needs a wikidata tag.--Polyglot (talk) 07:33, 31 July 2015 (UTC)
@Polyglot: Every place is notable in Wikidata if a 3rd party source contains any non-trival information about the subject.--GZWDer (talk) 05:23, 31 July 2015 (UTC)
For WD yes, for WP not necessarily. I had too many articles, even about mayors that have streets named after them, being flagged for NE. The fact that there is a street named after a person should, in itself, mean that the person is notable. But if one can't say more than some genealogical info, it's not enough for some WPs. So I'm giving up on the battle to add them anymore. Still if the subject is mentioned on a WP page, it would be nice to be able to say so in the WD entry. --Polyglot (talk) 07:33, 31 July 2015 (UTC)
If Wikipedia just mentions Florenville has several chapels, A, B, C, Q20741330, E, F ... Then there is no point in mentioning this on Wikidata. BTW, what is NE? --- Jura 08:34, 31 July 2015 (UTC)
@Jura1:Not 'encyclopedic', this may be something that's only used on nl WP. I'm working on too many language versions...--Polyglot (talk) 08:41, 31 July 2015 (UTC)
The whole notion of notability is so artificial. It is dead easy to find any excuse why something may find its place here. Encyclopedic is for Wikipedia and this is NOT Wikipedia. GerardM (talk) 11:04, 31 July 2015 (UTC)

100 parliaments as open data

An interesting initiative: 100 parliaments as open data, ready for you to use "aims to collect, store and share information about every parliament in the world, past and present—and it already contains 100 of them". It's a pity they are not storing that directly in Wikidata, but it will still be useful.

It uses the Popolo standard, a set of "international open government data specifications", and we can probably use that to determine any relevant parameters that we are missing. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:42, 31 July 2015 (UTC)

What does a Healthy Community look like to you?

Community Health Cover art News portal.png

Hi,
The Community Engagement department at the Wikimedia Foundation has launched a new learning campaign. The WMF wants to record community impressions about what makes a healthy online community. Share your views and/or create a drawing and take a chance to win a Wikimania 2016 scholarship! Join the WMF as we begin a conversation about Community Health. Contribute a drawing or answer the questions on the campaign's page.

Why get involved?

The world is changing. The way we relate to knowledge is transforming. As the next billion people come online, the Wikimedia movement is working to bring more users on the wiki projects. The way we interact and collaborate online are key to building sustainable projects. How accessible are Wikimedia projects to newcomers today? Are we helping each other learn?
Share your views on this matter that affects us all!
We invite everyone to take part in this learning campaign. Wikimedia Foundation will distribute one Wikimania Scholarship 2016 among those participants who are eligible.

More information


Happy editing!

MediaWiki message delivery (talk) 23:43, 31 July 2015 (UTC)

query wikidata for some category

I know I can get an entity as json by querying the wikidata by this url: https://www.wikidata.org/wiki/Special:EntityData/Qxxxxxxx.json but can I get more than one entity in one query/url, for example getting all the entities of american actors or french presidents best reagrds, -- – The preceding unsigned comment was added by 165.50.248.106 (talk • contribs) at 21:02, 13 July 2015 (UTC). // 21:02, 13 July 2015 (UTC)

In good time

well, I need, estem, do a project, which has several advantages, I'm here, because I want to be on wikipedia, so help with all kinds of articles in Spanish in English.  – The preceding unsigned comment was added by Mazdacited Mikedoffer (talk • contribs) at 19:53, 15 July 2015 (UTC). // 19:53, 15 July 2015 (UTC)