Wikidata

Wikidata:Project chat/Archive/2015/03

< Wikidata:Project chat‎ | Archive‎ | 2015
This page is an archive. Please do not modify it. Use the current page, even to continue an old discussion.

Wikibooks now gets language links via Wikidata

Hey folks :)

Last night we added Wikibooks to Wikidata. Wikibooks' language links can now be maintained on Wikidata. Please give them a warm welcome and keep an eye on Wikidata:Wikibooks for questions if you can.

Cheers --Lydia Pintscher (WMDE) (talk) 07:57, 25 February 2015 (UTC)

200000000th edit

See Special:Diff/200000000. @ValterVB:--GZWDer (talk) 00:54, 1 March 2015 (UTC)

Wow... Wagino 20100516 (talk) 03:33, 1 March 2015 (UTC)
  --ValterVB (talk) 07:52, 1 March 2015 (UTC)
\o/ --Lydia Pintscher (WMDE) (talk) 08:37, 1 March 2015 (UTC)
woot! Husky (talk) 10:52, 1 March 2015 (UTC)

Wikidata weekly summary #147

 
Here's your quick overview of what has been happening around Wikidata over the last week.


  • Development
    • Chaged a number of Lua modules (from deprecated function mw.wikibase.getEntity to mw.wikibase.getEntityObject)
    • Updated the json documentation
    • Always link to Wikidata on client pages that don’t have any langlinks. This affects users without JavaScript and logged out users, logged in users will still see the link item dialog. gerrit:168632
    • Fixes for the Wikibase qunit Jenkins job
    • Made Vagrant git-update also properly update Wikibase and dependencies
    • Fix for phabricator:T88254 (malformatted Wikidata entries appearing in Watchlist RSS feed in clients)
    • Final touches on new header design
    • Investigated how we can provide language fallback also in suggestions when search or adding new statements
    • More work on allowing additional languages in monolingual text datatype
    • Added missing backend piece for quantities with units. Now the remaining piece is the user interface.
    • Enabled Wikibase Client on Wikibooks
  • Monthly Tasks
    • Hack on one of these.
    • Help fix these items which have been flagged using Wikidata - The Game.
    • Help develop the next summary here!
    • Contribute to a Showcase item
    • Help translate or proofread pages in your own language!

Merge problems

Can somone merge Category:Israeli football chairmen and investors (Q8557014)

with Kategorie:Fußballfunktionär (Israel) (Q16861973) 47.64.246.241 15:11, 1 March 2015 (UTC)

Can someone merge Kategorie:Fußballfunktionär (Ukraine) (Q8960528)

with Category:Ukrainian football chairmen and investors (Q15145176) 47.64.246.241 15:17, 1 March 2015 (UTC)

Can someone merge Kategorie:Fußballfunktionär (Vereinigte Staaten) (Q8960530)

with Category:American soccer chairmen and investors (Q8248022) 47.64.246.241 15:25, 1 March 2015 (UTC)

"is a list of" on categories

Hoi, I have added over 2000 times "is a list of" to category items. I have the impression that someone is deleting them. If this is the consensus, it is easier to do them all at one time. I would however like some discussion first as it does impact the functionality of Reasonator aversely. Thanks, GerardM (talk) 18:48, 21 February 2015 (UTC)

category combines topics (P971) seems the option for categories. Neither works with Reasonator. --- Jura 12:22, 22 February 2015 (UTC)
Reasonator is down.. However, for all the categories that have the "is a list of" you WILL get a list of items that fulfil the statements in the qualifiers. Thanks, GerardM (talk) 12:54, 22 February 2015 (UTC)
Oh, a new feature? In the past, the qualifiers had no effect, so the lists were quite useless. --- Jura 12:57, 22 February 2015 (UTC)
It seems to work. It's just that in the past, I tried qualifiers that don't ;). For things like Cultural Asset of National Interest (Q1019352), it seems impossible.
Anyways, it does seem useful. We could just rename the property to "is a list/category of". --- Jura 13:19, 22 February 2015 (UTC)
Categories are lists in their own way... No reason to change this property.. Better not ... Thanks, GerardM (talk) 13:52, 22 February 2015 (UTC)
Well, the description of the property should at least explain how it's meant to be used. --- Jura 14:21, 22 February 2015 (UTC)
I'm against is a list of (P360) on categories. It is abuse of a property, only used by Gerard and is added only for a feature of Reasonator. We have category's main topic (P301). Sjoerd de Bruin (talk) 16:19, 22 February 2015 (UTC)
First how is a category anything but a list and also why is this abuse? It has been widely published that it has this use. Thanks, GerardM (talk) 19:30, 22 February 2015 (UTC)
Publishing is something different than a discussion. This is abuse because it's only being added to get a feature in Resonator working. The Reasonator should be adjusted, not these kind of stuff. Sjoerd de Bruin (talk) 20:09, 22 February 2015 (UTC)
A list is by definition a finite and ordered set. A category may not be finite and not be ordered. Therefore a distinction can be made between those concepts. To me, the question is is this distinction useful to us and how ? Casper Tinan (talk) 21:02, 22 February 2015 (UTC)
I also prefer category's main topic (P301) and category combines topics (P971) over is a list of (P360). Pichpich (talk) 17:02, 22 February 2015 (UTC)
Is there anything that can be done with P971? --- Jura 17:30, 22 February 2015 (UTC)
There are currently over 2000 categories done in this way. They help to prioritise what category to do next. They help in the process of setting up AutoList. With over 2,000,000 edits using this process it should be obvious that it helps us a lot.
When there is a process to automate the adding of statements based on categories in Wikipedias, these statements can be used to set that up and once that is done, they can be removed without adverse effect. GerardM (talk) 06:56, 23 February 2015 (UTC)
Actually, I see the usefulness of your approach, but I was wondering what Pichpich's approach would look like. --- Jura 07:01, 23 February 2015 (UTC)
Discussion now also continuing at Property_talk:P360#P360_and_categories Jheald (talk) 08:52, 2 March 2015 (UTC)

Proposing category combines topics (P971) for deletion

Main: Wikidata:Properties_for_deletion#.7B.7BPfD.7CProperty:P971.7D.7D

I'm going to propose category combines topics (P971) for deletion, because is a list of (P360) is a far better model.

The problem with category combines topics (P971) is that by just specifying that a category combines eg "Paintings" + "London", it gives no way to distinguish whether a category is for "Paintings of London" or for "Paintings in London". It also doesn't specify whether the items in the category are going to be instances of paintings, or instances of Londons -- you have to then check back to the item for painting and the item for London and guess which one is more likely.

Contrast that with how is a list of (P360) handles list of women engineers (Q15832361): it specifies precisely the inclusion criteria, in a form that is natively Wikidata, that precisely corresponds to what would or wouldn't be included on appropriately items. It tells us that the members of this list will have the fundamental nature instance of (P31) human (Q5), that engineer (Q81096) is relevant because it specifies these items' occupation (P106), and that female (Q6581072) is relevant because it specifies these items' sex or gender (P21).

Therefore (particularly @Sjoerddebruin, Multichill:, I hope you won't mind if I nominate category combines topics (P971) for deletion, with a view to migration to is a list of (P360). I think it's the right thing to do.

(As for the point that is a list of (P360) is for "lists", I don't see any value being added is we make that restriction. The way to tell whether something is a list or a category is to look at its P31. What P360 does is to present inclusion criteria, and makes sense to do that with a common syntax on a single common property). Jheald (talk) 11:07, 26 February 2015 (UTC)

Let me now go slightly off-topic, to explain why I particularly care about this at the moment, and the challenge that moving to P360 would help me with.
My key focus project at the moment is to try to develop a pipeline that will take images of maps for which we have particular lat/long and scale data, and try to automatically identify what they are of, and how they should be categorised. To assist this, it would be particularly helpful to strengthen Wikidata's P31 identification of things like battles, castles, cathedrals, churches etc; so that if I have a map with a particular lat/long, I can look at the items that Wikidata returns close to that position, and see that one of them has a P31 of an appropriate scale (ie the scale of a battle), so a strong contender to be what the map is of; and with the P31 I can also identify what CommonsCat the map should go into (ie plans of battles in that region).
However, at the moment our P31 coverage for items can be quite poor. For example, out of about 180,000 items with co-ordinates in the UK and Ireland [1], more than 30,000 (17%) currently don't have a P31. [2].
One way to improve this is to mine the wiki categories. But then I hit the same problem that everyone else hits with wiki categories. After an indeterminate number of levels, suddenly the category no longer defines a narrower and narrower subset of battles; but instead becomes a category devoted to a particular battle, containing particular people that fought in it. Alternatively, one could be working on painters from Italy, but suddenly one reaches a category for a particular painter's works (or a category for a family of whom some, but not all, were painters). To the extent that it's possible (and that is going to vary from category to category) putting in machine-readable descriptions using is a list of (P360) would be incredibly helpful for extractions of this kind -- a simple crawler can then return a tree of categories of items that match the criteria, excluding ones that don't.
Adding P360s is for me the clear front-runner of how to add this desperately needed information to items for categories. It is the way I think we should go forward on this. And that is why I think we should retire category combines topics (P971) without delay, and migrate any information currently in a P971 to a P360 as soon as possible. Jheald (talk) 11:07, 26 February 2015 (UTC)
I've posted a few more ideas about where I think categories are going, and what I think Wikidata has to offer categories and category-crawling, at the end of the Phabricator item T87686.
It got a bit long (and may be a bit easier to read if you cut and paste it into your favourite editor); but I'd be interested to know what people think, and especially @GerardM: how much of this is already possible, once you get good coverage with is a list of (P360) properties? Jheald (talk) 22:10, 26 February 2015 (UTC)

Automate "is a list of" harvesting

When a category item includes the "is a list of" statement, it follows that all the articles in that article that conform to whatever they are a list of, can be given the statements that are the qualifier. When this is automated, a bot can regularly revisit all the categories associated with the item and harvest the data. This has many benefits:

  • Wikidata will have more comprehensive data for the subject
  • there is no need to do this revisit manually
  • "category data" will include many more redlinks when seen from the Wikidata perspective

Obviously, when categories include the data in Wikidata, it can become possible to have all kinds of configurations that define what a project shows. A category may be shown or not. Red links may be shown or not.

In my opinion there are two ways of doing this.. Use the qualifiers in the categories themselves or have a list where other paramaters are included as well for instance the depth a query may look in a given Wiki project.

What do you think ??? Thanks, 14:46, 27 February 2015 (UTC)


Hi @GerardM: I don't think it should be done fully automatically; rather, I think a semi-automatic process that generates a list for Magnus's QuickStatements would probably be best, that can be manually reviewed before execution.
But it would be good to have a regularly-updating automatic process that identified how many items there were in the category, that don't match the inclusion criteria in the is a list of (P360) -- like a constraint violation count. It would also be a good thing for this count to be accessible by WDQ like a regular property, so one could easily create a report for a paricular query of categories, that could be used to identify areas that need attention.


There are various reasons why I would resist a fully automated property-adding bot, principally because there are a variety of reasons that might be behind an article being in a category, without matching its inclusion criteria.
  • it might be a missing property on the item, eg a missing P31 (still a big problem for us), or a missing one of the other properties specified in the P360 -- this is the case that could be fixed by automatic or semi-automatic addition.
  • it might indicate an additional set of acceptable criteria: ie an additional P360 that should be added to the category item. For example, the en-wiki version of category Category:Spanish Armada (Q8785755) includes ships that were part of the Armada; people who were commanders of it; events in which it was associated; and more -- each of which might need to be specified in a separate P360. So it's entirely possible that an additional set of acceptable criteria could have been missed.
  • some articles are typically included in categories as auxiliary articles, even though they don't match the usual inclusion criteria of the category, that would be expressed in the P360. The most obvious of these is a survey article on the topic of the category as a whole -- eg perhaps "Dutch painters", which is not itself a painter from the Netherlands. This will normally be indicated by category's main topic (P301). However, there may be additional auxiliary articles also conveniently included in the category -- eg "List of Dutch painters" perhaps. I suggest that these should be identified by a new property -- "Category auxiliary item", one might call it -- indeed, perhaps "Category-related list" is such a common occurrence that it should have a property of its own. But there are other reasons that an article might be an "auxiliary article", so one needs to investigate whether the constraint violation is one of these.
  • there's always the possibility that the sitelink may be wrong due to a language muddle, and that this category item is not the correct category item for the wiki-category on the particular language wiki being investigated. A correct category item might have subtly different inclusion criteria.
  • finally, there's always the possibility of human error -- that the article in question should perhaps never have been in that category in the first place.
For all of these reasons, I would be against a fully automatic approach.
But a supervised automated tool could IMO be very valuable. As also could be gamification through one of Magnus's toys ("does this article meet these criteria?") Jheald (talk) 16:30, 27 February 2015 (UTC)
I have proposed new properties List related to category, Category related to list, and Category auxiliary item to help define items that should be excluded from category-membership constraint violations. Jheald (talk) 18:30, 27 February 2015 (UTC)
Also Parent category. Jheald (talk) 19:36, 27 February 2015 (UTC)
An item would only be included with a statement when the qualifiers apply. So it needs to be a person only then can it be "educated at" "Harvard".. A manual phase helps how ? Thanks, GerardM (talk) 18:59, 27 February 2015 (UTC)
@GerardM: Because if you look at categories like "People educated at Harvard" you find they also contain things like "List of people educated at Harvard", which would get on the list of constraint violations -- and sometimes other sorts of auxiliary items, not so easily spotted as a simple person/non-person distinction (particularly if you're working in a non-person category).
I suggest we take it gently, step by step, until more people have experience of doing this in a machine-assisted semi-automated way first, so that we get a community of editors that understand and have experienced what sort of issues do or do not arise, and how often, before we take the definitive step to go full automatic. Other people need the chance to catch up in their experience and understanding to where you've got to. Jheald (talk) 19:36, 27 February 2015 (UTC)
When the first statement says .. "is a list of" "human" it cannot be a list or anything else so what is the problem? Thanks, GerardM (talk)
@GerardM: I thought we were talking about categories. If you look at a real category, at least on en-wiki, such as en:Category:Harvard University alumni, you will see that the second item is en:List of Harvard University non-graduate alumni, which is not a person; and the first is en:Harvard alumni health study which is not a person either. So one cannot just take the contents of the category, and tag every item with "alumnus of Harvard". Almost all, yes; but not quite all. Jheald (talk) 22:00, 27 February 2015 (UTC)
@GerardM: So you're assuming you can filter based on the P31 matching, and that will root out all false positives of items needing fixing. It may also remove a number of true items that need fixing, because our P31 coverage isn't always all that strong. But it's the rate of false positives that are the real concern, if we're using a machine to add content.
Does what you're suggesting make sense? It may do. Probably if a human was scanning a long list of items to be tagged, they too might not spot many false positives other than those with a blatantly wrong P31. But will that get everything? I simply don't know. I just haven't looked at enough 'non-usual' category entries to feel that I definitively know, one way or the other. What might be useful would be to scan a lot of categories, try to identify all the 'non usual' entries other than list articles and survey articles on the topic, and see whether we can convince ourselves that we then understand the full range of sorts of articles that can additionally appear in a category, other than regular ones that are included by virtue of a P360 inclusion criterion that can be stated for the category. Jheald (talk) 22:38, 27 February 2015 (UTC)
I do not care all that much if it gets "all" of them. I care that they are right. So when I harvest from the Indiana State Senators category, I will only include those articles that have an item and have a statement indicating that they are human. When this process is repeated at a later date, new items and items that have recently been identified as human will be included. Great. I do not care for the articles that are in the category and do not really fit. I harvest and for many categories Wikidata will be more complete than any Wikipedia category.. Stop thinking Wikipedia.. The proposal is to benefit Wikidata first. Thanks, GerardM (talk) 22:48, 27 February 2015 (UTC)
I care about the articles which do not really fit, because I want to make sure that they do not get into your list. You're right: filtering on P31 should catch most of them. But I want to know we can catch all of them -- and to be confident about that is the reason that I want to get a better sense of what a big enough sample of them look like. Jheald (talk) 22:57, 27 February 2015 (UTC)
The other thing is that I do care about what articles get missed out, because for what I want to do -- namely to understand what sort of objects are the items being returned close to a particular latitude/longitude -- to be able to do that I want as comprehensive P31 coverage of items like cities, castles, churches, cathedrals, battlefields, villages and towns as possible. And I need it as soon as I can get it. Jheald (talk) 23:05, 27 February 2015 (UTC)
(ec) There are still also the other potential issues that I listed above -- eg a bad sitelink, linking a category to a non-equivalent item, that has a slightly mismatched P360; or a category with additional routes for inclusion, in addition to those specified by its current P360s -- both of these could lead to attempts to add statements that are incorrect that might potentially be spotted by a human phase. Jheald (talk) 22:52, 27 February 2015 (UTC)
Another example: Stamford (Q1000662) included in en-wiki's Category:Castles in Lincolnshire (Q8346352) not because it is a castle, but because it has a castle (which is discussed in the article). I know that you're more interested in people articles with a confirmed instance of (P31) => human (Q5), which may be a lot safer; but even then, are there really never examples that a human being might spot as not quite fitting the criteria, despite being included in the category?
What (IMO) ought to be done with Stamford Castle is a redirect created on en-wiki, and linked to Stamford Castle (Q17644180), with the redirect (rather than the target) being put in the category Category:Castles in Lincolnshire (Q8346352). (Though I know you and Jane disagree with that). But even so, there's an issue, because Stamford Castle (Q17644180) => instance of (P31) => scheduled monument (Q219538), which is not a subclass of castle (Q23413). So perhaps really a new item should be created for Stamford Castle, that Stamford Castle (Q17644180) would be part of (P361). Jheald (talk) 14:03, 28 February 2015 (UTC)
Per advice from WikiProject Cultural Heritage, I have marked Stamford Castle (Q17644180) now additionally with instance of (P31) => castle (Q23413) and instance of (P31) => ruins (Q109607). Jheald (talk) 08:46, 1 March 2015 (UTC)
In the same tree there's also adulterine castle (Q16970488), which should => subclass of (P279) => castle (Q23413), rather than instance of (P31) => castle (Q23413). (Just collecting more examples of the kind of things that may need to be watched out for when crawling down a category tree. Is there an existing FAQ on this anywhere?) In this case this would be the kind of P360 exception that I would propose to use the suggested new property Category auxiliary item to mark up, so Category:Castles in England (Q7131216) => Category auxiliary item => adulterine castle (Q16970488)) Jheald (talk) 14:18, 28 February 2015 (UTC)
Hi @Jheald: What you are pointing at has little to do with what I propose. The proposal is to harvest what fits obviously in Wikidata. When it means that 30% of a category is not included, Is leave it for another day. When statements can be added to castles, the first line of business is to find that they are a castle and only then add new statements. When we do it in this way, we add a lot of information from ALL the Wikipedias. It is NOT a Wikipedia centric approach. It does not care at all about what problems a Wikipedia has by being inconsistent. The idea is not to replace the content of categories, it is to harvest categories and do a better job at that. Thanks, GerardM (talk) 10:01, 1 March 2015 (UTC)
Hi @GerardM: To be clear, my first interest here, just like you, is to be accurately harvesting in information from categories into Wikidata. The longer-term possibility to give category views on Wikipedias some nice new features is a free bonus, not my immediate priority.
For something like Category:Castles in England (Q7131216) and its first-level children (which actually are reasonably straightforward), the important property often missing from items is the instance of (P31) => castle (Q23413), which is therefore what I am so keen to harvest from the category, and the context in which I am so keen to explore the challenges that can arise to such harvesting. (More ambitious will be to do similarly for something like Category:Battles (Q7145329))
I appreciate now that you are asking about a slightly more specific case, where the P31 is already in place, and what you would be adding are additional properties based on the qualifiers of the P360 -- for such a purpose you can use the P31 to filter out the articles which are most obviously inappropriate, which makes what you are proposing likely to be much safer.
But I would still like to know: what sorts of false positives have you still encountered? Can you give examples? Jheald (talk) 11:57, 1 March 2015 (UTC)
Hoi, harvesting the "instance of" for an article is not something that can be safely automated. What can be automated is what makes the "whatever" more specific. Thanks, GerardM (talk) 17:59, 2 March 2015 (UTC)

[Global proposal] m.Wikidata.org: (all) Edit pages

 
MediaWiki mobile

Hi, this message is to let you know that, on domains like en.m.wikipedia.org, unregistered users cannot edit. At the Wikimedia Forum, where global configuration changes are normally discussed, a few dozens users propose to restore normal editing permissions on all mobile sites. Please read and comment!

Thanks and sorry for writing in English, Nemo 22:32, 1 March 2015 (UTC)

Duplicate

are P1720 (P1720) and P1719 (P1719) duplicate? --Rippitippi (talk) 14:52, 2 March 2015 (UTC)

Not at all, why? Sjoerd de Bruin (talk) 15:17, 2 March 2015 (UTC)
for the difference see en:Posthumous name and en:Temple name --Pasleim (talk) 15:21, 2 March 2015 (UTC)

Wikidata Skim - a (very) limited replacement for Wikidata Query / Autolist

Hi everyone, for those times when Wikidata Query (and Autolist as well) is down or very slow (like right now), i wrote a (very) limited replacement called Wikidata Skim, it's over here:

http://tools.wmflabs.org/hay/wdskim/

It just does one claim, with both property and item fields filled in, and it returns a maximum of 50 results. It's based on the 'linkshere' functionality of the API.

It's not much, but it's better than nothing, i suppose :)

Husky (talk) 16:39, 2 March 2015 (UTC)

I've added an option for images as well, and a simpler interface. So, here are all paintings by Frans Hals (Q167654). Husky (talk) 17:33, 3 March 2015 (UTC)

Wikidata API: Does Wikidata item exist?

What is the easiest way using the Wikidata API to find out given a Wikidata ID whether an item corresponding to that ID exists or not and if it exists whether it is an redirection or a proper item? --Jobu0101 (talk) 21:14, 2 March 2015 (UTC)

Probably /w/api.php?action=wbgetentities&format=json&ids=Q9817212&redirects=yes&props=info You can try with a normal item, a redirect or an item that don't exist --ValterVB (talk) 21:30, 2 March 2015 (UTC)
Thank you very much. --Jobu0101 (talk) 21:34, 2 March 2015 (UTC)
*Agrees* ·addshore· talk to me! 22:40, 2 March 2015 (UTC)

WDQ not updating

Hi all, is there a problem with WDQ using an older database than the live Wikidata? I am updating values for CLARA-ID (P1615) and though I updated Élisabeth Louise Vigée Le Brun (Q213163) on 25 February it still won't be listed in a "claim[1615]" query in autolist. Am I missing something? Thx, Jane023 (talk) 08:37, 28 February 2015 (UTC)

I just noticed that Carla Lavatelli (Q17404034) is listed now, and her entry for CLARA-ID (P1615) was updated on 15 February - weekly updates maybe? Jane023 (talk) 08:40, 28 February 2015 (UTC)
It's lagging these days. http://wdq.wmflabs.org/stats has the date of 2015-02-09T02:06:06Z --- Jura 08:43, 28 February 2015 (UTC)
Thanks! I am getting a bit farther, but that still doesn't make sense in this case because Barbro Bäckström (Q4941467) had the CLARA-ID (P1615) value added 10 February and though this is not reflected in the dump, the one I listed above for Carla Lavatelli (Q17404034) updated on 15 February is. I must be missing some sync option for the mix-n-match data. Pinging @Magnus Manske: Jane023 (talk) 09:08, 28 February 2015 (UTC)
Strange issue of unknown origin. YuviPanda and I will be on it in the coming days. --Magnus Manske (talk) 21:14, 3 March 2015 (UTC)

Wikibooks-related Badges

Would it be possible to add Featured Book, Featured Ingredient and Featured Recipe as qualifiers badges on Wikidata, now that b: items can be added to Wikidata? It Is Me Here t / c 20:08, 3 March 2015 (UTC)

You mean "badges" instead of "qualifiers"? To add them as badges, we would need an item created for each badge, like Q6540291 and then it is a configuration change. Aude (talk) 13:30, 2 March 2015 (UTC)
Sorry, yes, that's what I meant. User:Lydia Pintscher (WMDE) says that, if there is enough support in the community for this change, then WMF will implement it. It Is Me Here t / c 20:08, 3 March 2015 (UTC)

http://wdq.wmflabs.org down

Seems like http://wdq.wmflabs.org is down (at least if you request a query). Does somebody know anything about it? --Jobu0101 (talk) 21:47, 3 March 2015 (UTC)

Yuvi and Magnus are working on it but it might take a while to be resolved. --Lydia Pintscher (WMDE) (talk) 10:26, 4 March 2015 (UTC)

WikiProject

I'm searching for a WikiProject to add sources to dates for birth and dead. Thank you, Conny (talk) 10:58, 4 March 2015 (UTC).

List of songs

Is there any way how to link John Cale (Q45909) and list of John Cale songs (Q17280100)? Matěj Suchánek (talk) 16:39, 1 March 2015 (UTC)

I added "Performer:John Cale" as a qualifier to "Is a list of:song" on list of John Cale songs (Q17280100). Filceolaire (talk) 22:57, 1 March 2015 (UTC)
Thanks anyway, I more curious about the second way (from John Cale (Q45909)), though. Matěj Suchánek (talk) 15:22, 2 March 2015 (UTC)
@Matěj Suchánek: I believe you are looking for list of works (P1455). See Wikidata:WikiProject Music for more helpful music properties. Sweet kate (talk) 21:38, 4 March 2015 (UTC)

Inspire Campaign: Improving diversity, improving content

This March, we’re organizing an Inspire Campaign to encourage and support new ideas for improving gender diversity on Wikimedia projects. Less than 20% of Wikimedia contributors are women, and many important topics are still missing in our content. We invite all Wikimedians to participate. If you have an idea that could help address this problem, please get involved today! The campaign runs until March 31.

All proposals are welcome - research projects, technical solutions, community organizing and outreach initiatives, or something completely new! Funding is available from the Wikimedia Foundation for projects that need financial support. Constructive, positive feedback on ideas is appreciated, and collaboration is encouraged - your skills and experience may help bring someone else’s project to life. Join us at the Inspire Campaign and help this project better represent the world’s knowledge!

(Sorry for the English - please translate this message!) MediaWiki message delivery (talk) 20:01, 4 March 2015 (UTC)

Broken links

If a link in a reference is broken, do we have a way to mark the reference such? --Denny (talk) 23:59, 4 March 2015 (UTC)

Occupations

... are now very big mess. You can find many examples at Wikidata:Database reports/Constraint violations/P106.

In Property talk:P425#"Field of" qualifications or degrees? Gymel used very nice and simple scheme:

(My addition:)

However, the current state does not well recognize whether an item is an activity or an occupation.

My suggestions thus are:

  1. all targets of occupation (P106) should be profession (Q28640) (subclasses) and not activity (Q1914636) nor position (Q4164871)
  2. all targets of occupation (P106) should have field of this occupation (P425) (not occupation (P106))
  3. all targets of field of this occupation (P425) should be activity (Q1914636) and not profession (Q28640)

More discussions:

Matěj Suchánek (talk) 12:11, 26 February 2015 (UTC)

Hi, @Matěj Suchánek: An academic publication about this for example. I recommand to watch fig1. Their "activity" class says that all activity instances have a start and an end date. Every time somebody makes a hat, it also has a start and an endate, so hatmaking is a subclass of activity (every instance of hatmaking is also an instance of activity). I'd wrote

⟨ hatmaking ⟩ subclass of (P279)   ⟨ activity ⟩

. TomT0m (talk) 14:08, 27 February 2015 (UTC)

Re: #1, items with occupation (P106)  :activity (Q1914636), 39,638 out of 40,723 are because singer (Q177220) has been an activity since this claim by User:Giftzwerg 88, who therefore might like to weigh in on this discussion. --Haplology (talk) 14:32, 27 February 2015 (UTC)
I like this proposal and I have removed the ":" and the "Singer:instance of:activity" claims. 'Activity' is a process, not an occupation. 'Singing' is an activity. 'Singer' is an occupation. Filceolaire (talk) 11:05, 28 February 2015 (UTC)

I have added one more value type constraint to occupation (P106). Matěj Suchánek (talk) 15:39, 5 March 2015 (UTC)

date and time type

I tried add property with year only, and I found that system accept such input but set 0 to month and day value. Is such behavior intentional? In next edit I changed the value to 1st January because the zeros are not properly handled in my Lua module yet. Paweł Ziemian (talk) 21:59, 3 March 2015 (UTC)

I think you might need to read the precision first and then the significant digits. --- Jura 22:37, 3 March 2015 (UTC)
Yes this is intentional. If you do not provide the data for a part they are set to 0. The precision needs to be taken into account. --Lydia Pintscher (WMDE) (talk) 10:27, 4 March 2015 (UTC)
Thank you for clarification. Paweł Ziemian (talk) 22:14, 4 March 2015 (UTC)
I'm interested in what software parses the unique Wikidata date/time string. Would you be willing to provide more information about you're Lua module? Jc3s5h (talk) 17:57, 4 March 2015 (UTC)
The submodule is here. Generally I try to invent the wheel and provide some infrastructure for easier use the data in infoboxes, which handles such cases like date of birth (P569) of Frédéric Chopin (Q1268) or Isaac Asimov (Q34981) with help of that submodule. Paweł Ziemian (talk) 22:14, 4 March 2015 (UTC)
I see in the module linked by User:Paweł Ziemian a comment that the before and after fields are not understood yet. I think this is going to be a problem for Wikidata. The DataModel indicates that the uncertainty would be -before + after in units of precision. So if time were "+00000001850-00-00T00:00:00Z", precision were 9 (year) before was 4, and after was 5, it would mean the event occurred in 1850 -4 years + 5 years, that is, between 1856 and 1855. The problem is that the user interface only allows people to input the precision, not before or after. So I think we have to interpret before and after as always being at least 1, even if they are set to 0.
This means that if time is "+00000001850-00-00T00:00:00Z", precision is 9, before is 0, and after is 0, we should consider the event to have occurred in 1849, 1850, and 1851.
It is sad that poor implementation of the user interface has thrown doubt upon many dates stored in the database. Jc3s5h (talk) 02:52, 5 March 2015 (UTC)

┌────────────────────────────────────────────────────────────────────────────────────────────────────┘I have looked into this further. Near the beginning, DataModel states

The data model is conceptual ("Which information do we have to support?") and does not specify how this data should be represented technically ("Which data structures should the software use?") or syntactically ("How should the data be expressed in a file?"). Separate documents describe the serialization of the Wikibase data model in JSON and in RDF.

So apparently the real technical information on how the data is represented is in m:Wikibase/Notes/JSON. That says that before and after are "currently unused, may be dropped in the future". That means that precision is the only indication of precision. I will now start a new thread on what precision should mean. Jc3s5h (talk) 14:36, 5 March 2015 (UTC)

WDQ up again ...

Hoi, WDQ is working again.. I want to thank Yuvi and Magnus. Thanks, GerardM (talk) 05:51, 5 March 2015 (UTC)

Thanks guys! Husky (talk) 13:48, 5 March 2015 (UTC)
Thanks both of you.--Vyom25 (talk) 14:00, 5 March 2015 (UTC)

Precision in date and time

This thread pulls out a detail from the #Date and time thread above. As described in m:Wikibase/Notes/JSON, when a property such as a date of birth (P569) is exported, a number of fields are exported to describe the date and time. One of these fields is precision, which is set to an integer. For example, 9 indicates the precision is years, and 13 indicates the precision is minutes.

The description of what precision means is "To what unit is the given date/time significant?" I think users should be provided with a better description. One possible interpretation is to ignore all information less precise than precision and consider the event to fill the indicated time period. For example, if American Independence day 2014 is written with the time "+2014-07-04T00:00:00Z", everything after "04" should be ignored and the celebration of Independence Day should be considered to occupy the entire period from 00:00 hours to 24:00 hours, local time, July 4, 2014.

A different interpretation is that the exact value is unknown, but the time unit indicated by precision contains useful information, while smaller time units do not contain useful information. So if it were known that a certain person were born within six months of September 63 BC, it would be useful to know the month but it would be useless to try to guess the day, hour, minute, etc. Such a case could be written with the time "-0062-09-00T00:00:00Z" and precision 10 (months).

So how should we put this into the relevant documentation so users will be on alert that precision may have either of these meanings? Jc3s5h (talk) 15:07, 5 March 2015 (UTC)

The problem is that precision is often not this simple. From documents we maybe know NN got a son in the year 1234, but he was not mentioned in the census-records of 1203. That means he was born not earlier than 1204, and most unlikely not later than 1220. That is hard to describe in a simple way today, and I do not think the reality should adapt to the software.
I work a lot with astronomical objects here. The precision in the databases I have access to show the precision, not in the number of digits, but in how large a standard deviation is. That way of describing precision does not exactly tell you how many digits are valid. You have yourself to choose how many standard deviations wide the precision is when you display the number. -- Innocent bystander (talk) 17:02, 5 March 2015 (UTC)
There are two separate issues. First, should we improve the documentation to describe how the system works today, and let the user know that the current structure only gives them a general idea of the date, time, and precision? If so, how?
Second, should we make the structure more complex so we could make statements such as "the birth occurred between 00:00 hrs January 5, 1610 and 24:00 hrs September 27, 1615" and "the person died in 1430 BC with a standard deviation of 25 years". We already see widespread evidence of editors making the occasional hand edit, and bots being set loose to import from various data bases, with no hint that the editors or bot operators ever heard of the Julian calendar. So I wonder if a more complex structure would just insure more errors. Jc3s5h (talk) 17:15, 5 March 2015 (UTC)
And only a few know what happened in Sweden February 30, 1712! Julian and Gregorian are not the only calendars, I'm afraid. These datatypes were opened for editors premature, I'm afraid. All of Wikidata opened to early. -- Innocent bystander (talk) 17:31, 5 March 2015 (UTC)
Innocent bystander: It's the first footnote in my german article about Carl Linnaeus (Q1043) :) --Succu (talk) 17:47, 5 March 2015 (UTC)
And that's why we in Sweden used to celebrate the death of the great hero king twice a year. :) -- Innocent bystander (talk) 17:56, 5 March 2015 (UTC)

Ballet dancer

Just encountered ballerina (Q4070300) and ballet dancer (Q805221). I thought we didn't had gender specific professions on Wikidata... Sjoerd de Bruin (talk) 20:48, 2 March 2015 (UTC)

They both have separate links in ruwiki, so they can't be merged. --Jakob (talk) 21:26, 2 March 2015 (UTC)
We probably need to ask a Russian speaker to explain the distinction made on ru.wiki. Pichpich (talk) 23:31, 2 March 2015 (UTC)
You shouldn't use ballerina (Q4070300) as a profession. Use ballet dancer (Q805221) instead. Filceolaire (talk) 00:14, 6 March 2015 (UTC)

the fringes of notability

Today we got a new item, Sammy Zimmermanns, created and edited entirely by User:SammyZimmermanns. His gender is a disambiguation page and his name is a single by Dutch singer Ramses Shaffy, among other problems. He claims to be a marketing expert and an author, but strangely for an "author", his name returns nothing in Google Books. In a case like this, how would you assess notability? He has provided no references, and a quick web search yields only social or self-generated media. Also, simply being a marketing expert seems less than notable. For that, I think a person should be recognized by third parties in some way.

My intuition says he is not notable, but what are your rules of thumb? I have read WD:N, but it seems vague and I still wonder about this.

I just checked again, and his commons gallery has been deleted entirely.

Besides people, how about small, generic companies? --Haplology (talk) 03:06, 4 March 2015 (UTC)

It might be just one of the persons following the invitation to move their contributions from Freebase to Wikidata ;) --- Jura 04:44, 4 March 2015 (UTC)
@Jura: :]
@Haplology WD:n says that an entity can be notable if she "can be described using serious and publicly available references". While the phrasing might be a bit too lax, I think it already excludes someone who is not mentioned in any source we find. An European or American author wihout any VIAF entry sounds kind of suspicious.
For small business, I don't know. It may not a good idea to add them individually, but it might may sense to upload opendata repositories that contain data about companies. --Zolo (talk) 05:49, 4 March 2015 (UTC)
This is the same issue as the MusicBrainz conversation above. That "serious and publicly available references" wording in the second option in WD:N is very, very nebulous. What does "described" mean? Just proof of existence, or more data? What does "serious" mean? Is a business's online directory of employees "serious," making every single listed employee notable? Etc. Sweet kate (talk) 21:46, 4 March 2015 (UTC)
I would prefer, in the case of Wikidata, the notion of identifiability. Notability is terribly encyclopedia oriented and prone to value judgment. Having a wikidata item is not a proof of value, it should be ony that we need the notion or object to build a precise statement. TomT0m (talk) 20:06, 5 March 2015 (UTC)

Using uris in templates as item identifier

Hi, I'm happy to announce that if you now use the {{Q'}} template with the uri of an item, it works as if you had just typed the Qnumber. For example
{{Q'|https://www.wikidata.org/wiki/Q42}} or {{Q'|https://www.wikidata.org/entity/Q42}} will render
Douglas Adams (Q42)      or Douglas Adams (Q42)     .

This means we can now just copy/paste the uri of the browser \o/ TomT0m (talk) 20:11, 5 March 2015 (UTC)

The coolest thing about this is that we can actually drag&drop from the item search widget who is in every wikipage.

For example I searched this item, draged and dropped from the suggestions, and the url popped https://www.wikidata.org/wiki/Q12191 which gives with Q' Nantes (Q12191)     
This also works in {{C}} who uses {{Q'}}.

Merge problems

  --Csigabi (talk) 10:16, 6 March 2015 (UTC) Can someone merge: Kategorie:Société Générale (Q9119668) with french Catégorie:Société générale (Q9417197)

Can someone merge: Kategorie:Veolia (Q9158591) with english Category:Veolia (Q6400957) 88.70.25.20 04:36, 6 March 2015 (UTC)

Double redirect

Through merging, I accidentally created a double redirect: Q19359738 redirects to Q12883558, which in turn redirects to Q2890983. In the future, how should I resolve such double redirects? Gabbe (talk) 10:14, 6 March 2015 (UTC)

My bot is resolving double redirects within 24 houres after they pop up at Special:DoubleRedirects --Pasleim (talk) 10:19, 6 March 2015 (UTC)

Merge problems

Can somone merge German Kategorie:Casio-Produkt (Q16661005) and English Category:Casio products (Q8345851) ? Can someone merge German Kategorie:Casio (Q8918869) and English Category:Casio (Q7321342) ? 88.70.210.220 13:41, 6 March 2015 (UTC)

  Done--Mrutyunjaya Kar (talk) 13:45, 6 March 2015 (UTC)

Can someone merge German Kategorie:Casio (Q8918869) and English Category:Casio (Q7321342) ? 178.3.19.119 16:25, 6 March 2015 (UTC)

  Done --Marek Koudelka (talk) 17:01, 6 March 2015 (UTC)

Q7639918 protected ?

I am not able add the OrWiki page as the edit button is not visible. Is Q7639918 protected ?--Mrutyunjaya Kar (talk) 12:11, 6 March 2015 (UTC)

I see no sign of a protection. Has javascript been disabled in your browser? -- Innocent bystander (talk) 12:29, 6 March 2015 (UTC)
No, I recently edited more than 5 pages, the problem is 'only' with Q7639918. --Mrutyunjaya Kar (talk) 12:38, 6 March 2015 (UTC)
No, it is not protected. If you leave a link here, someone might be able to add it to the item.--Ymblanter (talk) 17:52, 6 March 2015 (UTC)
Maybe it's related to the "caching issue" mentioned at WD:DEV#Q2782419. -- Innocent bystander (talk) 18:43, 6 March 2015 (UTC)
  Done may be caching issue.--Mrutyunjaya Kar (talk) 06:31, 7 March 2015 (UTC)

Monolingual text

There are properties with monolingual type. Internally it is constructed as pair of language code and string value. The language code is described as UserLanguageCode, which is similar to BCP 47. However there is also stated that the value is based on the language preference setting of logged in Wikipedia user. The problem is that the edit box does not accepts any language code, which might be created with BCP 47 rules. I wonder if the list of all acceptable codes is compatible with this one, and why at all there is list of available language codes. This might be against to precise description of language used in native label (P1705), which could be some local dialect, and thus additional subtags in language code are required, or even language not supported in Wikipedia yet. Paweł Ziemian (talk) 21:32, 6 March 2015 (UTC)

User’s edits are not patrolled

Raid5 is doing lots of translations into Finnish. Many of these show up as unpatrolled in the Recent Changes list (example, skip to end of list); however, at the actual diff, there is no option to patrol the edit. AFAIK, the user should also be autoconfirmed and the edits therefore be autopatrolled. What’s going on here? —DSGalaktos (talk) 11:47, 7 March 2015 (UTC)

Seems like it’s only on the Wikidata: pages where I don’t get a patrol link; it works fine on the Translations: pages. Do you need special permissions to patrol in the Wikidata: namespace? —DSGalaktos (talk) 12:07, 7 March 2015 (UTC)
No, that can’t be it, I was able to patrol this edit without problems. Now I’m confused again. —DSGalaktos (talk) 12:09, 7 March 2015 (UTC)
Well, he is now autoconfirmed and in the future his edits are automatically marked as patrolled. --Stryn (talk) 17:14, 7 March 2015 (UTC)
Alright, thanks! —DSGalaktos (talk) 18:01, 7 March 2015 (UTC)

Help choosing English label

Can someone help me identify what the label for humanities scholar (Q16727193) should be in English? I'm assuming it's a name for an occupation related to humanities, but not 100% sure. NavinoEvans (talk) 23:31, 7 March 2015 (UTC)

Wikidata weekly summary #148

 
Here's your quick overview of what has been happening around Wikidata over the last week.

Statistics

Is there statistics about every language use in wikidata? And per user (according the language he/she uses)? Xaris333 (talk) 01:54, 8 March 2015 (UTC)

There is no data on the use of Wikidata. There is data on the use of Reasonator. Do not ask me why except that it was deemed to be meaningless in the USA. Thanks, GerardM (talk) 08:24, 8 March 2015 (UTC)
See Wikidata:Database reports. --Stryn (talk) 09:28, 8 March 2015 (UTC)

Q317279 Owain Glyn Dŵr

The interwiki link to cy:Owain Glyn Dŵr is correct on Wikidata (Q317279), but from the English Wikipedia only the "Cymraeg" link is to the invalid Nodyn:Llinach_Owain_Glyn_Dwr page (all other language versions of the article have the correct cy link). Seems like some sort of weird glitch -- can someone fix it? BabelStone (talk) 11:47, 8 March 2015 (UTC)

  Done The problem was here (the link had been transcluded by the template). Matěj Suchánek (talk) 12:07, 8 March 2015 (UTC)
Thanks, I would never have found that! BabelStone (talk) 13:02, 8 March 2015 (UTC)

Qualifier for escutcheons in coat of arms image

Hi. en:Template:Coat of arms, like other language versions of this template, is based on the following principles:

"Based on conventional heraldic practice, the following premise ensures a certain degree of uniformity and distinctiveness of the motifs even at low resolutions (20 pixels is the default width):
  • I: Escutcheons exclusively
All elements of achievement save the indispensable escutcheon are omitted.
E.g. for Belgium,   is shown instead of more elaborate variants such as  ,   or  .
Similarly, for Austria:   is shown instead of  .
  • II: Marshalling avoided
When an escutcheon has multiple variants, the one with least marshalling, i.e. the simplest, is employed.
E.g. for Sweden,   is shown instead of the following quartered variant:  .
Similarly, for Liechtenstein:   is shown instead of  ."

With regard to principle I, I propose a qualifier named "escutcheon" for coat of arms image specifying that a given file actually is an escutcheon, like  , and not a greater coat of arms (which often would be meaningless for such templates) like  . This change would certainly make Wikidata more useful for the use of heraldry in editions of Wikipedia. Could this be done? Thanks in advance. - Ssolbergj (talk) 14:32, 28 February 2015 (UTC)

It sounds like the guidance above needs to be added to the talk page for coat of arms image (P94) so that this property P94 is only used to link to the version of the escutcheon without marshalling and without elements of achievement. If the wikidata item for a location links to more than one image then one of the images should be marked as 'preferred'. Usually this will happen if the coat of arms changes over time and each one will have qualifiers for 'start date' and 'end date', with the current version marked preferred.
You could use depicts (P180):escutcheon (Q331357) as a qualifier to coat of arms image (P94) for the link to the plain version. Use shown with features (P1354) as qualifier to coat of arms image (P94) to note that an image has marshalling or elements of achievement or better yet don't use use P94 to link to these, unless I have misunderstood something - save this for the CommonsData item for these images when that happens. Filceolaire (talk) 22:25, 1 March 2015 (UTC)
Thanks for your answer. Do you or someone else know how for instance the coat of arms image for Sweden that according to the qualifier "depicts" an "escutcheon" can be accessed on English-language Wikipedia? en:Module:Wikidata does not seem to explain this. - Ssolbergj (talk) 22:30, 3 March 2015 (UTC)
Ssolbergj. To the left of each of the links to coat of arms images in Sweden (Q34) there are three little boxes. If you click on the edit button for an image then you can edit these little boxes and make the image you want the template to use into the "Preferred" image. as I have done for "Shield of arms of Sweden.svg". See Help:Ranking for more info. If a statement has more than one value then the template should treat the 'preferred' value as the default for that property. Filceolaire (talk) 00:31, 9 March 2015 (UTC)

Property creation

I now have the "Property creator" right on this project, and am working through the considerable backlog.

Would there be any objections to me creating properties which I have proposed, and which are uncontroversial, but long overdue, such as this one? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:28, 6 March 2015 (UTC)

None, then. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 10:59, 9 March 2015 (UTC)
Thanks for creating them! --- Jura 11:07, 9 March 2015 (UTC)

Founder of

Is there any inverse property for founded by (P112)? I mean: How to link from John Cale (Q45909) to Life Along the Borderline (Q12033514) or from Michael Eavis (Q2267003) to Glastonbury Festival (Q309066)? --Marek Koudelka (talk) 17:11, 6 March 2015 (UTC)

We generally don't have reciprocal properties; you can use What links here. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 18:50, 6 March 2015 (UTC)
What links here is a very blunt too as it shows all links to one item, not just the property founded by (P112). Wikidata query gives the results you want (what did Michael Eavis (Q2267003) founded?) but is a little more difficult to use. /ℇsquilo 09:21, 9 March 2015 (UTC)

A couple questions

I have been trying to figure out how to edit here and I think I am getting the hang of it but I had a couple questions related to things I have seen.

  1. If Wikidata has the interwiki link, should that interwiki link be removed from the corresponding Wiki if its still there? I have seen several cases where the interwiki link was on the Wikidata entry, but also duplicated in the source of the article as well.
  2. Does anyone know if there is an item property to designate a namesake of a person. or example, Smedley Butler and Hiram Bearss both have ships named after them and I have linked the ships to their Namesakes, but I cannot find a corresponding value to link the ships or other namesakes from their page.
  3. Do we use redirects here? If I put in the name Hiram Bearss for example, nothing comes up. I have to know that his entry is under Hiram I. Bearss and that seems somewhat less than ideal.
  4. There also seem to be a lot of properties missing with relation to ship and aircraft data. For example, I can add Gross tonnage of a ship, but I cannot add complement, draft, beam, range, speed or a number of other factors.

Thanks in advance for the help. Reguyla (talk) 15:19, 8 March 2015 (UTC)

Interwikis I can say at least that with your first question, it's best to check and see if the project in question wants to delete interwiki links or not. For instance, it is routine to delete them on w: but there is a big discussion on keeping them over at n:Wikinews:Water cooler/policy/archives/2015/January. —Justin (koavf)❤T☮C☺M☯ 15:49, 8 March 2015 (UTC)
Ok thanks. So far most of the ones I have found are on ENWP and can't do anything about them but I have also found a few on others as well. Thanks. Reguyla (talk) 16:02, 8 March 2015 (UTC)
Hi,
  1. 99% interwiki had been removed from all Wikipedias, most from Wikisource and Wikiquote. They are still present in some articles due to possible interwiki conflicts. Today I found a way how to track them using the search engine.
  2. Proposing is possible but creating is not sure.
  3. You propably mean aliases.
  4. Very common question: the developers are currently working on launching properties with units. They are not available at the moment.
Feel free to ask more if you want. Matěj Suchánek (talk) 16:14, 8 March 2015 (UTC)
Thank you. Reguyla (talk) 16:16, 8 March 2015 (UTC)
  1. Yes.
  2. The inverse property exists; named after (P138)
  3. No, we don't use redirects that way. We use aliases instead (third edit-button from the top).
  4. Most of those properties are proposed and are pending the creation of apropriate datatypes.
ℇsquilo 09:13, 9 March 2015 (UTC)

Addition of Wikidata-to-DBpedia classes/properties mappings

Hi everyone!

I started adding a few equivalency claims that map Wikidata classes/properties to DBpedia ones:

The procedure is as follows:

  1. The claim has property == equivalent class or equivalent property (depending on the item), and value == URI of the DBpedia ontology item;
  2. Add a qualifier with property == described at URL, and value == URL of the description of the DBpedia ontology item;
  3. Add an imported from reference to DBpedia

Do you agree with such modeling?

The next steps would be to automate the procedure through a bot, add all the available mappings in DBpedia, and let the community double-check their accuracy.

Looking forward to hearing your thoughts! --Hjfocs (talk) 10:31, 9 March 2015 (UTC)

Good work. Thanks. --- Jura 11:04, 9 March 2015 (UTC)

Films with multiple genres

I've been doing some batch editing to add film genres using Wikipedia categories, and have a questions relating to multiple film genre items. There are many categories in Wikipedia for crosses between genres, e.g. animated adventure films. My assumption is that it is much more useful in Wikidata to assign two separate genres to indicate this, rather than have hundreds of different 'combo genres'. Then someone can easily query for films that are both "animated films" and "adventure films" for example. There are quite a few items that I've found that are not being handled this way (e.g. Ninja Scroll (Q1052531)), so I would appreciate any feedback about this before I proceed with any more large scale editing. Note: I'm aware that there are some 'combo genres' that are accepted genres in their own right, like romantic comedy (Q860626), that will need to be treated differently. NavinoEvans (talk) 14:21, 1 March 2015 (UTC)

agree. Add genres, not Wikipedia genres. --- Jura 16:56, 1 March 2015 (UTC)
Thanks Jura. Any other votes for or against this? It would be good to get some consensus as this choice would imply that we should actually change, for example, every item with genre = comedy horror (Q224700) to have two separate statements instead - genre = horror film (Q200092) & comedy film (Q157443). It certainly makes sense to do so in my eyes - a horror-comedy should surely be returned if you run a query for horror movies (or comedies). NavinoEvans (talk) 21:32, 1 March 2015 (UTC)
Theoretically, you could add all three genres, in your example romantic comedy, plus comedy and romance, so the item will show no matter which filtering you use. Edoderoo (talk) 07:36, 2 March 2015 (UTC)
Yes, I think that's a reasonable option for well known combo genres like romantic comedy. The community would still have to decide which combo genres are notable enough to use though, as we certainly wouldn't want to have as many as are defined in the Wikipedia category system - I will start a discussion in Wikidata:WikiProject Movies to see if they could get help figure out which ones to include. NavinoEvans (talk) 12:31, 2 March 2015 (UTC)
If you're running a search for "horror movie", you should specify your search to include instances of any subclass of horror movie. Since horror-comedy is a subclass of horror movie, a movie with a genre of horror-comedy should be returned by such a search (unless specifically excluded), alongside other non-comedy movies. Since we have an item for horror-comedy, I would have thought it inevitable that some editors will use it as a specific genre, therefore your search should be specified to be able to cope with that. Jheald (talk) 14:03, 2 March 2015 (UTC)
The other advantage of marking its genre as comedy horror (Q224700) is that then it will be automatically included in the corresponding Reasonator page [3] -- something not the case if its genres are marked separately as horror and comedy. Having marked it as a horror comedy, it should not also be marked separately as a horror and a comedy -- it is simply clutter to include additional classes alongside a subclass, when that subclass already implies those classes. Jheald (talk) 14:27, 2 March 2015 (UTC)
Many thanks for that Jheald, very good points. That sounds like a good way to proceed for now. NavinoEvans (talk) 02:59, 3 March 2015 (UTC)
@Jheald, NavinoEvans: Not the best process to take a decision: using one visualization tool to define how you model data is a wrong idea. Reasonator can be outdated and your data structure becomes senseless. You can use constraints to detect movies which are not correctly tagged instead of creating complex query in order to be sure that all items will be correctly analysed in a query. Just imagine that you want to look for movies defined as horror, comedy and SF, how do you do to structure that kind of query ? Just imagine the number of combinations you have: horror, comedy, SF, horror-comedy, horror-SF, SF-comedy, SF-comedy-horror. And all this combinations have to be correctly connected. Snipre (talk) 16:39, 3 March 2015 (UTC)
@Snipre: If I want that combination, I would look for films that had a genre that was in the subclass tree of horror, and a genre that was in the subclass tree of comedy, and a genre that was in the subclass tree of SF. Or if one wanted (Comedy and (Horror or SF)), the query would be obvious. Let people give things the genre they think is most appropriate -- whether it's a hyphenated genre (preferable, if that is a recognised sub-genre, that the film is genuinely part of), or different separate root genres if it's a more random sort of collision. Wikidata queries have the flexibility to be able to cope. Jheald (talk) 16:50, 3 March 2015 (UTC)
Just to add to that, there are so many situations where a query should actually include any item within the subclass tree that this should often be included as a matter of course anyway. If you querying for "visual artists" for example, then you would always need to include any subclass of visual artist. There is no way for the querying user to know whether there are items with more specific instance of (P31) or subclass of (P279) without running another query or otherwise delving into the class tree - hence inclusion of all items within the subclass tree is always needed to make sure (or 'in administrative territory', 'part of' or whatever other hierarchy is applicable for the query). Even just looking for 'thriller films' this is an essential to make sure you get the psychological thrillers, political thrillers etc.
I guess my main point is that this extra step in the query can't be avoided anyway, so simplifying the query should not be used as an argument for defining the ideal structure of the data. NavinoEvans (talk) 01:04, 4 March 2015 (UTC)
There's an interesting question from my point of view that I've been pondering a while, and it's that there are wikis with the more general genres identified in articles already but not with the more specific genres. In this case, comedy horror is a genre that is only present on some 10 wikis. A wiki with a reasonably coded template might expect that they would be able to add a link to an object returned from Wikidata; in this case they would get a red link! Is that something we should care about? --Izno (talk) 17:40, 9 March 2015 (UTC)
This is an interesting point to consider! I wonder whether it might be good that it's a red link as this will speed up the creation of the article in the native language? NavinoEvans (talk) 15:03, 10 March 2015 (UTC)

Is there a way to get the events listed in wikipedia in 2011,2012?

Hi, is there a way for wikidata to retrieve the events listed in wikipedia in 2011,2012,etc.

For instance in http://en.wikipedia.org/wiki/2011, get the

......

January 1 – Estonia officially adopts the Euro currency and becomes the 17th Eurozone country.[2]

January 4 – Tunisian street vendor Mohamed Bouazizi dies after setting himself on fire a month earlier, sparking anti-government protests in Tunisia and later
other Arab nations. These protests become known collectively as the Arab Spring.[3][4]

......


Thanks,

Michael

Try https://tools.wmflabs.org/reasonator/?date=2011 --- Jura 21:49, 9 March 2015 (UTC)

Hi Jura, Well thanks but that's not really what i am looking for. Your query will bring anything from wikidata connected somehow to a date of 2011. I would specifically want the major, international events that have been registered in wikipedia in the specific page http://en.wikipedia.org/wiki/2011. There are about 50 major events per year.

GLAM partners

Hoi, I want to link Wikimedia chapters and GLAM partners.. What is the best way of doing this? Thanks, GerardM (talk) 08:39, 10 March 2015 (UTC)

  • This will probably need a specific property, if there isn't one for things like "Yahoo has a partnership with Microsoft". I agree it would be useful. Doing a comprehensive work for chapters will be hard though; for instance, having a partnership with the ministry or with a single soprintendenza is not the same thing, but the soprintendenza is just a branch of the ministry. --Nemo 10:50, 10 March 2015 (UTC)

Family trees

Hi everyone, I hacked up some code to analyze family trees. The program first takes all items that either have father (P22), mother (P25) or child (P40) from Wikidataquery. This way it finds 55399 items. It goes over all these items and was able to find 11015 subtrees that are not connected. The list of different subtrees that have more than 50 items in them (the item is just the first item in the tree I stumbled on):

The biggest tree by far is the European nobility with 14343 items. I wonder how many of these subtrees can actually be connected. Multichill (talk) 22:09, 9 March 2015 (UTC)

Magnus did a similar analysis some time ago: http://magnusmanske.de/wordpress/?p=232 --- Jura 22:18, 9 March 2015 (UTC)
Interesting, when I search children & parents of Aragorn (Q180322)      I get 206 results. For Isildur (Q206324)      I get 83 for TREE[] and 206 for WEB[] (same family as Aragorn (Q180322)     . Out of curiosity, how did you get these numbers?  – The preceding unsigned comment was added by Popcorndude (talk • contribs) at 01:10, 10 March 2015‎ (UTC).
Yes, I remember reading something about it. Just wanted to know what clusters we have that still could be connected.
The difference can probably be explained by two factors: Data in WDQ might be old and I'm only looking at father, mother and child, not spouse, brother or sister. Multichill (talk) 15:59, 10 March 2015 (UTC)
I also queried only father, mother, and child. Perhaps WDQ is looking from both directions and thus finds connections which are not listed at the other end? Popcorndude (talk) 00:16, 11 March 2015 (UTC)

Can small amounts of data be hosted on Wiki data?

Hi I'm new to the wiki data project and joined inorder to test hosting of small amounts of astronomical observations (< 1 GB) on wiki data. However I don't seem to be able to create a subpage with the data (in ascii), description and the python plotting programs associated with the data.

So should I look elsewhere for hosting the data, or can this be one on wikidata?

Thanks

Glen (chemistry@protonmail.com) --unsigned comment by User:Astrochemistry

Greetings and welcome to the project. I guess the first question that comes to my mind is, does an article for these observations exist in one of the other Wiki's? Reguyla (talk) 20:18, 10 March 2015 (UTC)
I love astronomical data, but I think you can assume that Wikidata almost certainly would not be a good host in this case. This is because Wikidata is intended for the kind of data that is well-established in reliable sources. Fresh observational data would not have gained the broad acceptance that is expected/required by the Wikidata community.--Anders Feder (talk) 01:37, 11 March 2015 (UTC)

What is a global account and where can I find a local administrator?

This weekend I suddenly got a global ban on a machine I am using to maintain the ProteinBoxBot. I was in the process of updating human gene items and correcting some constraint violations that were caused as a result of this process when the ban initiated. When I asked clarification - as suggested in the api output = on the reasons why, I got the following respons: 

   Thank you for your email.
   Unfortunately, your bot has no global account <https://meta.wikimedia.org/w/index.php?title=Special%3ACentralAuth&target=ProteinBoxBot> and it appears that you do not operate it on    en.wikipedia.org. Without a global account, we cannot set specific global rights such as global IP block exemption.
  When the SUL finalization happens, you will probably be prompted to create a global account. But for now, no such thing appears to be feasible given the great number of edits of the accounts under the username "ProteinBoxBot". In the meantime, we recommend that you have a local administrator put a local exemption on your IP xxx.yyy.zzz.aa. Have them use this page <https://www.wikidata.org/wiki/Special:GlobalBlockWhitelist>.

Where can I contact a local admin and what is a global account? Andrawaag (talk) 17:29, 9 March 2015 (UTC)

@Andrawaag: This is kind of weird, since I believe all accounts created on Wikidata should be global by default, since Wikidata was created after they took effect. Nonetheless, to fix this you should try going to Special:MergeAccount when logged in as the bot (or possibly Special:CentralAuth) and follow the steps. Jon Harald Søby (talk) 18:01, 9 March 2015 (UTC)
@Jon Harald Søby: I have gone through the steps suggested and got a confirmation that the account is unified. However, my IP address is still blocked. What steps should I take to get unblocked? As said before, I am not aware of doing anything out of the ordinary. So if this indeed is considered abuse, I first apologise for any inconveniences I might have caused and I am looking forward to instructions on how to prevent this.Andrawaag (talk) 12:50, 10 March 2015 (UTC)
  pywikibot.data.api.APIError: failed-save: Your IP address has been blocked on all wikis.
  The block was made by Vituzzu (meta.wikimedia.org).
  The reason given is Open proxy: + abuse from aaa.bbb.ccc.ddd

  * Start of block: 21:30, 6 March 2015
  * Expiry of block: 21:30, 6 March 2020
@Andrawaag: Send me an e-mail with your IP address, and I will add it to Wikidata's whitelist. :-) Jon Harald Søby (talk) 12:53, 10 March 2015 (UTC)
@Jon Harald Søby: Email has been sent. The bot runs in the cloud, meaning that IP addresses can change. I sending an email sufficient to keep track of which IP addresses are whitelisted?188.189.85.82 11:53, 11 March 2015 (UTC)
@Andrawaag: Yeah, sorry, I wasn't able to unblock the IP here because the original block was a range block (blocking several IPs at once). What you should do is post a request (with a link to this section I guess) on m:Steward requests/Global permissions#Requests for global IP block exemption, that should solve the problem. Jon Harald Søby (talk) 12:21, 11 March 2015 (UTC)

Invalid token

I seem to be getting an invalid token error when trying to update some of the properties in Wikidata today. For example, I tried to update [4] with an English description of Wikimedia category and it kicked this error. Reguyla (talk) 20:19, 10 March 2015 (UTC)

Log out, restart the browser, log in, and make a burn offering of a imaginary chicken to the internet gods! -- Innocent bystander (talk) 20:30, 10 March 2015 (UTC)
Apparently the gods aren't in the mood for chicken tonight now I get another error "We are experiencing technical difficulties. Your "save" could not be completed." Oh well, I'll try again later I suppose. Reguyla (talk) 23:41, 10 March 2015 (UTC)
The hangovers he whose name should not be pronounced can never be overestimated! Both me and my bot got "invalid tokens" yesterday. I think I have resolved the problem with the bot (repeated logins) but the manual problems are more difficult to solve. -- Innocent bystander (talk) 07:19, 11 March 2015 (UTC)

Input needed for watchlist integration improvements

Hey folks :)

Data quality and trust is what we're currently concentrating on in the development around Wikidata. A big part of that is improving the integration of Wikidata in the watchlist of Wikipedia and other sister projects. I just opened a page to collect input on how we can improve it. Please help me with spreading this to the Wikipedias and other sister projects. Wikidata:Watchlist integration improvement input --Lydia Pintscher (WMDE) (talk) 15:49, 11 March 2015 (UTC)

Merge problems

Who can help ? I can't merge German article de:Johann Graf (Unternehmer) (Q19519288) with English article en:Johann Graf (Q1694243). 92.77.57.31 16:19, 11 March 2015 (UTC)

Done. --Marek Koudelka (talk) 16:33, 11 March 2015 (UTC)

Stop all the bots they don't have the AI needed

I am worried that the bots have created a mess (huge mess?) (10% mess?) with wrong interwiki links. Also cleaning up the mess manually is too laborous. I haven't really paid attention to interwiki linking, it did not need any attention, but now it needs a attention. --Pasixxxx (talk) 20:26, 11 March 2015 (UTC)

Which bots? What mess? Do you have examples --Izno (talk) 21:07, 11 March 2015 (UTC)
Are you talking about User:Mjbmrbot? I also noticed that one to create nonsense with Wikivoyage links, even merging items which are already marked as "do not merge". I cleaned up many of the mess last night, maybe 80% were of the merges were wrong - but haven't checked the plain adding of interwiki of Wikivoyage items not yet having items. I at least added some statements to the still empty Wikivoyage items wrongly merged, so it can be easier avoided in future. Example merge was GR 20 (Q14209685) with GR 20 (Q262374), or the wrong interwiki to Surat Thani (Q240463) [5]. Ahoerstemeier (talk) 09:48, 12 March 2015 (UTC)

Merge problems

Can someone merge en:George P. Livanos (Q5543123) with German article de:George Livanos (Q1507761) ? 92.77.57.31 21:16, 11 March 2015 (UTC)

  Done. Jared Preston (talk) 22:17, 11 March 2015 (UTC)

Another question

Hi,

Do you know a bot who can check an entitle wikipedia and automatically delete the interwki on articles when those interwiki links are on wikidata ? I remember that a russian have a bot like that but I don't remember this name.

My problem is that few thousands articles was created on wp:fr on 2014, about polish settlements, with a interwiki links on there articles. Those articles was automatically linked on wikidata, but the interwiki on the articles wasn't delete. Ex : fr:Walentynów (Lipsko). --Nouill (talk) 16:03, 8 March 2015 (UTC)

Interwiki links I'm surprised if a bot doesn't come by to delete them. If not, it doesn't actually hurt for them to be there: they are just unnecessary. Are the interwiki links all accurate as far as you can tell? —Justin (koavf)❤T☮C☺M☯ 16:05, 8 March 2015 (UTC)
Addshore's Addbot and Emaus' EmausBot. Matěj Suchánek (talk) 16:18, 8 March 2015 (UTC)
My bot is and has been inactive for a while now! ·addshore· talk to me! 11:10, 10 March 2015 (UTC)
As Justin says, usually these interwiki links are not a problem, just unnecessary, but the inline links override the Wikidata links, so when there are changes in the Wikidata item (like a title change of the related article), these will not be reflected on the article. Therefore it is still useful to remove the superfluous links. Sometimes inline links are being used to link articles of a different item, because the same topic is split over several items. There is not yet a good solution for this problem. Bever (talk) 20:55, 12 March 2015 (UTC)

Wrong interlinking?

Q6607 (guitar) has a whole slew of interwiki links to wikibooks which happen to be about guitars: [6]. Can we agree that this is an inappropriate application of interwiki linking? For instance, the guitar item is linked to wikibooks:fr:Apprendre la guitare ("Learn [to play] the guitar"). What if someone made a French wikibook titled "How to build a guitar". Which of the two wikibooks should now be linked to the completely generic concept of "guitar"? If the answer is "let's pick one of the two at random", then what value does the link even have?--Anders Feder (talk) 01:15, 11 March 2015 (UTC)

I can agree that it's inappropriate for Wikidata-sitelinking, but I think each project should feel free to use old style interwiki [[w:Guitar]] <-->[[b:How to build/eat/play a guitar]] whenever they like. -- Innocent bystander (talk) 07:10, 11 March 2015 (UTC)

I went ahead and killed those links.[7]--Anders Feder (talk) 22:56, 12 March 2015 (UTC)

Help

must divide link about Athinaikos A.C. (the multisport club) and link about Athinaikos F.C. (football team of the club). Some there are here (Q16329826) and other there are here:(Q16329826). How will it become;

should be   Donehttps://www.wikidata.org/w/index.php?title=Q301074&action=history Oursana (talk) 23:11, 12 March 2015 (UTC)

Collections of entries

Hi all,

I've started a rough index of "things which are on Wikidata" - the idea is for this to be a list that sits alongside Wikidata:Notability and helps give some examples of:

a) things which are all in Wikidata (eg every MEP);
b) things which will all be in Wikidata at some point (eg every species, every populated place);
c) things which will not all be in Wikidata (eg "every person", "every building")

It's provisionally at Wikidata:Comprehensive groups of items but a better name would help ;-)

If you know of something that would fit category A or B - especially if you know of something we definitely have covered already! - please do add it. Andrew Gray (talk) 09:19, 12 March 2015 (UTC)

I suggest these should be added to Wikidata:Notability/Exclusion criteria and Wikidata:Notability/Inclusion criteria. Filceolaire (talk) 23:46, 12 March 2015 (UTC)

Merge problems again

Hello, Could someone look into merging commons category: "Category:Paris 15e arrondissement" and "Catégorie:15e arrondissement de Paris". Also, Wikimedia links don't show on the left side in "other projects" on this Wikipedia page: "15e arrondissement de Paris" , on the English page even Wikivoyage does not show up. Thank you for any help.--DDupard (talk) 10:25, 12 March 2015 (UTC)

item is perfectly linked and languages show up on the left. on the English page even Wikivoyage does not show up I do not understand, wikivoyage is given on wikidata --Oursana (talk) 22:27, 12 March 2015 (UTC)

Thank you Oursana for looking into it, here is the page with no wikivoyage or wikimedia on the left-hand side: en:15th arrondissement of Paris

Merge problems

Can someone merge German article de:Georgios Ikonomou (Q17321904) with english article en:George Economou (Shipbuilder) (Q16062040) ? 92.76.115.31 13:59, 12 March 2015 (UTC)

  Done by User:Mjbmr Oursana (talk) 22:35, 12 March 2015 (UTC)

New property proposal to help solve the Bonnie and Clyde problem

See Wikidata:Property_proposal/Generic#Redirect_to_(if_no_sitelink_available_in_a_language) for a proposed new property to help solve the Bonnie and Clyde problem. Filceolaire (talk) 00:22, 13 March 2015 (UTC)

SUL finalization update

Hi all, please read this page for important information and an update involving SUL finalization, scheduled to take place in one month. Thanks. Keegan (WMF) (talk) 19:45, 13 March 2015 (UTC)

Adding P31 to items without any properties...

Before, when I edited an item with no property at all, when adding a property, the instance of (P31) was always the top choice, which made it easy to add P31.

Now, I have to click on the Property cell and manually type the name of the property, which is much longer, especially when editing a long list of items...

Could you please help me with that ? did something change in the behaviour of the interface ? I don't remember changing my preferences... :/ --Hsarrazin (talk) 11:30, 14 March 2015 (UTC)

I remember that the entity suggester was working without any current statements, but that was only when it was just introduced last year. But, I   Support this so new users don't forget to add the most useful property on Wikidata. Sjoerd de Bruin (talk) 12:00, 14 March 2015 (UTC)
I support too the possibility to display automaticaly the P31 property (without entering any text in the property field). Do you know where one has to go for asking such modification? Pamputt (talk) 12:12, 14 March 2015 (UTC)
I'll look into it. Do you know when it changed roughly? --Lydia Pintscher (WMDE) (talk) 12:35, 14 March 2015 (UTC)
I do not remember precisely, maybe a couple of weeks. Pamputt (talk) 13:37, 14 March 2015 (UTC)

Maybe  , but suggestions should include subclass of (P279) and part of (P361), too. --Diwas (talk) 00:12, 15 March 2015 (UTC)

Wikidata weekly summary #149

 
Here's your quick overview of what has been happening around Wikidata over the last week.

Allow the basic templates in edit summary

Is there a way to allow the resolve of templates like Q and P (maybe C and St) in the edit summaries?--Diwas (talk) 17:28, 8 March 2015 (UTC)

I think you just need to write [[Q1]] and generally it gets resolved. --- Jura 17:58, 8 March 2015 (UTC)
{{Q|1}} gives you Universe (Q1) and {{P|1000}} gives you record held (P1000). Note that with this template the item and property names are changed each user sees the name in the default language of their browser. Filceolaire (talk) 23:59, 8 March 2015 (UTC)
@Filceolaire: That doesn't work in edit summaries. --- Jura 05:08, 9 March 2015 (UTC)
Thank you both. Until all users will get learned all Item IDs and property IDs properly, it would be helpful to show labels in edit summaries. Yes it will be a big task or a bundle of it, but wikidata is a multilanguage project, more than commons. It would be helpful particularly with section headers in talk pages, project pages and maintenance pages to have multilanguage labels in item related section headers and related edit summaries shown on watch lists and histories. --Diwas (talk) 01:15, 11 March 2015 (UTC)
@Diwas: Watchlists already have meaningful labels for items and properties. For histories, you can use PropertyNames.js; despite the name, it also handles items. --Ricordisamoa 13:38, 15 March 2015 (UTC)
Thank you @Ricordisamoa: that is helpfull. But there in the first row you can't see the names and the → link isn't finding the section. Same there and if you see that on your watchlist. --Diwas (talk) 19:58, 16 March 2015 (UTC)

Please help test next step for header redesign

Hey folks :)

We've been working on completely rewriting the header section and moving it towards the new design. We're not there yet but the next step is ready for testing and then rolling out on Wikidata. Please go ahead and test it thoroughly at http://wikidata.beta.wmflabs.org/wiki/Q27946. This will allow you to collapse the in other languages box for example and adds a hint about how to configure the displayed languages. The remaining known issues are tracked at phabricator:T75654.

In addition to the new header the next deployment will bring a lot of under-the-hood changes and bug fixes. The most relevant changes for you are:

  • we made the diff for time values more meaningful
  • we fixed a lot of bugs in the time datatype
  • edit links are no longer cached incorrectly based on the users permission (This lead to users sometimes seeing edit buttons on pages that they could not edit and no edit buttons on pages that they could edit.)
  • we fixed some issues with propagating page moves and deletions on the clients (Wikipedia, etc) to Wikidata
  • we corrected an issue where you would see new data in the old part of a diff (This affected qualifiers mainly.)
  • the sitetointerwiki gadget now also works on diff pages
  • the precision is now detected correctly when entering a quantity in scientific notation
  • we added mailto as an accepted protocol for the URL datatype

Unless you find big issues during testing we plan to deploy this on Wikidata on 24th of March. --Lydia Pintscher (WMDE) (talk) 12:39, 14 March 2015 (UTC)

triangles to move claims or qualifiers are still missing Oursana (talk) 13:46, 14 March 2015 (UTC)
header redesign. Sjoerd de Bruin (talk) 13:48, 14 March 2015 (UTC)
Where do I leave comments? In general, it really looks awesome! Much better than the current version. My only quibble is in the head that the aliases are above the description. For me, the label and the description form a unit, whereas the aliases are mostly for alternatives to find the item. So having the label and description broken by the aliases is weird. I would hope that this semantic unity would trump aesthetic considerations (i.e. 'the long list of aliases looks weird under the descriptions'). It's also the order in the entity selector (as conveniently illustrated in the section above) But other than that, wow, this is really neat! --Denny (talk) 16:12, 16 March 2015 (UTC)
Yay! About the aliases: We're currently fiddling around with options for that in phabricator:T87579. Some of those will make this obsolete. Comments there welcome. --Lydia Pintscher (WMDE) (talk) 16:28, 16 March 2015 (UTC)
Commented there. You seem to be going even further away from how labels, descriptions, and aliases are meant to work. Please don't do that. --Denny (talk) 16:57, 16 March 2015 (UTC)
Oh, and one wish, but that's more like another feature request: one button in the language box that says 'show all' and displays all labels, descriptions, aliases, in all languages. --Denny (talk) 16:16, 16 March 2015 (UTC)
Agreed. That'd be phabricator:T92759. There is a bit of a conceptual issue lined out in the ticket. Comments there welcome. --Lydia Pintscher (WMDE) (talk) 16:28, 16 March 2015 (UTC)
Yay! Quite happy about this! --Denny (talk) 16:57, 16 March 2015 (UTC)

Broken merges

A few days ago, someone merged the items for two different people together - see here. I've corrected all the properties, but there's a lot of labels/aliases/descriptions that got merged across meaning that Q640096 has the wrong name/description in pl, de, es, nl, fa, nn, nb, da, sv, & fr. I can't figure out how to remove them. Any advice? Andrew Gray (talk) 13:05, 14 March 2015 (UTC)

If you go to the gadgets tab in the preferences, there's a "labelLister" gadget which adds a "Labels list" tab where you can edit labels in other languages. Is that what you're looking for? - Nikki (talk) 16:28, 14 March 2015 (UTC)
Looks great, thanks! Now all I need to do is turn off the damned "British English" autosuggest box ;-) Andrew Gray (talk) 22:23, 16 March 2015 (UTC)

Request for review

Item Q19594883 seems to be self-promotion (with some statements that don’t make any sense). What’s the correct course of action? And in general, is there a better place for “I noticed this while patrolling recent changes, need assistance”? —DSGalaktos (talk) 16:22, 15 March 2015 (UTC)

Deletion requested. The entry does not fulfill the notability requirement. Pichpich (talk) 16:27, 15 March 2015 (UTC)
to request deletion add the item to Wikidata:Requests_for_deletions. There is a tool you can add to your top toolbar via your preferences to help with this. Filceolaire (talk) 22:48, 16 March 2015 (UTC)

sex or gender (P21) via Virtual International Authority File (Q54919)

There is a section "about" [8], where sex or gender (P21) is filled in - but there is no source. Does anybody know where the entry comes from? Thank you, Conny (talk) 11:27, 16 March 2015 (UTC).

This will come from one of the contributing libraries - however, very few record gender in their cataloguing data. I believe most times it appears it's come from the BNF in France, but I could well be wrong. Andrew Gray (talk) 22:20, 16 March 2015 (UTC)

titel rules

How to handle titel Squirrels of the World (Q19597701)/Squirrels of the world/Squirrels of the World/Squirrels Of The World? Thank you, Conny (talk) 16:50, 16 March 2015 (UTC).

It's the name of a book in English so follow the capitalisation used by the author and publisher. Filceolaire (talk) 22:18, 16 March 2015 (UTC)

business vs. company

Right now, a business is not a subclass of a company (it's a separate branch of organization: organization->company vs. organization->operation->business) so something that is a business is often not a company, and causes constraint violations with these properties, which require companies:

AFAIK "business" is not used as a constraint at all.

The most information I have found about the distinction is here. If there's more anywhere, please let us know and add a link in an obvious place.

This leads me to ask if the subclass relationships of business and company should be altered (I vote yes), or if some other solution should be pursued.

This causes widespread constraint violations when someone makes a specific entity an instance of a subclass of business which is often not a company. It includes some with "company" in the name, like holding company (Q219577) and subsidiary (Q658255), which are not companies. Seemingly obvious companies like Google (Q95) are not companies (it's a multinational corporation (Q161726).)

Faced with an apparent logical contradiction, users may be bold and alter the subclass relationships, and if their edits are undone, they may be upset and ask why, or more likely just decide to give up and leave. When automatic warnings for constraint violations come online, I imagine this might become a problem then too. --Haplology (talk) 14:58, 15 March 2015 (UTC)

My understanding is that "Company" is a particular way of legally incorporating a business with benefits (limited liability) and duties (to file accounts and register a place of business and names of directors). This suggests that the appropriate arrangement is "company->business->organization->operation" but I'm not going to edit this - I leave it for who ever is more interested in the Wikidata "High Level Ontology" (HLO).
At the moment there doesn't seem to be enough interest in our HLO to support a wikiproject so I think users should be bold about unravelling this with any necessary changes to constraints being editted afterwards. You could publicise the change here too in case someone does turn up with an interest in our HLO. Filceolaire (talk) 23:03, 16 March 2015 (UTC)
Thanks for your reply. I'm concerned because the HLO is the only mechanism that exists right now to manage data quality, and I'm afraid that there's a hostile environment to changes to it combined with an absence of discussion about it. This was my experience with being bold. I'm not upset about that edit individually, but if it's at all symptomatic of the general situation, we have a big problem. --Haplology (talk) 01:01, 17 March 2015 (UTC)
Usually when I get reverted like that I go talk to the user. Given our reliance on IW links, usually a "that's the wrong concept" revert stems from some IW link not matching particularly well in one of the languages with your own native language. --Izno (talk) 02:37, 17 March 2015 (UTC)
Filceolaire "At the moment there doesn't seem to be enough interest in our HLO to support a wikiproject..." Wrong, we have plenty of discussions (see here as example) but the problem is be able to meet enough persons with some knowledge in ontology. The wikiproject will be a real help to centralize discussions. Snipre (talk) 13:32, 18 March 2015 (UTC)

Help Merge

Item 15th arrondissement of Paris (Q191066) and Category:15th arrondissement of Paris (Q7470156), these are about the same topic. Thanks.--DDupard (talk) 17:52, 16 March 2015 (UTC)

@DDupard: an item and its category are not to merge, but to link using category's main topic (P301)   and topic's main category (P910)  . TomT0m (talk) 18:34, 16 March 2015 (UTC)
@TomT0m:, Merci TomT0m pour la réponse, mais je ne sais pas ce qu'il faut faire, l'objectif étant d'une part de faire apparaitre le lien Wiki commons, dans la colonne de gauche des pages sur le sujet 15e arrondissement de Paris , 15th arrondissement of Paris, etc, et d'autre part comment trouver un titre "trans-langue" ou quelquechose comme ça afin que tout se rejoigne..... Enfin , des explications me seraient nécessaires, Thanks--DDupard (talk) 19:10, 16 March 2015 (UTC)
@DDupard: The sidebar on the left is for the second of those uses, not the first -- i.e. it is meant to show pages that directly correspond to the current page. So by default only the page Category:15th arrondissement of Paris (Q7470156) will get a sidebar link to the Commons category.
(One can specify a relationship between an article-like page like 15th arrondissement of Paris (Q191066) and a Commons category using Commons category (P373). In future it is likely that this will allow 1-to-1 sidebar links from article pages to directly corresponding Commons category pages, eg from 15th arrondissement of Paris (Q191066) to c:Category:Paris 15e arrondissement; but (I think) this is not yet available).
For other pages, eg pages on buildings and streets in the 15th arrondissement, the best solution is to create a template on fr-wiki to go on the bottom of the page of all such articles. Such a template can include anything you want -- including a link to the appropriate Commons category page. Jheald (talk) 20:06, 16 March 2015 (UTC)
@Jheald:, Ok, Thank you Jheald, for looking into it, --DDupard (talk) 20:44, 16 March 2015 (UTC)

Solved, Thank you.--DDupard (talk) 06:55, 17 March 2015 (UTC)

Gender from names

Recently, we finished matching all entries in the Oxford Dictionary of National Biography to Wikidata (using Oxford Dictionary of National Biography ID (P1415)). I've been using their database to sanity-check our own gender information - of the ~50,000 entries where both had gender data, about 30 had the wrong gender on Wikidata. An error rate of >0.1% is pretty good, but there's a pattern, and almost all of them seem to be cases like Shirley Brooks (Q7498705) - men with a "female" first name, or women with a "male" first name. But not all generally-male names were always male, and some (there's a few Jeans) are male in one language and female in another.

I suspect these have all been "auto-gendered" using a script at some point in the past, and it's a reminder to be careful about what we can automatically assume, especially with gender. Perhaps it might be worth double-checking sex or gender (P21) on people with unclear first names? Andrew Gray (talk) 22:30, 16 March 2015 (UTC)

You might like this item for qualifiers: male given name borne by a female (Q18220911). --- Jura 23:00, 16 March 2015 (UTC)
I'm not convinced this is a perfect solution. It won't address the "Jean" problem, for one thing! More generally, there are a lot of names this can apply to, and tracking them all will be hard work. Andrew Gray (talk) 12:32, 17 March 2015 (UTC)
It needs to be read in conjunction with the items used for given names (and its P31 - "female given name" or "male given name" or "unisex name"). For "Jean", there are already two distinct items so the item would rarely be used. --- Jura 12:38, 17 March 2015 (UTC)
@Andrew Gray: could you post the list of 30, please? I'd be interested to track down where the addition of a gender occurred. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 00:07, 17 March 2015 (UTC)
I'll get this up tonight, probably after fixing them all! Had hoped to do it yesterday but ran out of time. Andrew Gray (talk) 08:59, 17 March 2015 (UTC)
Okay, did the list in my lunchbreak :-). Will correct them tonight (if you do them before then, please note).
Female in ODNB but male in Wikidata - all now fixed

(For reference, there are 161 Jeans in the ODNB, of whom 54 are female. This one definitely seems to be a problem for us.)

Male in ODNB but female in Wikidata - all now fixed

I've removed a few where it was a real mismatch or otherwise not applicable, giving 27. Andrew Gray (talk) 13:02, 17 March 2015 (UTC)

  Question (@Jura1: especially:) For usages that shifts in time, isn't it enough to use something like :

Then a query can extract those who stopped to be, for example, given only to boys. TomT0m (talk) 16:23, 17 March 2015 (UTC)

I'm not sure we can get a reliable cut-off date, and of course a lot of people simply don't have birthdates listed yet. Andrew Gray (talk) 18:06, 17 March 2015 (UTC)
@Andrew Gray: In Wikidata dates supports uncertainty, so the date does not really to be precise, something as vague as a century can be a date. TomT0m (talk) 18:20, 17 March 2015 (UTC)

A Swedish farm name (Q10512538) can always be applied to all genders, even if the nature of the name itself can look male/female. The reason is that the name is connected to a place, and the name of the place is connected to a person. While the place have no gender, the namesake of the place has. -- Innocent bystander (talk) 16:43, 17 March 2015 (UTC)

en:WP says that "Before the publication of the novel Shirley by Charlotte Brontë in 1849 Shirley was an uncommon, but distinctly male name and would have been a very unusual name for a woman." You need to be careful about guessing sex from the name alone. Filceolaire (talk) 22:18, 17 March 2015 (UTC)

Assigning genders to names is a bad idea for two reasons:
  1. The gender of names depends on the fads of the day. For example, Ashley used to be an exclusively male name and is now an exclusively female name. Sometimes these changes can take place in a matter of a few years, for example, in the 70s, Ryan was exclusively a male name. Then in the 80s there was a surge of women named Ryan. Unfortunately, there is no reliable global data on name usage by gender, so even adding date ranges is merely an educated guess (and likely to be hopeless biased towards U.S. usage).
  2. The gender of names depends on the culture and language. For example, until recently, Camille was a male name in French and a female name in English. (Now it is unisex in French.)
I don't think we have enough manpower (or data) to keep up with problem #1 and I don't know if we have enough expertise to keep track of problem #2. In most cases, the use of these properties will simply be one person's opinion at one point in time, not objective long-lasting metadata. Kaldari (talk) 23:37, 17 March 2015 (UTC)
I am not sure it's a bad idea. The bad idea is to add statements to items about persons based on that idea. Here, there is a name-law who tell which name can be given to each gender. We can make statements based on that. But since people easily can do otherwise than what the law tells you, it's not a good idea to add statements to person-items based on it. -- Innocent bystander (talk) 06:10, 18 March 2015 (UTC)
SOURCES. Most of the problems come from the fact that people has adding data WITHOUT any reference, just using their knowledge and thinking that is common knowledge. Snipre (talk) 08:59, 18 March 2015 (UTC)

Citation for Wikidata

We are writing the second edition of our AI textbook (http://artint.info/) and would like to reference Wikidata. Is there a paper to refer to that overviews Wikidata? It is much better to refer to an overview than just to the website (which is a bit overwhelming for our audience).

@Lydia Pintscher (WMDE): We need a credit page and a official publication about Wikidata. Snipre (talk) 13:13, 18 March 2015 (UTC)
We have Wikidata:Introduction. Is it what you need ? Snipre (talk) 13:38, 18 March 2015 (UTC)
You probably want to cite Wikidata: A Free Collaborative Knowledgebase (Q18507561). --Lydia Pintscher (WMDE) (talk) 13:48, 18 March 2015 (UTC)
--- Thanks! The CACM paper was exactly what I want. I was surprised there was nothing I could find pointing me to the paper. The main page seems to assume everyone knows what Wikidata is about.

Next office hour

Hey folks :)

We'll be doing another office hour on 31st of March at 16:00 UTC. See here for your time zone. It'll be in #wikimedia-office on Freenode IRC. Denny and I will be there to answer questions about the Freebase move. I'll give a short update on recent developments and I was asked to put the RfC about admin inactivity criteria on the agenda as well. Hope to see many of you there!

Cheers --Lydia Pintscher (WMDE) (talk) 16:30, 18 March 2015 (UTC)

Merge problems

Can someone merge German de:Kategorie:Jesch-Atid-Mitglied (Q17203774) and English en:Category:Yesh Atid politicians (Q7809003) ? 178.11.14.88 16:59, 18 March 2015 (UTC)

  Done Pamputt (talk) 17:02, 18 March 2015 (UTC)

Can someone merge German de:Kategorie:HaBajit-haJehudi-Mitglied (Q16659773) with English en:Category:The Jewish Home politicians (Q7714698) ? 178.11.14.88 17:11, 18 March 2015 (UTC)

Done --Marek Koudelka (talk) 17:17, 18 March 2015 (UTC)

Can someone merge German de:Kategorie:Konservatives Judentum with English en:Category:Conservative Judaism ? 92.74.60.4 18:58, 18 March 2015 (UTC)

OK, done. --Marek Koudelka (talk) 19:11, 18 March 2015 (UTC)

Search function

I haven't found a discussion about the search function being down for a few weeks now. Is there any timeline for when it will be fixed? I'm not talking about the auto-search (typing in the field and getting the list of matches) but the regular search. It seems that if someone were to search for an existing entry and clicked "Enter" instead of looking at the match list they'd be told that it didn't exist and go to create a duplicate. Thanks, Hazmat2 (talk) 15:25, 19 March 2015 (UTC)

Hmm, I never had any problems with the search, and so right now, I am able to enter e.g. "Albert Einstein" in the search field, press enter and get a list with 93 results. --YMS (talk) 15:32, 19 March 2015 (UTC)
@Hazmat2: It also always worked fine for me. Double-check your default search settings: in the search results, click "Advanced" below the search button and check your default settings. To change them permanently, click the checkbox "Remember selection for future searches" at the bottom of that frame. -- LaddΩ chat ;) 17:07, 19 March 2015 (UTC)
Strange, leave it until the day I mention it for it to work for me again. As an aside, I do fiddle with my search settings (very) often, but because of that it's the first thing I checked. Nonetheless, it works again, so I'm not going to worry about it. Thanks for getting back to me! Hazmat2 (talk) 02:14, 20 March 2015 (UTC)

Extortion or Blackmailing?

Can someone please me to clearify the distinction between extortion (Q6452087) and blackmail (Q34284) and sort out the messy interwiki? /ℇsquilo 19:37, 16 November 2014 (UTC)

I am not a lawyer and I did not unterstand languages. I have looked at de,fr,en,sv: Google translator says that fr and de Chantage blackmail (Q34284) subclass of (P279) extortion (Q6452087) Chantage is Extortion with a distrubution of informations about the victim of the Chantage. w:sv:Utpressning seems to be similar to fr/de Chantage, but I am not sure if the sv:Utpressing meanings includes extortion with other harm like violence too. blackmail (Q34284) seems to be splitted. I guess en Blackmail ist not equal to fr/de Chantage. --Diwas (talk) 21:24, 16 November 2014 (UTC) PS: w:it:Estorsione blackmail (Q34284) seems maybe similar to en:blackmail blackmail (Q34284) or to en/de/fr/... extortion (Q6452087) but not to fr/de chantage blackmail (Q34284).
That's a general problem with legal terms, as legal definitions of offenses vary a lot between countries, and so do legal terms. What is covered by one legal term in one country could be split into two distinct offenses in another. So in many cases, it might have been okay to link articles to articles about similar legal concepts in other languages, but from a database POV, it's probably not a really good idea to try to somehow stuff them all into one data set. --95.90.51.227 05:33, 19 November 2014 (UTC)

Sometimes the words "extortion" and "blackmail" are used interchangeably in colloquial English. But more precisely, "extortion" (as in extortion (Q6452087)) is generally any demand with coercion (Q325980): it can use any threat; it might have the threat of violence or it might threaten in some other way. I would say that extortion (Q6452087)subclass of (P279)coercion (Q325980), at least in English. Some kinds of extortion (in the general sense, not necessarily legal sense):

  • "Blackmail" (as in blackmail (Q34284)) is more specific extortion: The threat is to spread embarrassing information, and the information might be true or not.
  • racket (Q1283181) is almost always extortion (Q6452087), though sometimes the victim does not know it: The victim might know that the criminal is involved with the source of the problem, or the victim might not know, and instead think that the problem is caused by something else.
  • ransom (Q1414572): a person or thing is kept (usually kidnapping (Q318296) or theft (Q2727213)) until demands are satisfied

There are many types of extortion (and similar behavior) listed at en:Extortion#Similar crimes. But of course that is for English. Maybe coercion (Q325980) is the broadest term. (Also, there is duress (Q1192757), which is a legal defense, saying that the perpetrator was a victim of coercion (Q325980).) I have seen instances on Wikidata where English-language concepts are subclasses of German-language concepts that have no English-language label; that may be useful here, but requires someone fluent in more than one language. --Closeapple (talk) 14:20, 9 December 2014 (UTC)

"Erpressung" in German is something the perpetrator does to achieve a certain reaction of the victim. The perpetrator threatens to spread an information or to harm the victim or a relative. The perpetrator holds a child of the victim hostage to achieve money from the victim, this is a typical case of Erpressung. Blackmailing has a colloquial equivalent in German "Anschwärzen" = denigrate. There is no specific legal term for it, but if it comes to court it is "Verleumdung", however the perpetrator must tell lies or twist the truth to get convicted. You can not get convicted by telling the truth about a secret affair or something similar. Erpressung can represent the threat of spreading unwanted Information, Verleumdung represents the action of spreading lies an twisted information. The main difference between a case of Erpressung and a case of Verleumdung is the threat and the action.--Giftzwerg 88 (talk) 13:48, 19 December 2014 (UTC)
Blackmail is such an ugly word... (TV tropes) -- Jheald (talk) 13:54, 19 December 2014 (UTC)


adding comment to get this archived at some point... --LydiaPintscher (talk) 09:55, 22 February 2015 (UTC)

Why?

Why has the GUI only gotten 100 times worse each time it has changed? Wikidata has an unfamiliar format for anyone who is familiar with editing Wikipedia pages - there is no edit tab at the top. In the original format, there was a "section" for each item, with an edit tab for each, and a blank section at the bottom to add a new item. That worked well, once you figured out how to use it. The only difficulty was if you tried to add something that was in its own dummy item, which you had to delete first, or merge.

 

The second version took away the one section per item and consolidated the edit link to one link that unintuitively had to also be used if you wanted to add a link. That version made the time required to add links about 100 times as long. To save time though, this link can be used, just substituting the Q number you want to edit for Q42. https://www.wikidata.org/wiki/Special:SetSiteLink/Q42

It would be extremely helpful to place a link to that page with the link labeled "add" at the bottom of the Wikipedia section of every page, populated of course with that page's Q number, as it is about 100 times faster to use than the present method of adding new links. The only thing you have to remember to do is put in the full wiki name, like enwiki, instead of just en.

 

I was not able to find a screenshot of that iteration.

Which brings us to the current abomination. Which removed the most important functionality of all, for me, the ability to copy and paste in tab separated columns. The second worst feature of this version is to move the GA/FA to the right of each article name instead of to the left where they have always been. And the third, using side by side columns.

 

So the question is, why does each new GUI become worse than the last one, instead of better? Like this one. Wixty (talk) 00:54, 9 February 2015

Redirects from

Acapulco Wikipedia page (Q2378103) is redirected from Acapulco, Mexico

Is it possible to discover this data based on Wikidata dumps \ API? If so how?

Otherwise, any chance it could be consider to add it ass a feature in the future?  – The preceding unsigned comment was added by 92.74.76.120 (talk • contribs).

Landing pages for partners

Hey folks :) In my job I get approached by institutions, companies and so on about getting involved in Wikidata. Over and over again I have to explain the very basics like "no you can not just dump your data here" or "yes it will be open to editing by anyone" or "yes you can use our data but it'd still be awesome if you give credit". Since me telling this over and over again does not scale any longer I started two landing pages - one for partners who want to use data from Wikidata and one for partners who want to donate data to Wikidata. They're in my user space at User:LydiaPintscher/Donating data and User:LydiaPintscher/Using data. This should also be a great resource for anyone else who gets approached about such topics since I am sure I am not the only one. It'd be lovely if you would help me expand them and make sure they reflect community consensus. Edit away. Depending on feedback I'd love to move them to the project namespace next weekend. --LydiaPintscher (talk) 09:58, 7 March 2015 (UTC)

Awesome? Terrible? Don't care? Would love some feedback or even better edits ;-) --LydiaPintscher (talk) 12:47, 14 March 2015 (UTC)
I might suggest one page, something like "Involve your organization" or similar (that's a little too DO SOMETHING for a good name IMO) since you're going to have a lot of similar information that an org will need to know. I would expect the General disclaimer to be linked in both. The overall idea is a good one however. --Izno (talk) 02:57, 17 March 2015 (UTC)
How about suggesting to the institutions, companies and so on that approach you as partners that they set up their own MediaWiki page as a first step, Lydia? Here, for example, is CC wiki World University and School's initial MediaWiki installation - http://worlduniversityandschool.org/mediawiki-1.24.1/index.php?title=Main_Page - and per the post on this page just below this, WUaS would like to offer online CC free accrediting university degrees in many large languages while developing multi-lingually in Wikidata. Thanks for helping to flesh out these "landing pages for partners" questions. --Scott WorldUnivAndSch (talk) 015:14, 18 March 2015 (UTC)

Update: I have now merged the content of one into Wikidata:Data access and moved the other to Wikidata:Data donation. --LydiaPintscher (talk) 16:07, 20 March 2015 (UTC)

Help needed, message copied from Help talk:classification (Awards)

I WOULD LIKE TO ADD THIS AWARD TO MY LIST OF AWARDS: 

        HONORARY DOCTORATE IN THE ARTS from THE UNIVERSITY OF THE ARTS, PHILADELPHIA, PA.

         Rosalyn Drexler

 – The preceding unsigned comment was added by 47.18.103.221 (talk • contribs).

This has been done. --Denny (talk) 14:58, 20 March 2015 (UTC)

Translation markup help please

Could someone who understands the use of <Translate> markup assist at Wikidata talk:Property proposal/Proposal preload#formatter URL, please? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 21:26, 19 March 2015 (UTC)

  Done with additional fixes. Matěj Suchánek (talk) 14:13, 20 March 2015 (UTC)

SNB doesn’t work

Hi

For Srečko Kosovel (Q368139), the SBN identifier doesn’t work I’m afraid. Can someone help? Jihaim (talk) 13:21, 20 March 2015 (UTC)

FAQ for Freebase users

Hey folks :) At the end of the month Freebase is going read-only and I expect we'll get an influx of new people. I started an FAQ for them at Help:FAQ/Freebase. Cheers --Lydia Pintscher (WMDE) (talk) 16:56, 20 March 2015 (UTC)

Would it be a good idea to redirect help desk to project chat? –Be..anyone (talk) 21:18, 20 March 2015 (UTC)

Marge Q11324611 (니모) to Q1804807 (Nimo)

I found Q1804807 (Nimo), and Q11324611 (니모 and ニモ). It must be marged Q11324611 to Q1804807. / 저는 Q1804807 (Nimo)와 Q11324611 (니모와 ニモ)를 찾았습니다. 그것은 Q11324611을 Q1804807로 병합해야 합니다. --아라는 다 알아 (talk) 02:45, 21 March 2015 (UTC)

done. You know Special:Preferences#mw-prefsection-gadgets to check the Gadget Merge?--Diwas (talk) 04:29, 21 March 2015 (UTC)

Wikidata weekly summary #150

 
Here's your quick overview of what has been happening around Wikidata over the last week.

Removing sitelinks from items

Why is it no longer possible to remove a sitelink from an item? It makes merging items unnecessarily complicated because you first have to delete the first item and then add the sitelinks to the other. -- Liliana-60 (talk) 12:12, 22 March 2015 (UTC)

It is possible, if I get you correctly. You need to click on an icon with a trash can next to the item.--Ymblanter (talk) 14:15, 22 March 2015 (UTC)
There is no trash can icon anywhere. In the old interface there used to be "edit | delete" links next to every sitelink. In the new interface, there is only an edit link at the top of the sitelink box, but no remove link anywhere. -- Liliana-60 (talk) 14:19, 22 March 2015 (UTC)
There is a very small trash can icon at every site link, last time I used it was today.--Ymblanter (talk) 15:29, 22 March 2015 (UTC)
I think now I see where the problem is. Noscript was blocking the scripts, but it didn't actually tell me there are blocked scripts on the page so I couldn't tell. And apparently, then I got served a version of the page which has no way of deleting pagelinks. -- Liliana-60 (talk) 16:26, 22 March 2015 (UTC)
@Liliana: If you want merge two item is better to use the Merge gadget. You can found it in your preferences. --ValterVB (talk) 16:39, 22 March 2015 (UTC)

Templates

Infovarius believes Wikipedia:Templates (Q11871615) should be used instead of Wikimedia template (Q11266439) for project template items. I agree with him, however there are 190,000 templates in Wikidata. Matěj Suchánek (talk) 11:41, 22 March 2015 (UTC)

I'd generally agree. Anyone have a bot that could go through and change it? Ajraddatz (talk) 17:28, 22 March 2015 (UTC)
  Oppose, there is nothing wrong with the current situation. This only causes spam on peoples watchlists. Sjoerd de Bruin (talk) 06:38, 23 March 2015 (UTC)

Search Engine Optimization

What is the actual Syllabus of OFF PAGE & ON PAGE SEO ?

What? This seems like spam. I rolled back and then reverted myself just in case... —Justin (koavf)❤T☮C☺M☯ 04:02, 23 March 2015 (UTC)

Add self link is no longer possible

See Talk:Q6294369. --Liuxinyu970226 (talk) 05:17, 23 March 2015 (UTC)

Protection indicators

Tracked in Phabricator
Task T92384

Katie worked on a script that shows a padlock on protected items. You can enable it on Test Wikidata and see it in action here. I propose that it be enabled as a default gadget on Wikidata, until a real server-side solution is available. --Ricordisamoa 12:28, 16 March 2015 (UTC)

I don't see it (the padlock) there. Jared Preston (talk) 14:19, 16 March 2015 (UTC)
You need to enable it in your preferences. It's a gadget there - not enabled by default. I also missed that first... --Lydia Pintscher (WMDE) (talk) 15:04, 16 March 2015 (UTC)
Ok, the padlock is so far out to the right side, that I missed it. But other things, that I do not see in for example Project:Village pump (Q16503) makes it still more visible on that project than here. For example the black [edit]-buttons with mouseover-text who explains that the page is protected. -- Innocent bystander (talk) 15:10, 16 March 2015 (UTC)
Anyone else? People not seeing they can't edit has been brought up many times as a big issue to me so it'd be good to get a few more yay or nay on this :) --Lydia Pintscher (WMDE) (talk) 20:20, 22 March 2015 (UTC)

One question I had was if there should be padlock indicators on all pages (e.g. including wikitext pages) or only items and properties? I think there would be no objection to the latter, and then it could be extended to all pages if people want.

Regarding the [edit] buttons with mouseover text... it is nice but is inconsistent with how MediaWiki core (non-item/property pages) work, is buggy implementation and has other issues. The indicator will work more reliably and eventually we would like to have it work via a MediaWiki extension or such instead of a gadget. The gadget is more of a short term, easy solution for now. Aude (talk) 16:53, 24 March 2015 (UTC)

Q2533601

This item seems to have at least two different topics mixed together, but I don't know how to separate it into two items. If someone could help me with that (or point me to a page that explains how), it would be much appreciated. Mr. Granger (talk) 18:24, 21 March 2015 (UTC)

for me the wp-articles fit in one item, nl emphasizes different but also belongs there. Or what do you think does not fit. Oursana (talk) 19:13, 21 March 2015 (UTC)
Well, the en.wp and en.wv articles give a list of scams, whereas the fr.wp and nd.wp articles seem to discuss one particular type of scam (apparently the one described in section 8.15 of the en.wp article). In particular, the item is labeled as an "instance of Wikimedia list article", which seems appropriate for en.wp and arguably also for the Wikivoyage articles, but certainly not for nl.wp, fr.wp, and de.wp. Mr. Granger (talk) 19:27, 21 March 2015 (UTC)
Mr. Granger This is what we call the "Bonnie and Clyde problem". Where some wikis have multiple focussed articles and others have one general article but none have both. We have to split the wikidata items because otherwise we can't make statements that are true about both the general and the specific items but if we do split the items then the sitelinks are broken. See my proposal for a property to help fix this.
To split an item first create a new item (menu item on the left side).
Delete sitelinks from the old item then add them to the new item.
Alternatively enable the "Move" gadget; (on your preferences/gadgets). This should add a "move" link next to 'edit' above each group of sitelinks
Enable the "LabelLister" gadget as well. This adds "Label Lister" to the left side menu. Clicking this will show you every Label, Description and Alias in every language for the current item. Go through these on the old item and delete any that no longer apply to this item and add these to the new item if appropriate. You'll be surprised by how many languages you will understand enough of to do this.
Hope this helps. Filceolaire (talk) 00:24, 22 March 2015 (UTC)
Thank you! Mr. Granger (talk) 11:29, 23 March 2015 (UTC)

fecha de publicación dentro de versión

Buen día. Agregue la propiedad fecha de publicación a la entrada de fluxbox y desde wikipedia podía usarla {{Propiedad|P577}}; sin embargo, la han eliminado como declaración y la dejaron dentro de la propiedad versión. ¿Me pueden orientar como recuperar ese valor en wikipedia? Muchas gracias. petrohs (gracias) 18:25, 23 March 2015 (UTC)

@PetrohsW: Hay una pagina por discusion en espa~ol aqui: Wikidata:Café per no hay mucho actividad... Mi espa~ol no es muy bueno pero, puedo tratar ayudarte. No se entiendo tu pregunta... ?Hay informacion aqui en Wikidata y en w:es y los dos son diferentes? —Justin (koavf)❤T☮C☺M☯ 02:00, 24 March 2015 (UTC)
Thanks @Koavf: My english is poor. I add the property publication date to the fluxbox and from es.wikipedia could use {{Property|P577}}; however, have been removed and set in property version. Can I be oriented as retrieve that value in wikipedia?. petrohs (gracias) 02:58, 24 March 2015 (UTC)
@PetrohsW: Si la propiedad fue eliminada en Wikidata, entonces será imposible llamar el valor desde Wikipedia pues ya no existe. No podría explicar por qué fue eliminada, pero sí que fue eliminada por un administrador de Wikidata. Si acaso la propiedad no puede volver a ser colocada en la entrada, entonces deberás añadir la fecha de publicación manualmente en el artículo de Wikipedia. Allan Aguilar (talk) 14:53, 24 March 2015 (UTC)
Muchas gracias don @Allan Aguilar: Esperaba que existiera algo como {{Propiedad|P348|sub=P577}} para obtener valores anidados. petrohs (gracias) 16:56, 24 March 2015 (UTC)

Help verify WikiGrok data before bot starts running

As I've mentioned here before, the WMF mobile web team has been experimenting with micro-contribution interfaces for adding metadata to Wikidata from within Wikipedia articles. These experimental interfaces are very similar to Magnus's Wikidata game, but instead of posting the results to Wikidata immediately, we have been collecting the results in a database so that they can be aggregated for better accuracy. Now that we have collected a large number of responses, we would like to try posting some of the aggregated data to Wikidata and get community feedback on the quality and usefulness of the data. User:Atlasowa has suggested that before we do a test run of the bot, we first post some of the edits that the bot would make and get the community to scrutinize them for errors. That should give us some idea of the quality level that can be expected from WikiGrok edits and help inform how and if we should move forward with the feature (and also help the community decide how to proceed with our bot approval request). I've posted a list of the first 100 potential WikiGrok edits here:

Please help to vet the accuracy of these potential edits by filling in the table on that page.

To learn more about our immediate plans with WikiGrok, please read through our bot approval request. To learn more about WikiGrok in general, please see our extension documentation. If the project ends up moving beyond a purely experimental status, we will also be creating an FAQ-type page for more community-focused documentation. Ryan Kaldari (WMF) (talk) 22:24, 23 March 2015 (UTC)

Ryan Kaldari (WMF), you [deleted the manual vetting that i did, without message or comment. Thanks, buddy. <sarcasm/> Now Pasleim has done the same work a second time. I'm done wasting my time on this. --Atlasowa (talk) 10:52, 24 March 2015 (UTC)
@Atlasowa: As explained in the text that I added above the Delete section, the correctness of the deletions is purely dependent on the correctness of the corresponding insertions. That's why they were originally filled in with "N/A" which means "not applicable". For example, if adding the claim "studio album" is not correct for an item, it is also not correct to delete the "album" claim. I'm sorry if that was confusing. Ryan Kaldari (WMF) (talk) 17:24, 24 March 2015 (UTC)
The list of proposed edits do not indicate what citations would be provided to support the edits. So I would rate them all unacceptable at this point. Jc3s5h (talk) 12:39, 24 March 2015 (UTC)
The issue of citations is explained in the bot request. I'm currently only asking for people to vet the accuracy of the claims in the table I have posted. The acceptability should be discussed at the bot request. Thanks. Ryan Kaldari (WMF) (talk) 17:24, 24 March 2015 (UTC)

How to deal with lists split by first letter

There are hundreds of cases on Wikidata of items like Q6596289: Wikipedia lists that are split by their first letter (in this case D–F). There are a few potential problems here:

  1. The items are often not disambiguated, for example Q3834238 has the same label and description as Q6596289. Should they be disambiguated in the label, the description, both, neither?
  2. They often don't align between wikis. For example, one wiki might have 1 long list, while another one divides the list into 26 smaller lists, and another one divides it into 10 smaller lists.
  3. These items don't really represent distinct concepts, just arbitrary divisions of a long list. Would it make more sense for us to merge them up into the parent list item (in this case Q2593450)?

Kaldari (talk) 00:45, 24 March 2015 (UTC)

Suggestion:
As they are separate pages at Wikipedia, we end up having an item for each. For use in statements for other items, these are rarely of much use to Wikidata. --- Jura 07:28, 24 March 2015 (UTC)


@Kaldari, Jura1: I dealt with this in this item some time ago list of Unicode characters (Q17081188)     . It's an interesting example because the list is huge and is split into sublits, and subsublists ... Oh no, it was Unicode block (Q3512806)      what a mess /o\ Oh no, it was more like items like Unicode characters from 1B000 to 1BFFF codepoints (Q3513269)      I dealt with ... If it can be useful, I wrote a script https://github.com/TomT0m/wp_scripts/blob/master/scripts/wikidata_unicode_table_fr.py to do some of the work. It is old so it might not work anymore, an it needs to be worked on to complete the work a little bit more. TomT0m (talk) 10:27, 24 March 2015 (UTC)

An item for every street?

I'm concerned that Wikidata:Requests for permissions/Bot/RobotMichiel1972 2 had little discussion or community consensus, for a 240,000-item import and a precedent for several magnitudes more than that. I have asked the bot operator to pause the job - which I appreciate is a good-faith effort - until that discussion has has taken place in a more prominent venue. I also think we need a longer period for the approval of bot tasks (or at least, for those of such size) than 48 hours. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 21:56, 10 March 2015 (UTC)

Yes, a pretty fundamental question. The natural thing to happen would be that in the long run everything that can be imported into Wikidata will be copied into it. Hard to draw a line somewhere, except to exclude low-quality data (the abhorrent example is Lsjbot which copied the worst database it could find, leaving us with thousands of fictitious taxa). - Brya (talk) 06:03, 11 March 2015 (UTC)
I think we already have all streets in Paris. --- Jura 06:07, 11 March 2015 (UTC)
I don't see anything wrong with items for streets. We might need stricter notability rules for living people/companies, but for places and other sorts of things, I clearly tend towards inclusiveness. In this case, the streets are already referenced in an external database, with make the statements easily verifiable. That said, this bot run could be certainly improved (as discussed in User talk:Michiel1972#Street items . When bot create item with potentially disputable notability, it may also be a good idea to put some efforts in adding proper references. In the long run, references sounds like a promising tool for quality/notability analysis. --Zolo (talk) 08:24, 11 March 2015 (UTC)
As for numbers, in my city (population ~1000 000) there are >1500 streets. So the ratio is roughly 1 street per 600 men, which can give us up to 10 millions streets. From europeo-centric POV. --Infovarius (talk) 09:48, 11 March 2015 (UTC)
There are maybe 1500 people where I live and much more than 4 streets, but the number is not the main problem for me, it's the volatile nature of postal code (P281). It's the internal concern of the largest postal company and they can change without notice. In larger cities located in the administrative territorial entity (P131) can also change every four years. -- Innocent bystander (talk) 10:05, 11 March 2015 (UTC)
As it's about Dutch streets, they have the same postal code for 30 years, as long as we have a postal code. I'm not that concerned that it's volatile, and if it is, we should forbid the property at all, instead of trying to not use it for Dutch streets ;-) The good thing I see, is that for every Dutch item in Wikidata we can start adding a location connected to a street, and Reasonator can show all Wikidata-items in a certain street once it gets properly populated. I will definitely add this for all items in my own home town, so I believe this import is gold. Edoderoo (talk) 10:37, 11 March 2015 (UTC)
I don't see any problem with one item for every street. I don't see any problem with any doubtful volatility either, if postal code (P281) or located in the administrative territorial entity (P131) should change, so change it on Wikidata as well. Hopfully we'll have sources to use and bots to update the data if needed. /ℇsquilo 10:44, 11 March 2015 (UTC)
What is the limit ? Do we want to add all addresses too ? Do we want to create an item for every human ? For me streets are not common knowledge, just temporary information (who knows the streets of Roma under Julius Cesar) ? The limit in that case is the notability and then the existence of WP articles to justify an item creation. We have to reduce the items creation speed and to focus on increasing the data of existing items. Snipre (talk) 11:01, 11 March 2015 (UTC)
"We have to reduce the items creation speed and to focus on increasing the data of existing items." Well, if we were employed to improve the database, I would agree, but we are not. We are volunteers, and we have to work with what we like, otherwise we will leave. -- Innocent bystander (talk) 11:16, 11 March 2015 (UTC)
These streets help with increasing the data of existing items. We can replace P969 (P969) with located on street (P669) on lots of items, so you can get a trusted overview of buildings at a street. Sjoerd de Bruin (talk) 11:19, 11 March 2015 (UTC)
That seems to be an argument for creating an item for a street when we have an item for something on that street, not for creating an item for every street. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 18:28, 11 March 2015 (UTC)
Agree with bystander
In this case, Michiel1972 even makes a good effort to add meaningful statements to the items. --- Jura 11:21, 11 March 2015 (UTC)
User:Innocent bystander If your argument is to use blackmail, I prefer you leave: WD is a collaborative project not a game for egoistic gamers.
User:Sjoerddebruin But for your example do you need all streets of all countries ? How many streets have monuments, buildings which have an item in WD ? The problem here is the goal of the addition of all streets.
User:Jura You can always add statements to an item: you can take a phone book and for each phone number add the first name and last name of the owner. Do you want to add all phone numbers in WD ? Snipre (talk) 14:44, 11 March 2015 (UTC)
User:Snipre: My argument is about why you cannot force me or anybody, who isn't interested in the subject, to maintain items about fictional characters in the Universe of Tolkien/Marvel/Disney/StarWars. You could if you paid me, but you don't. Now, instead, driven by my egoism, I maintain the set of items about minor planets. I have an interest in the development of the Solar System. Because of that egoistic interest, I will make several items without related articles, no matter how many else are interested. Because of that, I hope, it will be easier to maintain lists of minor planets on WP. Several WP-projects have had bot-projects to create such articles, and obviously today have problems with the maintenance of them. I hope my work will help with that too, but it will most likely never be a large scale cooperation with a large set of users on WD.
I do not understand your arguments about every street in every country. Most countries do not have databases about such subjects, so it will never happen. In Sweden, that database is private property of a private company, and is copyrighted. It cannot be added here, until after it has been out of date. I know, because I have worked for a company who subscribed to that database. Every mailbox in Sweden is in that database, it includes the names of everybody who uses the mailbox, together with the coordinates and height over the sea level. But as I said, it is protected by copyright, and is out of date when it becomes free. -- Innocent bystander (talk) 19:05, 11 March 2015 (UTC)
I'm not sure if we need every street in every country. What's nice about these Dutch streets, is that they are linked already to the appropriate town, therefor we can already list every known street in a certain Dutch town. If we can do the same for France, India or Guatamala, with the same level of knowledge, then yes, I would like that. If it is just the names of all streets without the town? Then it's indeed just a sjid-load of info that doesn't make much sense (yet). I can imagine that *all Dutch streets* are not of big interest to people in the US, but to people that populate Dutch wikidata-items it does makes sense. It adds extra information to existing items, and therefor helps in our strategy to reveal knowledge. Step by step, bit by bit. Edoderoo (talk) 15:31, 11 March 2015 (UTC)
Edoderoo What is knowledge ? What is information ? The sun is brightening today, that's information but can that information be in WD ? You say correctly that all Dutch streets can interest Dutch people and perhaps some others dozens of thousands persons, but can this be interested for WD which have a more larger focus: several billions of persons in 2015 and the others billions of persons in 2020 or 2030 ? Everything is information, not knowledge. The transformation of information to knowledge is more complex. Snipre (talk) 17:25, 11 March 2015 (UTC)
Dear Snipre, 99.9% of all info in WikiData will never be read by me, but that is NO REASON AT ALL to forbit others to add it. Comparing streets from my beautiful country with the sun shining doesn't make sense. You bring it like Michiel is adding non-info, but I already explained some use-cases. People, please see some of those streets, before you come here to complain that you don't care. I don't care either, so live and let live. Edoderoo (talk) 18:57, 11 March 2015 (UTC)
If there is no problem to import hundreds of thousands of items, I don't see why it should be prohibit by some contributors. There is a lot of things to do with these data : % of streets with a female name, most popular name, celebrities born at my address, etc. Pyb (talk) 17:50, 11 March 2015 (UTC)

I think, we should add the information which streets are connected. --Molarus 18:40, 11 March 2015 (UTC)

Just a note from the technical side: Wikidata definitely won't scale to have an item for every street. Especially not if we apply similarly lax criteria to other data as well. But to be honest I think the community will stop scaling long before the technology will. So please please be careful. --Lydia Pintscher (WMDE) (talk) 18:45, 11 March 2015 (UTC)

I have doubts about this import. Not about the import itself, I know Michiel will do a good job, but about the maintainability of the data. For each item we have, we need a (tiny) bit of community time. Having so many items about streets won't scale. Multichill (talk) 19:51, 11 March 2015 (UTC)
Special:Search/street in paris does look like good stuff. If we can't maintain these 3200 items, maybe we should work on the technical side to make it possible. --- Jura 20:00, 11 March 2015 (UTC)
I'm not sure that Rue Ordener (Q3449347) (for example) having multiple shares border with (P47) is helpful... Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:05, 11 March 2015 (UTC)
To see how it's used, have a look at fr:Rue Ordener. --- Jura 20:07, 11 March 2015 (UTC)
That seems like overkill; but even if the data is needed, "shares border with" is not the appropriate property. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:12, 11 March 2015 (UTC)
On further reading, I'm not sure why we need pages like fr:Cité_Nollez, either. Why are we trying to rebuild OSM? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:24, 11 March 2015 (UTC)
It is not only overkill, it is improper use of shares border with (P47). crosses (P177) would be better, but generally terminus (P559) is sufficient information for streets. /ℇsquilo 10:08, 14 March 2015 (UTC)
Looking at cité Hermel (Q2974666) on OpenStreetMap, and Google Streetview, it's hard to see why we would need it. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:12, 11 March 2015 (UTC)
@Multichill: if we can't maintain these items, then we need a special treatment like claim protection. There is actually no need to modify a claim if it is sourced from another database. Maintainability means periodic imports of the databases if there is some. Why would it be expensive ? In particular, why would a lot of items would need more attention than their initial import ? What would maintain mean actually ? We hear that a lot in this kind of discussion, maybe it would be worth thinking of what is actually maintenance in Wikidata and the cost. If the good maintenance unit is the database, the number of item is not really relevant. If it's the claim and every claims needs to be updated every day, it's impossible. (talk) 20:18, 11 March 2015 (UTC)
TomT0m: „There is actually no need to modify a claim if it is sourced from another database”. So all databases agree? I have strong doubts. --Succu (talk) 22:42, 11 March 2015 (UTC)
@Succu: The right process in Wikidata spirit, as concepted by its creators is not to modify the claim, but to add each a claim for each database, and if some database is known to be wrong mark its claim deprecated. TomT0m (talk) 08:36, 12 March 2015 (UTC)
I am working with items now that have data imported from databases, and I find a lot of mistakes. I don't blame the uploader, a database could sometimes be difficult to understand. And many claims were added before the "rank"-function was added, which I have to add now. -- Innocent bystander (talk) 08:41, 12 March 2015 (UTC)
@Innocent bystander: Can you be more specific on the kind of mistakes you have to deal with, examples maybe ? From what you say here I can't decide whether the mistake was due to a bug in the robot or from the databases datas. TomT0m (talk) 08:52, 12 March 2015 (UTC)
@TomT0m: In 1744 Harriet (Q122010), Palomar–Leiden survey (Q2048368) is marked as discoverer, I am not sure that is a good interpretation of the database. van Houten, van Houten-Groeneveld and Gehrels are the Palomar-Leiden Survey. So they are mentioned twice here.
All results in provisional designation (P490) are correct, but 6557 P-L is the primary provisional designation and should be marked as preferred. It is marked in bold in the database, and can also be found in lists like these.
The labels of minor planets have no universal standard. enwp normally write them as "1744 Harriet" while dewp write them as "(1744) Harriet" and others as "Harriet" with "asteroid" in description, and some languages "transcribes" the names (even if they have the same set of alphabet). Sometimes the bots have removed the parts of the sitelink which are within brackets. Most often here, they shouldn't, the brackets are often parts of the name. It doesn't matter which system you use, but you have to be consistent, at least within one language.
I have found named after (P138) linked to disambig-items and I have found other linking to list-items. -- Innocent bystander (talk) 09:40, 12 March 2015 (UTC)
@Innocent bystander: I guess a bug in the import bot should be reported to the bot owner, it's pretty easy to undo what he did automatically with a few special cases checking in the code and add the code to do it correctly, especially when there could be a lot of similar errors. Anyway I don't think this cotradicts what I say, except a bug in the initial import, in which case we need to solve the bug, there is no reason to touch an imported claim after the import (except to mark it deprecated :) ) TomT0m (talk) 20:26, 12 March 2015 (UTC)
@TomT0m: I'm afraid the bot owner who made the import from JPL and MPC is not active anymore. I do not think we should let obvious mistakes stay even as "deprecated", like when disambig-items are used to describe relations. There is a minor planet with the name Chariklo, it is named after the wife of the mythological person Chiron. (You see him in films sometimes.) I am not aware of any item about her (yet), but that is not a reason to use the disambig-item with the title Chariklo instead, not even as a deprecated statement. Because nobody has ever, not even the stated source, supported that claim.
My ambition now is to create a chain of follows (P155)/followed by (P156) between all minor planets in the mpc-database. When that is done, I intend to ask other bot owners to fill all these items with claims. (It's easier when the chain is complete.) The reason a full P155/P156-chain is missing, is that not all articles exists in all languages. One language may have nr 4567 and another has 4568. The link between them is not obvious, and I guess it can be tricky for a bot to connect items in Asian languages with those in Latin or Cyrillic alphabets. But as manual user, I can do that. -- Innocent bystander (talk) 06:17, 13 March 2015 (UTC)
I share Lydia's doubts on us, the community, being already prepared to scale to the required size to have all streets, planets, books, songs, and insect species in Wikidata right now. It is clear that effective measures to deal with the data and to avoid spam, vandalism, and inaccuracies have to be in place and tested before we grow by a factor of 10 (which should be sufficient for all of these use cases). But on the other hand, if we don't actually try it out, we will not grow to be able to deal with that data size and requirements. I would suggest that we selectively allow for such datasets to be added, whenever the person who wants to do the upload also at least sketches how to aim for a sustainable data quality.
So, I would very much support to gather datasets like all streets, etc., and to integrate them. There is no rush, but our vision should be to actually have this data in some day, not too far in the future. Especially since we already have a well-established data set for Paris streets, New York streets, etc.
Regarding the technical scalability, Wikidata has right now a few bottlenecks (terms table, change propagation), but terms table is being tackled and change propagation has no direct relation to number of items, so I think that technically we can scale to 100M. But again, I am more worried that we become a data dump or a data desert than I am worried about us being not able to technically scale. --Denny (talk) 15:23, 12 March 2015 (UTC)
@Denny: I can import around 26 millions of items with minimum 3-4 statements and around 100 millions with 2 statements: molecules data. When can I start ? 141.6.11.12 10:37, 13 March 2015 (UTC)
Read Wikidata:Comprehensive groups of items. --Succu (talk) 10:45, 13 March 2015 (UTC)
Succu Thanks for the link. But if I read the page, I think that the street problem should never appear. Snipre (talk) 12:51, 13 March 2015 (UTC)
It's a new page, see #Collections of entries. --Succu (talk) 12:58, 13 March 2015 (UTC)
Hello anonymous! Sounds great. But one condition was to provide at least a sketch of how to maintain the data in the future, i.e. how to avoid lasting spam and vandalism and at the same time be open for useful contributions. Any plans for that? --Denny (talk) 15:08, 13 March 2015 (UTC)
IMHO Wikidata is not a land register, and even bots would have a hard time keeping all those items up-to-date. It is time to better define our notability boundaries and focus on quality instead. I don't know much about OpenStreetMap, could such data be hosted there? --Ricordisamoa 13:04, 15 March 2015 (UTC)

I think it'd be good if, say, streets over one mile long would be considered notable. --AmaryllisGardener talk 23:26, 15 March 2015 (UTC)

The length of a street looks like a very strange criteria. The streets in the older parts of a city are often more interesting than the longer suburban. In svwp I guess almost all streets in the "old town" of Stockholm have articles, and they are all very short and narrow. -- Innocent bystander (talk) 07:02, 16 March 2015 (UTC)
Hmmm, good point, perhaps age would be a better criteria? --AmaryllisGardener talk 21:48, 17 March 2015 (UTC)
I don't think so. Many villages in the Netherlands and the rest of Europe have streets which are much older than most streets in New York. Dinsdagskind (talk) 12:20, 19 March 2015 (UTC)
I don't think bots should be used for that type of items... only streets that have a wp article (with real information in it) or streets where a notable monument exists (or where a notable person lived, for ex.) should be created…, for linking purpose. OSM is much more appropriate for geographic data like that... --Hsarrazin (talk) 21:30, 24 March 2015 (UTC)

Process for approving bots

Discussion above concentrates - not unreasonably - on the specific import; but I also raised the question of the process by which bot approvals which constitute major changes to Wikidata are made. Anyone have any views on that? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 18:28, 11 March 2015 (UTC)

Unfortunately, the bot is continuing to create pages, most recently 't Zand (Q19563267) just now. Perhaps it could be blocked (with a block summary noting that no malicious intent is ascribed) until this matter is resolved? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:02, 11 March 2015 (UTC)

Nobody voted for Michiel1972 request. That's bad! Ymblanter approved the job anyway. --Succu (talk) 22:32, 11 March 2015 (UTC)
Actually, for most requests nobody votes. I always suggest users to add the permission page to their watchlists and comment on requests, but nobody is doing this.--Ymblanter (talk) 08:26, 12 March 2015 (UTC)
As I've noted elsewhere, this particular request was open for comments for less than 48 hours before you approved it. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 10:43, 12 March 2015 (UTC)
If the community decides requests should be open longer, I will gladly do it. Until there was any community decision, I will close requests when I think they are ready to be closed. This is fully within my authority.--Ymblanter (talk) 10:48, 12 March 2015 (UTC)
Please don't bite Ymblanter for doing his job. He's basically the only bureaucrat working on Wikidata:Requests for permissions/Bot and he's doing a good job. He leaves requests open for as long as needed, but he'll close requests if a request looks good and nobody objects. That's exactly what happened here. Multichill (talk) 20:27, 12 March 2015 (UTC)
I've made no criticism of Ymblanter; I'm questioning the process (or lack of it), by which he and anyone else is operating. The request was open for under 48 hours; given that there is clearly an awareness that the page is under-watched, that's surely not adequate? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 10:06, 13 March 2015 (UTC)
I really have no intention of personalising this, but since you mention it, and speaking generally - what authority? By whom was it granted, to which role (admin, crat, or other), and where is it documented? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 10:06, 13 March 2015 (UTC)
Please note Wikidata:Bots#Approval process. Considering there often is no input by the community, extended silence (i.e. no voiced disagreement) is also interpreted as consensus. If someone finds a request approved they would have disagreed with, they are of course free to post to WD:BN. Discussion can be reopened then and if demanded the approval also revoked, until the discussion has lead to consensus for further action. Regards, Vogone (talk) 14:10, 13 March 2015 (UTC)
Thank you for the link. I'm having trouble seeing "under 48 hours" as "extended silence". I think the policy needs to be clearer on this point. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 21:23, 13 March 2015 (UTC)
I agree the (that) quick approval was a bit unfortunate, normally we would wait a bit longer in case of possibly controversial requests. Vogone (talk) 23:27, 15 March 2015 (UTC)
The most simple way is to modify the approval of data import by bots: a data import should be discussed in the framework of a project first and once a consensus between persons having an common interest and an global objective is reached, the request for the bot import can be organized with the link to the discussion. Data import should be the result of a project and not of the action of one person or two-three persons working without visibility on some global objectives.
Currently too many actions are the results of some persons or some groups from local WPs and without global/international view: people use WD to do what they are doing in their local WPs without coordination with others WPs or global projects. Snipre (talk) 13:09, 13 March 2015 (UTC)

Please stop the bot

As of the last couple of minutes, the bot continues to create items for streets (see Valkenhof (Q19576703)). Please can someone stop it, until we conclude this discussion and determine the community's consensus? If the consensus is to allow it, the delay will cause far less harm (and work for others) than the deletions necessary if the consensus goes against it. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 10:09, 13 March 2015 (UTC)

Hi, my bot has been blocked. I was at 95% of the database, now it is stopped and incomplete after a month of work. What's the next step in this community-consensus-process? In my opinion wikidata can contain all streets, all books, all albums, all monuments, etc. as long as there is sufficient content and a free source. Michiel1972 (talk) 14:34, 13 March 2015 (UTC)


Should I continue?

Please vote at Wikidata_talk:Requests_for_permissions/Bot/RobotMichiel1972_2, so we can see how the community thinks about this task. Michiel1972 (talk) 08:19, 14 March 2015 (UTC)

No you should not, at least until the above discussion has concluded. It's not helpful that you've started a poll elsewhere, without even mentioning it in the discussion here there. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 23:04, 15 March 2015 (UTC)
Ehm, I see a mention here... It's only to get a clear consensus instead of making this discussion more harder to follow. Sjoerd de Bruin (talk) 23:17, 15 March 2015 (UTC)
Sorry, typo; I meant there. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 14:10, 16 March 2015 (UTC)

Redux

Well, the bot has been unblocked and continues to create items about Dutch streets. The conversation above appears to have petered out. Is that consensus to continue in this fashion? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 21:29, 19 March 2015 (UTC)

As said, I'd like to see such data upload proposals to be partnered with a rough data maintenance sketch. And somehow more visibility for such data upload proposals, maybe have them be announced on the project chat as well, so that people are aware when a few hundred thousand items might be added? I am in general for such uploads, but they have to be done careful and with a sense of sustainable ownership. --Denny (talk) 15:02, 20 March 2015 (UTC)


So, the bot was unblocked a couple of days ago and the end result is:

  • around 240.000 dutch street items were created/imported by bot, which are not used
  • around 1.800 dutch streets that already have a dutch WP article and existing wikidata items were not enriched with the source according to spot checks (Q3094222 Q12012495 Q3032101)
  • there is no maintenance concept beyond "There was an additional question of maintainability, which is very limited because of the static information of postal codes"

What is next for the wikidatadump? Maybe all US streets (something between over 1 million roads and 26 Million Road Segments in Continental US)? Or, as said above, molecules data (around 26 millions of items with minimum 3-4 statements and around 100 millions with 2 statements)? Quite obviously this talk about "such uploads have to be done careful and with a sense of sustainable ownership" or "to provide at least a sketch of how to maintain the data in the future, i.e. how to avoid lasting spam and vandalism and at the same time be open for useful contributions" (quote Denny) this is just empty words: mass imports are supported regardless and hundreds of thousands of unused items are created by bot. No deletions, no bot block. --Atlasowa (talk) 15:05, 23 March 2015 (UTC)

They will be used, it just takes the right bot to do a job like that. Sjoerd de Bruin (talk) 15:10, 23 March 2015 (UTC)
"right bot job"? What do you mean, maybe a bot importing the dutch yellow pages so that the streets can be used as address property? Or a bot job to automatically generate dutch street "wikidata-articles" at, let's say: Navajo Wikipedia (Q17275421 Q17275427 Q17275459)? --Atlasowa (talk) 15:18, 23 March 2015 (UTC)

BTW, Re: "I think we already have all streets in Paris" That was completely different, existing fr.Wikipedia article items were enriched. 6,583 items in frwiki Category:Voie_parisienne, 5477 wikidata Items with Paris city digital code (P630), 5475 of those with sitelink to frwiki. --Atlasowa (talk) 15:11, 23 March 2015 (UTC)

To do things like this flawless, hence there are a lot of streets with that name. Note that creating or feeding Wikipedia articles is not the main goal of Wikidata. Sjoerd de Bruin (talk) 15:20, 23 March 2015 (UTC)
Come on, Sjoerddebruin, this is a bulk data import. Even if you could map most of the Rijksmonumenten to the new ~240.000 items, there would still be ~200.000 dutch streets at wikidata unused. And i bet that the Rijksmonumenten will map disproportionally to the ~1.800 dutch streets that already have WP articles. What else, will you import all home addresses of dutch politicians, artists, journalists, sportsmen? Or the dutch business directory? Is that the main goal of Wikidata? Actually, feeding Wikipedia articles is a main goal of Wikidata (or was?). --Atlasowa (talk) 17:39, 23 March 2015 (UTC)
The vandalism mentioned above is real. Please maintain your items. Thanks --Haplology (talk) 14:01, 24 March 2015 (UTC)

Mergers

It looks like Q15623447 should be merged with grant (Q230788), but there's a series of conflicts. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 00:06, 15 March 2015 (UTC)

Likewise triple-alpha process (Q336225) and helium fusion (Q1144695). Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 00:16, 15 March 2015 (UTC)
I guess the triple-alpha process (Q336225) is always a helium fusion (Q1144695) but a helium fusion (Q1144695) is not always the triple-alpha process (Q336225). --Diwas (talk) 00:48, 15 March 2015 (UTC)
@Diwas, Pigsonthewing: Does anybody understands what
⟨ triple-alpha process (Q336225)      ⟩ composed of Search ⟨ 336225 ⟩
of (P642)   ⟨ subset (Q177646)      ⟩
 is supposed to mean ??? The qualifier seems not to make sense. If it's supposed to mean every times a triple alpha reaction occurs, a helium fusion occurs, the statement
⟨ triple alpha reaction ⟩ subclass of (P279)   ⟨ helium fusion ⟩
means exactly that. Overwise their is absolutely no semantics defined. See Help:Classification and Markus' comments on this rfc for more informations. TomT0m (talk) 18:58, 15 March 2015 (UTC)
I think you mean
⟨ helium fusion (Q1144695)      ⟩ composed of Search ⟨ 336225 ⟩
of (P642)   ⟨ subset (Q177646)      ⟩
. I have tried to figure that sometimes helium fusion (Q1144695) is a chain of several processes and one of this processes may be the helium fusion (Q1144695). helium fusion (Q1144695) is sometimes a single process, but sometimes it includes more processes. --Diwas (talk) 07:15, 16 March 2015 (UTC)
@Diwas: That make sense, but the of (P642)   qualifier ??? what is that supposed to mean ? TomT0m (talk) 18:32, 16 March 2015 (UTC)
@Diwas: I see you built several of those kind of statements ? What are those ? https://www.wikidata.org/w/index.php?title=Q19361458&direction=next&oldid=199846730 for example ? TomT0m (talk)
Maybe several but not dozens, so it will be not a problem to revise it. Q19361458 is mostly a part of one of the several types of places/streets but not parts of common streets. I am not sure about the of (P642)  , but it seems to be a restriction of the statement. Do we have a specific property for qualifiers stating that (mostly) one of several values applies or one of several statements applies? --Diwas (talk) 19:12, 16 March 2015 (UTC)

I can't understand TomTom's post or the replies to it. However, I note that triple-alpha process (Q336225) includes a link to lb:Heliumbrennen. No one has addressed Q15623447 / grant (Q230788). Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 21:42, 19 March 2015 (UTC)

I had added some (maybe not perfect) statements to triple-alpha process (Q336225) and helium fusion (Q1144695), because I think that it both are not always the same. triple-alpha process (Q336225) produce only carbon (Q623) and gamma ray (Q11523), but helium fusion (Q1144695) may produce oxygen (Q629) and neon (Q654) too. But I am not a physicist (Q169470). I guess , too. I guess you can see lb:Heliumbrennen as triple-alpha process (Q336225) with additional informations or as a helium fusion (Q1144695) with a too specific header section.


Maybe Q15623447 should be merged with dotation (Q1251158) but I don't know. Hello ..., are there any users speaking cs, hy, kk, lt,pl, ru, sah or uk and maybe de or sv too and be familiar with finance? --Diwas (talk) 02:14, 21 March 2015 (UTC)
well, AFAIU, and in French financial system, a "grant" is money given to an organisation, that it can use freely... a "dotation" (Q1251158) is money given generally according to a regulation, and for a specific purpose… for example, the Government may give "grants" non-governmental associations, but will give "dotations" to local services, for their official function. - so, it's not the same purpose, and they are not given according to the same regulations, nor by the same organisations… to me, these 2 should NOT be merged. --Hsarrazin (talk) 21:21, 24 March 2015 (UTC)

Milestones template

FYI, I created {{Milestones}}, using code from Jura's user page:

-- Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:45, 23 March 2015 (UTC)

Also {{Milestones-long}}:

Universe (Q1) Q10 Boston (Q100) Gabon (Q1000) Dutch Wikipedia (Q10000) Cadier en Keer (Q100000) Matkaoppaat (Q1000000) Category:Panagyurishte (Q10000000) Q100000000
Earth (Q2) Norway (Q20) 2 (Q200) Q2000 Crawling (Q20000) Dúné (Q200000) Stasina americana (Q2000000) Bogenschiessen (Pauly-Wissowa) (Q20000000) Q200000000
life (Q3) United States of America (Q30) Sagas of Icelanders (Q300) Herzogtum Lauenburg (Q3000) 119 (Q30000) Ars-en-Ré (Q300000) list of mayors of Westdorpe (Q3000000) The Synergistic Activity of Thyroid Transcription Factor 1 and Pax 8 Relies on the Promoter/Enhancer Interplay (Q30000000) Q300000000
death (Q4) Austria (Q40) Jenna Jameson (Q400) Mahajana College (Q4000) Kim Chaek University of Technology (Q40000) Tomorrow will be better (Q400000) Hildegard Gölz (Q4000000) Kverneland (Q40000000) Q400000000
human (Q5) Q50 Citrus × limon (Q500) Emperor Shengzong of Liao (Q5000) victory (Q50000) Tim Jitloff (Q500000) Template:Snooker tournaments (Q5000000) El Refugio (Q50000000) Q500000000
Q6 New York City (Q60) Ichiro Suzuki (Q600) 1331 (Q6000) Saetgang Station (Q60000) Czechoslovakia at the 1980 Winter Olympics (Q600000) Erik Gabrielsson Emporagrius (Q6000000) ATP synthase-associated protein, putative (Q60000000) Q600000000
Q7 Bern (Q70) Swedish Chef (Q700) Silbermond (Q7000) Rougemont (Q70000) SMAD protein (Q700000) Giuseppe Murnigotti (Q7000000) 3,5-dimethyl-3H-pyrazole (Q70000000) Q700000000
happiness (Q8) Tim Berners-Lee (Q80) Costa Rica (Q800) Japanese Grand Prix (Q8000) Childersburg (Q80000) Ivanovka (Q800000) Category:Medalists at the 1984 Winter Olympics (Q8000000) Medal (Q80000000) Q800000000
Q9 Paris (Q90) Kazan (Q900) Alpignano (Q9000) Rodolfo Wirz (Q90000) Gyula Czigány (Q900000) Category:Langen Brütz (Q9000000) ZFOURGE UDS 19909 (Q90000000) Q900000000

-- Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:06, 23 March 2015 (UTC)

Also:

{{Milestone properties-long}}

P1 (P1) video (P10) P100 (P100) record held (P1000) Research Vocabularies Australia ID (P10000)
P2 (P2) place of death (P20) inflows (P200) CPDL ID (P2000) P20000 (P20000)
P3 (P3) continent (P30) ISO 3166-2 code (P300) marriageable age (P3000) P30000 (P30000)
P4 (P4) child (P40) platform (P400) has fruit type (P4000) P40000 (P40000)
P5 (P5) author (P50) exclave of (P500) P5000 (P5000) P50000 (P50000)
head of government (P6) P60 (P60) Wine AppDB ID (P600) water footprint (P6000) P60000 (P60000)
P7 (P7) P70 (P70) Kemler code (P700) DigitalNZ ID (P7000) P70000 (P70000)
P8 (P8) P80 (P80) notable work (P800) electron configuration (P8000) P80000 (P80000)
P9 (P9) P90 (P90) P900 (P900) World History Encyclopedia ID (P9000) P90000 (P90000)

-- Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 23:55, 24 March 2015 (UTC)

ISSNs for web and print

Faraday Discussions (Q385884) (for example) has two values for ISSN (P236), one for print the other for the online version. Should we use qualifiers, or separate properties to distinguish them. In either case, what should these be, or be called? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:21, 23 March 2015 (UTC)

  Notified participants of WikiProject Periodicals Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:48, 23 March 2015 (UTC)
eISSN is sometimes (not always) used to distinguish the electronic one, but many online-only journals report a single ISSN without calling it anything special. I would suggest just using both and a qualifier if needed. Andrew Gray (talk) 21:27, 23 March 2015 (UTC)
clearly, in future years, paper editions will tend to disappear, while e-editions will appear… a qualifier seems more than enough to me :) --Hsarrazin (talk) 22:58, 24 March 2015 (UTC)

Taxonomy questions

(moved to WikiProject Taxonomy)

@Pengo: You will probably want to go to WT:WikiProject Taxonomy. (My personal view on items 1a and 1b is that there should be 2 items if there are two different names in different language Wikipedias, and that the links should be stored according to the name. YMMV.) --Izno (talk) 22:15, 24 March 2015 (UTC)
Thanks. I've moved my questions there to avoid duplication. Pengo (talk) 22:23, 24 March 2015 (UTC)

merging items in the Dictionary of National Biography (Q1210343)

"Pulcherius (DNB00) (Q19089232)" is a page in the wikisource Dictionary of National Biography which is effectively a redirect to "Mo Choemoc mac Beoain (Q19049325)" - another page in the DNB. It is a redirect in the DNB itself however so it is reproduced verbatim (but enhanced with a wikilink) rather than being a wiki redirect. I think we should merge these pages and delete the sitelink to "Pulcherius (DNB00) (Q19089232)". Anyone else have an opinion? Filceolaire (talk) 14:52, 21 March 2015 (UTC)

I was wondering about DNB entries as well. DNB entries at wikisource seem to be problematic.
Another question is if these sitelink should on an item about a person (P31:Q5) or on an item about a biographic article.
The wikisource entry could be in a statement on an item about a person, but probably not a sitelink on such an item. --- Jura 15:03, 21 March 2015 (UTC)
@Charles Matthews: is thinking about what to do with these. Personally, my preference would be to create a "DNB entry" property and put Wikisource links in there, from the item about the subject (similar to Oxford Dictionary of National Biography ID (P1415)). However, crosslinking from the subject to the DNB entry item would also work. Andrew Gray (talk) 15:24, 21 March 2015 (UTC)
@Andrew Gray:: Agree. Here is the proposal: Wikidata:Property_proposal/Person#DNB_entry_at_Wikisource. --- Jura 17:07, 21 March 2015 (UTC)

Well, a decision needs to be made about the future. The initial point by @Filceolaire: is about the type of "soft redirect" from the Dictionary of National Biography that is called a "See article" on Wikisource. These articles have not been systematically created on Wikisource, so far. I don't think they should have Wikidata items as such.

The other point brought up by @Jura1: and @Andrew Gray: is more complex, and I have been discussing it by email. Before a property is created, could we consider the virtues of having the biography here as a separate thing? There are some attractions to me in having a complete "DNB quartet", namely a pair of a Wikisource page and matching Wikidata page, linking both ways, and another pair of a Wikipedia page (English) and matching Wikidata page. The biography Wikidata page should link to the conventional Wikidata page through a "main subject" statement, and the precise edition (which matters for the DNB) through a "part of" statement.

One reason this all interests me is that a large effort has gone into linking DNB Wikisource pages to English Wikipedia articles, with over 75% matched. This means most of the "main subject" links could be put in by a bot. With a caveat or two, that are also revealing. Checking along the links could show up two things: (a) cases where the proposed "main subject" link was not to a page carrying an ODNB number (OBIN); and (b) cases where the Wikipedia page linked to does not yet have a Wikidata item. Case (b) means the items should simply be created here. Case (a) can happen when, for example, a biography of X on Wikisource gets linked to X & Co., the company X founded.

This is all useful to understand, and underlines the way that Wikidata brings rigour, denying that "instance of" should just be over-ridden, as well as indexation.

So I do see quite some virtues in the apparently complex version, with four pages linked round in a sort of cycle. St. John, Spenser Buckingham (DNB12) (Q19053982) is a sample of the sort of page I mean. I suppose @ GZWDer: should be involved in this discussion? Charles Matthews (talk) 17:56, 21 March 2015 (UTC)

Charles Matthews, Jura I created a new item DNB redirect page (Q19648608) and made Pulcherius (DNB00) (Q19089232) and "instance of" this item. Does that work as a solution?
I am strongly opposed to having separate wikidata items for WP articles and for wikisource/DNB pages. If these are about the same person then they should link to the same wikidata item which will have various statements about that person.
Statements about the DNB itself will be on the wikidata item about that edition of the DNB and will have site links to the wp articles about that edition and to the wikisources table of contents page for that DNB edition.
OK? Filceolaire (talk) 23:33, 21 March 2015 (UTC)
No, as I don't think WikiSource has a policy of only accepting DNB. --- Jura 04:21, 22 March 2015 (UTC)
Jura Do you mean that wikisource could end up with two or more pages about the same person? Say a DNB item, a DNB redirect, a 1910 Britannica item and an author page? Or did you mean something else? Filceolaire (talk) 04:44, 22 March 2015 (UTC)
Sure. It could even have several versions of DNB about the same person. Only author pages seem fine to add to items with P31:Q5 (Sample: Q937). --- Jura 05:13, 22 March 2015 (UTC)

@Filceolaire:: Thank you for presenting a solution for the "soft redirect" issue.

On the broader point that has been raised here, you wrote "I am strongly opposed to having separate wikidata items for WP articles and for wikisource/DNB pages". OK, but we have to look further.

There can only be one link from a given item to a given Wikisource version. As things stand, the Author: namespace pages are privileged, and if there is an Author page linked from an item, there can be no further enWS links from that item. Therefore in some cases the DNB page would necessarily be two clicks away from Wikidata (this is how things stood until recently). In other cases they would be linked from Wikidata?

This would be a system with a serious inconsistency built into it, and not I think good for the longer term. What @Andrew Gray: has proposed does address it, but I see a lost opportunity in it for bringing the metadata from Wikisource over here (as well as the link-following advantages I detailed before).

It will occur often enough that a given person has a DNB, Britannica (different editions) and say Catholic Encyclopedia page, and this person need not be an author. Wikisource has disambiguation pages, though the system is not as thoroughly developed as we now expect on Wikipedia. So the unique enWS link could be to that, in case there isn't an Author: page. So "Author overrides disambiguation" could be the solution. I discussed this aspect of the issue some time ago with @Magnus Manske:.

Let me try and summarise a bit:

  • I do see operational advantages in having "data item" links from Wikisource DNB pages (and other pages), and these can only be achieved by having items here that link to them.
  • With a special property, as Andrew suggests, there would not be a "data item" link, but from a bot point of view a "string lookup" could find the relevant item. Better for machines than for humans.
  • The privilege given to Author: namespace pages (for enWS) as the link from Wikidata is convenient rather than something that works specially well.
  • The decision about how to handle these matters does relate to data, namely whether the metadata for pages such as the DNB's will be included in Wikidata, or simply left in Wikisource headers.

Please note also that the DNB has 69 volumes posted on Wikisource, so it is has numerous tables of contents. Charles Matthews (talk) 07:12, 22 March 2015 (UTC)

Wikidata:Notability says "On Wikisource, items for mainspace pages, ... are valid, ... The status of subpages of mainspace pages (for example, individual chapters) is undetermined." but why these DNB pages are named "xxx (DNB00)" instead of "Dictionary of National Biography, 1885-1900/xxx"?--GZWDer (talk) 11:40, 22 March 2015 (UTC)
The use of a suffix rather than a prefix for the DNB was decided in 2008, before I was active on the project. It does have some advantages over subpage style, for those typing in page titles, which I do all the time. It is surely a side issue in this thread. Charles Matthews (talk) 16:46, 22 March 2015 (UTC)
Charles Matthews, GZWDer, @Andrew Gray:, Jura the situation, as I see it, is as follows:
  • Wikispurce can have multiple pages about the same person, taken from different encyclopedic texts.
  • The Wikidata software does not allow us to have sitelinks to all these pages from the one wikidata item about this person so we have to have multiple wikidata items, each with a sitelink to a different wikisource page about that person.
  • we would like there to be an easy way to put a template on pages related to this person linking to the other pages about this person.
  • To do this we need a property which can be used to create a statement linking from the main item about a person to all the wikisource pages about the same person.
  • We may also need a symmetric property to put statements on the wikidata item for the wikisource pages and link this back to the main item. This can then be used to create a disambiguation template, to put on the wikisource page, linking to all the other wikisource pages about this person.
Is that right? What should we call these new properties? I would like it to be more generic than "DNB entry at wikisource" so it can be used for other reference works besides DNB. Could we use described by source (P1343) for this property or do we need a new property? Filceolaire (talk) 16:26, 22 March 2015 (UTC)
In that (expansive) scenario, using main subject (P921) to link one way, and described by source (P1343) the other, looks to me like a comfortable solution. I suppose that from the point of view of harvesting disambiguation content on Wikisource, the URL on Wikisource needs to go in the described by source (P1343) entry. Charles Matthews (talk) 16:57, 22 March 2015 (UTC)
described by source (P1343) doesn't seem suitable to identify unique items for specific DNB entries. --- Jura 12:54, 23 March 2015 (UTC)
there is also another problem here… one wikisource (en, for now) may have numerous biographic items from different biographical dictionaries, but the same person may also appear in more numerous biographical dictionaries in many other languages... on fr, we have a "Dictionnaire" project, that groups more than 250 dictionaries, among them, a lot are biographical collections. A system should be designed to allow linking to all articles about the same person, on every wikisource… that cannot certainly be achieved through a wikidata linking for every biographical article… and those articles CAN NOT be considered as Q5 - which many of them have been marked by bots… a reflexion between wikidata and wikisources projects should be engaged to find a satisfactory solution for both projects, before creating thousands of completely erratic items... also, those articles are often used as "source" for wp articles… --Hsarrazin (talk) 21:46, 24 March 2015 (UTC)
Here we go: Wikidata:WikiProject DNB. --- Jura 11:57, 25 March 2015 (UTC)

How to sorting statements?

Help:Statements#Sorting statements seems not be working. --Diwas (talk) 21:32, 24 March 2015 (UTC)

Sorting statements was disabled as the code was not very good, so far as I'm aware. --Izno (talk) 22:13, 24 March 2015 (UTC)
see Wikidata:Contact the development team#Why are the triangles vanished --Oursana (talk) 16:32, 25 March 2015 (UTC)

Discussion on en.wiki

This is a notifcation that there is a disucssion of wikidata at en.wiki's Reliable Sources Noticeboard. See https://en.wikipedia.org/wiki/Wikipedia:Reliable_sources/Noticeboard#Is_Wikidata_a_.27reliable_source.27_.3F Stuartyeates (talk) 08:39, 25 March 2015 (UTC)

Replied. TomT0m (talk) 12:05, 25 March 2015 (UTC)

Photographer

Hi all, I just changed the item for photographer (Q33231) to be a subclass of artist, not author. Does anyone know why it was set to author? Thanks, Jane023 (talk) 17:49, 24 March 2015 (UTC)

Since this edit by Andreasmperu (there had been no 'subclass of' before). Matěj Suchánek (talk) 18:04, 24 March 2015 (UTC)
OK thanks - maybe he meant creator not author, which is also possible. What is the difference between creator and artist in this hierarchy anyway? Jane023 (talk) 18:46, 24 March 2015 (UTC)
creator (Q2500638) has subclasses artist (Q483501) and author (Q482980) (and a few more). My point of view: artists' field is art, autors' field is written works. Matěj Suchánek (talk) 19:38, 24 March 2015 (UTC)
Got it, thanks - I totally agree. Jane023 (talk) 08:35, 25 March 2015 (UTC)
Creator would be better; some photopgrahy is art, other (medical imaging, astro-photgraphy, etc.) is just pressing a button. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 09:31, 26 March 2015 (UTC)
All linked articles seem to be concerned with photographers as performers of a craft, often after absolving a vocational education (there are other professionals too, like journalists and architects, whose output is generally protected by copyright law but we would not necessarily view upon them primarily as artists). -- Gymel (talk) 09:58, 26 March 2015 (UTC)

next step for header redesign and bug fixes are live

Hey folks :)

As announced earlier we've just deployed the next step towards the new header design. We're not there yet but this is the next step. This will allow you to collapse the in other languages box for example and adds a hint about how to configure the displayed languages.

In addition to the new header the next deployment will bring a lot of under-the-hood changes and bug fixes. The most relevant changes for you are:

  • we made the diff for time values more meaningful
  • we fixed a lot of bugs in the time datatype
  • edit links are no longer cached incorrectly based on the users permission (This lead to users sometimes seeing edit buttons on pages that they could not edit and no edit buttons on pages that they could edit.)
  • we fixed some issues with propagating page moves and deletions on the clients (Wikipedia, etc) to Wikidata
  • we corrected an issue where you would see new data in the old part of a diff (This affected qualifiers mainly.)
  • the sitetointerwiki gadget now also works on diff pages
  • the precision is now detected correctly when entering a quantity in scientific notation
  • we added mailto as an accepted protocol for the URL datatype

Please let me know if you encounter any issues. --Lydia Pintscher (WMDE) (talk) 18:35, 24 March 2015 (UTC)

@Lydia Pintscher (WMDE): Look at some edit summaries in logs like RecentChanges or Contributions. Previously you could see labels of changed properties (instance of (P31)), now you see just P31. Matěj Suchánek (talk) 20:03, 24 March 2015 (UTC)
Thanks! I've filed phabricator:T93804 for it. --Lydia Pintscher (WMDE) (talk) 20:10, 24 March 2015 (UTC)
Fix for this is being worked on right now. Should go live in the next hours. --Lydia Pintscher (WMDE) (talk) 09:49, 25 March 2015 (UTC)
Great, thanks. Matěj Suchánek (talk) 14:31, 25 March 2015 (UTC)
  • Too many usability and design bugs...
  •  
    Duplicate information
  •  
    Separators join with items
  •  
    Languages unknown for me
  •  
    Too small area
  •  
    Slow opening and closing
  •  
    Invalid virtual keyboard icon location
Lidia, could you say when development team fix bugs and redesign sitelinks section? Please, it is too hard to work with it now. — Ivan A. Krestinin (talk) 20:07, 24 March 2015 (UTC)
Thanks. I'll look into these. I don't understand the issue with the second image. Can you please clarify it for me?
Sitelinks section has gotten fixes and will get more. --Lydia Pintscher (WMDE) (talk) 20:14, 24 March 2015 (UTC)
Items in aliases section are not separated enough. All aliases are looked like single non-separated text line. Pipes are too small separators in line with uppercased letters. — Ivan A. Krestinin (talk) 20:30, 24 March 2015 (UTC)
Ah ok. Thanks. Now I understand. Will see what we can do to make that more obvious. --Lydia Pintscher (WMDE) (talk) 20:33, 24 March 2015 (UTC)
Good work on the redesign and bug fixes. I'm glad to see a "configure" link next to the "In more languages" section. I was hopeful that you had finally included all languages in this section too, as the collapsible section implies a large amount of information might be inside. But I was disappointed that most languages still remain completely hidden from the user (without any clue given to their existence), creating a confusing and incomplete experience for the user. At the end of the "In more languages" section there still ought to be a "more" button to reveal the other languages, or simply include all the languages in the table (as it's now collapsed by default anyway). User-preferred languages might go at the top of the list, and/or be highlighted. Glad to see some visible development work on a Wikimedia site all the same, and thanks all involved. Pengo (talk) 20:37, 24 March 2015 (UTC)
Thanks, Pengo! You are absolutely right. This is the way I want to go and we do have a ticket for it at phabricator:T92759. There is still one open question that you might want to comment on there. --Lydia Pintscher (WMDE) (talk) 20:42, 24 March 2015 (UTC)
Is it a bug or a feature that when I click on a label or an alias in the collapsible table, the value is removed? It doesn't happen all the time, but occasionally. —Wylve (talk) 20:56, 24 March 2015 (UTC)
That does sound like a bug indeed. I will try to reproduce it. --Lydia Pintscher (WMDE) (talk) 21:00, 24 March 2015 (UTC)
I have tried to reproduce this and it does not happen for me. Which browser are you using? Is anyone else having the same issue? --Lydia Pintscher (WMDE) (talk) 09:56, 25 March 2015 (UTC)
I'm using Firefox 36. I looked more closely at the symptoms and it appears that when I click on the label (this applies to descriptions and aliases as well) it is highlighted for a brief half-second then the entire text disappears. If I highlight a label then switch tabs and switch back to the tab containing the item, the label disappears. This only applies to unsaved changes. Hope this helps. —Wylve (talk) 10:34, 25 March 2015 (UTC)
Mpfh. No that still works fine for me. I will try it on some other computers here. --Lydia Pintscher (WMDE) (talk) 15:10, 25 March 2015 (UTC)
Thanks Lydia. Also, is it possible to make the languages appear in the collapsible table in the order I put them in my babel box? —Wylve (talk) 21:05, 24 March 2015 (UTC)
What's other people's opinion on this? It is probably possible (but I'd need to ask to be sure if more people think this is useful.) --Lydia Pintscher (WMDE) (talk) 21:09, 24 March 2015 (UTC)
Seems like a good idea to me. In my case, I naturally ordered my own babel boxes by knowledge/interest, as I suspect most people would. It also makes it more configurable (which the user must be interested in to have bothered to add babel boxes in the first place). Pengo (talk) 22:29, 24 March 2015 (UTC)
Ok I have filed a ticket to track it at phabricator:T75654. --Lydia Pintscher (WMDE) (talk) 10:00, 25 March 2015 (UTC)
Another rather interesting bug: when I rollback an edit I see a message informing me about successful rollback (that's normal). However, there is also another irrelevant message about some revisions being unavailable. Matěj Suchánek (talk) 21:21, 24 March 2015 (UTC)
Yep, I see the same:
Reverted edits by xx.xxx.xxx (talk | block); changed back to last revision by Stryn (talk | contribs | block).
Return to blaa blaa blaa.
2 revisions of this difference (123 and 456) were not found.
This is usually caused by following an outdated diff link to a page that has been deleted. Details can be found in the deletion log.
--Stryn (talk) 21:32, 24 March 2015 (UTC)
Bad. I have filed phabricator:T93866 for this. --Lydia Pintscher (WMDE) (talk) 10:08, 25 March 2015 (UTC)
@Ivan A. Krestinin: you should create a user page with something like {{#babel:ru|en-3|....}} so the software knows what languages you want to see. Multichill (talk) 21:24, 24 March 2015 (UTC)
Very strange... Base language is configured in settings. Additional languages are configured in absolutely another place. This is looked like usability bug. {{#babel:...}} is used to show known language set to another users. But description section require languages used for working. — Ivan A. Krestinin (talk) 21:58, 24 March 2015 (UTC)
another edit summary bug: this edit summary has two superfluous brackets. --Pasleim (talk) 23:21, 24 March 2015 (UTC)
@Lydia Pintscher (WMDE): And this one? Matěj Suchánek (talk) 14:31, 25 March 2015 (UTC)
I missed this. Thanks. I filed phabricator:T93896 for it. --Lydia Pintscher (WMDE) (talk) 15:09, 25 March 2015 (UTC)
@Pengo: You can activate LabelLister gadget in your user preferences to see and edit all labels, descriptions and aliases in all languages. I dunno why this is not activated by default fo all users. The language presets is defined in three different ways: 1. Language selector, 2. language fallback chain (assistant languages) 3. babels on user page. There should be one spot to define everything for improved usability.--Giftzwerg 88 (talk) 22:23, 24 March 2015 (UTC)

Descriptions

Today I cannot fill in descriptions. Pages that have No description defined yet and No aliases defined are missing edit buttons for descriptions and aliases. How do I circumnavigate this? --Gereon K. (talk) 18:38, 24 March 2015 (UTC)

That was really fast :) Click edit next to the main label. Matěj Suchánek (talk) 18:56, 24 March 2015 (UTC)
Thanks. Worked for me. :) --Gereon K. (talk) 20:18, 24 March 2015 (UTC)

Changes

The way pages present seems to have changed with regards different languages, is there anyway we can change it back to the way it was in personal preferences.--KTo288 (talk) 20:11, 24 March 2015 (UTC)

Can you please let me know what you don't like about it? (It was announced here and available for testing.) --Lydia Pintscher (WMDE) (talk) 20:15, 24 March 2015 (UTC)
Conservatism I guess, I've grown used to a page looking the way it did, and working with it as it was. Sorry but I normally don't visit this page. The first I knew of the change was 30 minutes ago. For me, I guess the best way to describe it is the irritation you feel when you sit at your desk and reach out for something and not find that its not in its normal place, then look and see that your desk has been rearranged, logically no doubt, but not in the way you're used to.--KTo288 (talk) 20:36, 24 March 2015 (UTC)
I just discovered it too. It's great that this languages part of an entry takes less place, but I am a bit bothered to see that long labels/description are cut... At the very least, I'd like to be able to read them in full in a tooltip on mouseover. For the "In more languages" button, I think a nice evolution would be to have it under the languages box, to show the languages that are not in my babel list (for example, I don't speak Spanish, but it would sometimes be useful for me to see if there is a label in that language), and maybe along with another link : "Add a language" to add a label/description in a language that is not yet on the list... -Ash Crow (talk) 00:10, 25 March 2015 (UTC)
For the too short descriptions we now have phabricator:T93807. For allowing to edit more languages we have phabricator:T92759. --Lydia Pintscher (WMDE) (talk) 10:13, 25 March 2015 (UTC)
What on earth are we supposed to do with those, install what, where, how?--08:09, 26 March 2015 (UTC)

New GUI broke several tools

  • Since the deployement of the new GUI, it seems that the Game answers do not appear in items anymore... no use adding genre to an item, it's just not taken into account :(
  •  
    screenshot of my recent edits in the Game
  • also User:Joern/altLabels.js does not appear any more… was quite useful.
    ping Joern. Matěj Suchánek (talk) 14:31, 25 March 2015 (UTC)
  • claims added through Wikidata useful do not appear with re-loading the page manually, which may lead to multiple addition, or worse...
  • it should be possible to edit directly the "main" language fields, like before…, and allow the "drag-and-drop"... - very convenient from the "Preview" gadget... this table, table, as it is, is not logical, not easy to use, and much too small for long labels and tired eyes... --Hsarrazin (talk) 20:51, 24 March 2015 (UTC)
also, the recently redesigned rule, that allowed direct access to project links has disappeared... it's not useful anymore for accessing claims, but for sister projects, it still is ;) --Hsarrazin (talk) 23:36, 24 March 2015 (UTC)

Suggested target items menu

Not sure if this is related, but today when fixing a statement for P21, I typed "male" and Q6581079 did not appear at all. --Haplology (talk) 02:00, 25 March 2015 (UTC)

I can reproduce it. We're looking into this as well. --Lydia Pintscher (WMDE) (talk) 13:48, 26 March 2015 (UTC)

Comment for changed label

Tracked in Phabricator
Task T93853

Today the last one from me: this edit summary is wrong (I have seen more, they all were changes of labels). Matěj Suchánek (talk) 21:49, 24 March 2015 (UTC)

A fix for this is being worked on right now and should go live in the next hours. Tracking at phabricator:T93853. --Lydia Pintscher (WMDE) (talk) 10:10, 25 March 2015 (UTC)

An anonymous user vandalized Queen Victoria by renaming her, but the change showed up as "Created a new item" --Haplology (talk) 03:49, 25 March 2015 (UTC)

I was able to reproduce this when editing via a special page. Created a new task for this issue. Aude (talk) 07:22, 25 March 2015 (UTC)

Wikidata property (P1687)

Is it ok to link more than one Item in Wikidata property (P1687)? Conny (talk) 16:51, 25 March 2015 (UTC).

RKDartists (Q17299517) seems a better place for one of the properties. On Netherlands Institute for Art History (Q758610), you might want to use "see also" instead. --- Jura 19:35, 25 March 2015 (UTC)

No Category Assignments on Wikidata?

Category assignments (article<category and category<category) are not in Wikidata.

  • They are present on DBpedia (dct:subject and skos:broader respectively).
  • Is there a particular reason why they are not included in Wikidata?
  • In contrast, the topical relation article~category is in Wikidata (and is skos:subject in DBpedia dump files)

We need them for Europeana Food and Drink. --Vladimir Alexiev (talk) 04:11, 30 March 2015 (UTC)

Because those things are not always the same in every Wikimedia project. Sjoerd de Bruin (talk) 05:36, 30 March 2015 (UTC)
@Sjoerddebruin: So what? Assume article A1 on enwiki has cats C1,C2; article A2 on dewiki has cats C3,C4. Assume A1=A2 and C1=C3 but C2<>C4. Then A1=A2 will get 3 cats: C1=C3, C2, C4. Don't see what is the problem. --Vladimir Alexiev (talk) 17:16, 30 March 2015 (UTC)
@Vladimir Alexiev: I don't think we need category assignments in wikidata. Category is no knowledge but an arbitrary way to sort items that depends on the wikimedia project. It would not be possible to source that kind of statement and it would generate a lot of repetitions with the other statements. --Casper Tinan (talk) 20:29, 30 March 2015 (UTC)
@Casper Tinan: Categories are most assuredly knowledge. A bit messy, lacking in organization (true) but very comprehensive. Wikipedians are very serious about their categorization (see numbers in the linked report). We need them for http://vladimiralexiev.github.io/pubs/Europeana-Food-and-Drink-Classification-Scheme-(D2.2).pdf, or how else would you delineate a domain as wide as Food and Drink and its reflection in Culture? We also don't source Labels and Descriptions, but do you think you could live without them? As for "duplication", I am not sure what you mean. Cheers! --Vladimir Alexiev (talk) 22:16, 30 March 2015 (UTC)

See Help:Basic membership properties, topic's main category (P910) and Wikidata:Property proposal/Generic#Parent category. Visite fortuitement prolongée (talk) 20:14, 31 March 2015 (UTC)

Possible mistakes in Wikidata

Hello, As you may know I wrote Kian, the first artificial neural network to serve Wikidata, I used results of Kian to report possible mistakes in Wikidata based on Wikipedia. These are two reports based on French and Dutch Wikipedia. For example: 

Q425430: 1 (d), 0.198071051892 (w) [0, 0, 8, 3, 0]
1 (d) means Wikidata thinks it's a human (it has P31:Q5 statement).
0.198071051892 (w) means French Wikipedia thinks it's not a human with 80.2% certainty.
Other parts are debugging stuff. and as you can see French Wikipedia is right. Note that some results may be outdated.

You can easily check them and fix problems. User:Sjoerddebruin: I think you like this.

If you want please tell me and I run this for other languages. Amir (talk) 16:21, 20 March 2015 (UTC)

Yes! A German list would be pretty cool! --FischX (talk) 20:07, 20 March 2015 (UTC)
Do I understand it correctly that, for now, this is checking only P31:Q5 for each of the given Qids? --18:22, 20 March 2015 (UTC)
How is anyone supposed to make sense of these codes? The only part of "Q35820: 0 (d), 0.93274580186 (w) [5, 0, 8, 4, 0]" that I understand is "Q35820" Kaldari (talk) 08:44, 21 March 2015 (UTC)


@FischX: User:Ladsgroup/Kian/Possible mistakes/de. @Kaldari: Yes, you are right. I created a more-human readable version in User:Ladsgroup/Kian/Possible mistakes for all languages. What do you think? Amir (talk) 14:01, 21 March 2015 (UTC)

Thanks, thats great! --FischX (talk) 21:51, 22 March 2015 (UTC)

This is result for English Wikipedia. I did some more advanced feature engineering and the results are pretty damn accurate. Please have a look and fix items. Thanks Amir (talk) 14:50, 23 March 2015 (UTC)


Good evening,

Editing genders frow the Game, I found today an awful lot of items wrongly marked as Q5… (never seen so many incorrect markings since I cleaned up the P107 (P107) a few months ago)… and when I checked the historic, I found "Dexbot (discussion | contributions)‎ . . (1 093 octets) (+621)‎ . . (‎Affirmation ajoutée : Adding P31:Q5 from English Wikipedia (Powered by Kian)) "

a lot of them had that kind of edit summary, so I guess, there must be something twisted in the way the bot has been "powered" by Kian… a lot of these items are music groups, music, films, books, families, duos, firms, but also lists of people and even ethnical groups !!

Ladsgroup, could you please check how Kian works… I usually take the time to manually edit that kind of errors, but there are so many that today, I just flagged them… and I obviously was not alone… Wikidata:The_Game/Flagged_items.

Thanks --Hsarrazin (talk) 23:53, 24 March 2015 (UTC)

Hey, Kian becomes more accurate as the time goes by (It has high precision and recall even now) It works based on categories and in some cases pretty bad categorization is responsible for the error. e.g. w:Tamil Trinity. I check the errors and fix them by hand or by bot. I maybe change the system to to use w:tf-idf of certain words and analyze based on this. Anyway thank you for informing me. Amir (talk) 02:55, 25 March 2015 (UTC)

@Hsarrazin: Let me give a prespective on how accurate Kian is. I added about 50K P31:Q5 statements based on Kian and 500 of are identified as error. It's 99% and It's a great number in AI, even though we should make it more precise as the scale is so huge. Another way of seeing accuracy of Kian is to see how many errors it detected and compare it on how many errors it created. about 3500 errors are detected by Kian (Only based on English Wikipedia and not other languages). It's seven times more than errors it created. What do you think? Amir (talk) 03:21, 25 March 2015 (UTC)

well, when working on huge numbers, 1% is a rather high number of errors... ;) - I'm really conscious that Kian is in development, and must improve... I just wanted to inform you of some types of very usual mistakes... working through categories explains a lot about the errors, if you include subcats - all works from a creator being categorized as Q5 ,like Bach's)... LOL
personally, I tend to prefer semi-automated work, like Magnus's tools, that make an educated guess about an item, but lets a user decide if that guess is right or wrong... :)
one thing, IMHO, worsens the automatic work.... Kian's claims are "sourced", which tends to make them more reliable in the eyes of a patroller... do you add a special tag, so that those edits can be specifically patrolled for mistakes detection ? - for now, I use the gender-Game to detect such mistakes, which are easily detected there...
thank you for that tool, that, at least, unburied old items that were not identified as humans because of strange names...
keep up the good job, Amir, and maybe, if you could put on a page how we can help you improve Kian ? --Hsarrazin (talk) 08:45, 26 March 2015 (UTC)

Thanks, In order to improve Kian, please flag errors in the wikidata game or here Amir (talk) 12:06, 27 March 2015 (UTC)

Inferences and redundancy

Is Wikidata able to make inferences about statements, or redundant statements need to be made?

For example: if there is a statement listing occupation as "politician", this implies that the subject is human. So, saying that the item in question is an instance of human should ideally be redundant. (If I understand correctly, the same applies to sex/gender: "male" implies "human" - there is "male animal" non-human sex/gender too.) What I'm seeing though are logically redundant statements about items: a "politician" who is also an instance of "human". Is this supposed to work differently in the future? GregorB (talk) 18:12, 25 March 2015 (UTC)

@GregorB: Wikibase does not make any inference at all. It's basically just a store and a user interface atm. There is atm. no plan to implement this see Wikidata:Development plan, the query engine who will replace {{WDQ}} is far higher on the priorities, and words lately has been it will be SPARQL based on the backend at least. But the door to inferences has not been closed according to User:Denny, one of the creator of wikidata in an earlier discussion. But not in the forseable future.
But if the software itself can't do it, the multiple clients might be smarter if community decides some rules about this :) TomT0m (talk) 19:24, 25 March 2015 (UTC)
Thanks! I suppose the best way to get rid of redundancy would be to enhance the query engine with inferential capabilities. So, when asked "is John Doe a human?" the engine would say "yes" if John Doe is a politician.
Also, I suppose it's currently possible to create contradictory statements: e.g. one cannot be both a "human" and a "cat". Hopefully down the road Wikidata will have the ability to reject (or at least detect) these. GregorB (talk) 20:08, 25 March 2015 (UTC)
Fortunately, some contradictory statements are detected already: Wikidata:Constraint violation report input. GregorB (talk) 20:49, 25 March 2015 (UTC)
I would assert that a politician may not always be a human. How? There are fictional characters whom we might want to assign an occupation and those persons are not necessarily human. As well, there are animals increasingly (according to certain countries legislatures and their courts) being treated as a "person" for the purpose of certain classes of things. Now, I find it unlikely that there should ever be a non-human person who ends up with an occupation ("showanimal" might be an interesting one that exists today)... Right now, we have property constraint violations as you have yourself noted, and so we fix the ones that actually are violations and leave the others as-is... But we have to add the constraints ourselves, of course. Which would probably be no different than if we were able to do this under the hood of Wikibase rather than over-the-hood with a bot. --Izno (talk) 23:54, 25 March 2015 (UTC)
Kellessar zh'Tarash is a politician and not human! And she is not female, at least not a female of a kind we are used to. She is a zhen, one of four genders among the Andorians. -- Innocent bystander (talk) 07:13, 26 March 2015 (UTC)
...and there is a character in Spaceballs who is both a human and a dog. :-) This is not a showstopper, it's just that certain inferential rules do not extend to fictional characters. So, if X is a politician and is not a fictional character, then X is human - otherwise, all bets are off. :-) GregorB (talk) 10:13, 26 March 2015 (UTC)
I do not think you have to be fictional to have heavy positions in the society. Incitatus (Q935895) for example! -- Innocent bystander (talk) 10:31, 26 March 2015 (UTC)
Why look to fiction? Catmando (Q1050083), politician and party leader of the Official Monster Raving Loony Party (Q1421227), is a nonfictional cat. For other examples, see non-human electoral candidate (Q2740853). --Yair rand (talk) 20:51, 26 March 2015 (UTC)
@Izno: Of i guess this inference would have a guard, forming a logical implication (Q7881229)      to be made only when the stuff is not fictional :) If we assume every inference on humanity has to be safe in a fictional world, we won't get anywhere :) TomT0m (talk) 11:01, 26 March 2015 (UTC)
Stubbs (Q7627362) --Denny (talk) 18:32, 26 March 2015 (UTC)
@Denny: Thanks for reminding me why i'm translating paraconsistent logic (Q426592)      in french :) TomT0m (talk) 11:04, 27 March 2015 (UTC)
For me the big redundancy we should get rid of is the cases where we have inverse properties like 'part of' and 'has part' but first we need the user interface to show the statements which link to an item as well as the statements which link from it. Filceolaire (talk) 17:05, 26 March 2015 (UTC)
I don't really think it is a case of big redundancy. Duplicating some artist anthology several times in different places would probably be way harder to maintain and to model with constraints, for example. TomT0m (talk) 11:09, 27 March 2015 (UTC)

Multiple coordinate location (P625) in the same item

Some Wikidata items have ambiguous coordinate location (P625) values for the same geographic location. It does not make any sense and in addition multiple coordinates cannot be displayed on one Wikipedia page and they break down templates that use global Wikidata item properties instead of locally defined coordinates. For instance, item Q15992735 has two different coordinate locations that breaks down the template in uk:3-я дільниця (see {{#coordinates:}}: error message there). We've got in total at least 43 such cases and thus we are hesitant to use coordinate location (P625) instead of local data for other templates. What solution would you recommend? Do we have to stick to local data for now? And what to do existing problem pages? --Pavlo Chemist (talk) 10:09, 26 March 2015 (UTC)

You can design the templates to import only the proffered value or only one value if there are more than one of the same rank. -- Innocent bystander (talk) 10:35, 26 March 2015 (UTC)
In the case of 3rd Uchastok (Q15992735) it should be a single value, there were just two values imported from two different sources. In this case, I don't see any problem to delete the less accurate value, and probably the other cases which cause your problems are similar. However, multiple locations is something which could happen, think of a village located next to a river and rebuilt at a new location after destroyed in a big flooding - then the two location values should have qualifiers to mark the time at which each of the statements was valid. Or some external sources may give a bogus location, and to avoid confusion we add it here with the "deprecated rank". Ahoerstemeier (talk) 10:45, 26 March 2015 (UTC)
Thanks for explanation, I see the point. BTW, can anybody please show me an example how to choose only one value out of several (the first one, with higher priority and by qualifier)? --Pavlo Chemist (talk) 11:23, 26 March 2015 (UTC)
You have to look at your module:Wikidata page ~in your Wp to see how to perform rank sorting. This is not a common code: each WP develops its code according to its need. Snipre (talk) 13:03, 26 March 2015 (UTC)
If a property has several values and only one of them is marked as rank preferred then by default Lua should only give you back the one marked as preferred. That should be the easiest way to solve the issue of items where multiple coordinates are actually ok to have. --Lydia Pintscher (WMDE) (talk) 13:44, 26 March 2015 (UTC)
Thanks for additional clarifications. Indeed, using "deprecated rank" actually works. --Pavlo Chemist (talk) 14:17, 26 March 2015 (UTC)
My understanding is that where are multiple values then the current value or the more accurate value should have 'preferred rank'. Values which were correct at some time have 'normal rank'. Values which are wrong but have been mentioned in sources or widely used should be 'deprecated rank' but left in so they don't get re-added in error. Wrong values which don't have references should not be included. The software should ensure wikipedia templates only get the preferred value. Filceolaire (talk) 16:55, 26 March 2015 (UTC)
It seems that es.wikipedia is now almost exclusively using Wikidata for coordinates, I suppose it appeared to be usable enough. :)
@Pavlo Chemist. you template should still be able to cope with cases where there are two preferred values, or tow normal values and no preferred value. It is best done through a Lua module. The simplest solution is probably a "numval" option in a Wikidata-data retrieval module. fr:Module:Wikidata. If you set numval to 1, it only returns the first value that matches your query.
In the case of 3rd Uchastok (Q15992735), I think putting rank = "deprecated" to the value from uk.wikipedia was the right solution: it signalled an issue for Ukranian Wikipedia users, but now that the issue has been fixed in Wikipedia, the statement can probably be deleted altogether in Wikidata. --Zolo (talk) 09:16, 27 March 2015 (UTC)

Strange problem

For some weird reason, my recent Widar edits were marked as "reCh" instead of "Widar". Does anyone know why this might be? --George (Talk · Contribs · CentralAuth · Log) 19:28, 20 March 2015 (UTC)

I got the same problem 2 days ago ... https://www.wikidata.org/w/index.php?title=Q19662808&diff=prev&oldid=206595526 --Hsarrazin (talk) 08:31, 26 March 2015 (UTC)
in fact, I just saw that ALL my recent edits from the Wikidata Game were wrongly tagged Special:Contributions/Hsarrazin (Balise : reCh [1.0])
I don't even know what reCh is, and AFAIK, never used it, but only Widar (since the Game uses Widar) :(( --Hsarrazin (talk) 21:45, 27 March 2015 (UTC)
reCh is the patrol tool I introduced last week. Can you go to https://tools.wmflabs.org/pltools and https://tools.wmflabs.org/widar and logout on both pages and then login again? --Pasleim (talk) 21:51, 27 March 2015 (UTC)
Oh, thanks, I made a try at your tool, Pasleim, but was not clear enough to understand how to use it, then :)
should I login back on both again ? or just widar ? --Hsarrazin (talk) 21:55, 27 March 2015 (UTC)
You can login on both pages and then tell me if the problem still exists. --Pasleim (talk) 21:58, 27 March 2015 (UTC)
seems ok on both sides now… difficult to patrol between very various languages, though :/ - do you think you could add a link/button on "new items" to ask for speedy deletion, when obvious spam, like Q19689142 ? nice tool, just difficult to understand what to do ;) --Hsarrazin (talk) 23:04, 27 March 2015 (UTC)

Disambiguation page descriptions

I'm creating a bot to add descriptions to items for disambiguation pages. I was intending to use the labels of Wikimedia disambiguation page (Q4167410), but it was pointed out that some of those labels would not make very good descriptions. I have made a list of the labels, would anyone be willing to help verify/correct them? User:Popcorndude/botcode3/labels Popcorndude (talk) 01:02, 27 March 2015 (UTC)

The ones on MediaWiki:Gadget-autoEdit.js should be fine (or at least, I had been using them). --- Jura 05:58, 27 March 2015 (UTC)
While most disambiguation pages list articles whose connection is that their names are spelt alike be aware that some disambiguation pages group things that have other connections. These include 'surname' pages and 'given name' pages. Some of these can be recognised because they are the target of 'surname' or 'given name' statements on other items. Where I find these I have added the description 'surname disambiguation page' or 'given name disambiguation page'. Filceolaire (talk) 22:41, 27 March 2015 (UTC)

addressee property?

Is there a property for the addressee of a letter? Hazmat2 (talk) 18:29, 27 March 2015 (UTC)

I'm not aware of it. On Wikidata:Property proposal you can propose new properties. --Pasleim (talk) 21:54, 27 March 2015 (UTC)
Will do. Thanks, Hazmat2 (talk) 23:43, 27 March 2015 (UTC)

What are the Challenges in Paid Media, AdWords

Please share you thoughts...--Mithilesh88 (talk) 06:03, 4 April 2015 (UTC)Mithilesh (Digital Marketer)

How to Model Complex Items (Vectors, Matrices)

To give you some context I want to know what is the best practice to model a time series in wikiData? Is a time axis a group of items? Or is it modeled as a key in a complex type? .. Is an instance of a day an item in the wikiData store (e.g. December 1, 2014: "https://www.wikidata.org/wiki/Q18064193") ? And if yes, who controls the insertion of this category? The general contributors?

So now to the main part of my question ..

What is the best practice to represent items in wikiData if they are of the type e.g.

  • vector
  • or matrix, table

I guess in wikiData every value in the structured object is stored as an item. The remaining question though is how to model the relationships between the items in wikiData to represent the structured object.

Here are my assumptions:

  1. to create an axis one would create an item to represent a group of column headers and one to represent an optional group for the row headers
  2. a duple of column header and row header form the parent of a cell (is there a query type which allows queries like "give me item where [a] is a parent and [b] is a parent"?)

So far so good ...

I guess it will be tricky to model something when both axis' are dates ...

But how would an order relationship in one group be represented in wikiData? e.g. item a >= item b

Is left to the querying client to apply the order relationship?

Regards  – The preceding unsigned comment was added by Smartkatt (talk • contribs).

In the general case for ordering, yes (though timelines are notable non-general case since years are certainly implicitly ordered).
I would avoid trying to model a matrix in Wikidata right now since I can't think of any elegant fashion to do so (nor would I want to, since we don't have the number with dimensions data type, which would be useful for most tables). Vectors (1D matrices) can maybe be modeled since they can be qualified in some ways. --Izno (talk) 03:06, 28 March 2015 (UTC)
@Izno: From the linear algebra perspective, it would not make sense to do just vectors or just matrices. I wouldn't want to restrict the vectors we can model to just the field of real or even complex numbers. So many other things, such as tensors, can form vector spaces. For example, earthquakes' moment tensor solutions would be useful data to store.--Jasper Deng (talk) 05:17, 28 March 2015 (UTC)
I don't see how that relates to what I suggested, whatsoever. I simply stated "I don't think this is a good idea to attempt" and nothing more. I can see modeling vectors, but right now I think a matrix is beyond the power of statements + qualifiers. --Izno (talk) 05:34, 29 March 2015 (UTC)
Where a table contains data with each line a separate datum then there are two ways to include this in wikidata.
If each line is data about a different item then each line can be encoded as a series of statements on the item for that line, with a different property for each column.
If all the lines of the table are related to a single item (say a league table for one season) then each line is a statement (name of participant) and each column is a qualifier to that statement (played, Won, Drawn, Lost, Goals for, Goals against, points).
As long as the same properties/qualifiers are used in the same way then (at some point in the future) it should be possible to create the table from this data, but it should also be possible to create other tables and graphs too. Filceolaire (talk) 13:43, 28 March 2015 (UTC)

Badge needed for wikisource items...

  WikiProject Books has more than 50 participants and couldn't be pinged. Please post on the WikiProject's talk page instead.

Hello, wikisourcians and wikisourcerers on wikidata !

Now that some bot has imported almost every text in wikisource fr (and probably many more wikisources), without minding the ws.fr community and the state of the texts, we need to be able to recognize texts that have been corrected and validated from incomplete texts... we intended to add to wikidata ONLY texts that were in good state, but hey, it's now past :/

I think it would be necessary to have 2 labels, one yellow for texts that have been checked by at least one corrector, and one green for those that have been validated by 2 readers...

I am almost sure there is a Wikiproject for wikisource, but I could not trace it. If someone could ping this group, it would be nice :) --Hsarrazin (talk) 12:29, 28 March 2015 (UTC)

Hsarrazin, Wikidata:Wikisource; I suppose this is what you want. And I concur with badge demand.--Vyom25 (talk) 12:52, 28 March 2015 (UTC)
yes, that was it, the discussion page - since I was looking for a Wikiproject, I just overlooked it  ;)
in fact, I checked icons on fr.ws... we use   for texts that have been corrected andfor texts that have been validated... but other wikisources may use different ones… so, what do you think ?
they should be distinct from wp badges, and be instantly recognized by the ws community… --Hsarrazin (talk) 13:32, 28 March 2015 (UTC)
I added the ordinary badges in Q15635617 (as a test) some time ago, and to me it looks like we would do fine with the same as WP uses! -- Innocent bystander (talk) 13:59, 28 March 2015 (UTC)
don't know wp badges, since I'm not really wikipedian... the gold one would be for "validated" and the silver one for "Corrected" ?
pb stands, the label on the badge is for "quality article" not for "validated text" ;) and those are Quality badges, like medals, wikisource's are more "completion" badges - LOL --Hsarrazin (talk) 15:21, 28 March 2015 (UTC)
The criteria for getting a quality-badge on Wikipedia is unique on every project. On svwp, I know there are three levels. Two of them demands a voting procedure. The third level can be set by anybody without any voting.
You can design how they look locally! -- Innocent bystander (talk) 15:58, 28 March 2015 (UTC)
Yellow and green would be common to every wikisource I suppose. So if that is possible then it would be ideal.--Vyom25 (talk) 06:28, 29 March 2015 (UTC)

Strange on ru sitelink

What is the difference in ru sitelink in Q4035560 and Q4402964? --ValterVB (talk) 22:35, 27 March 2015 (UTC)

seems Q4035560 is latin-alphabet encoded (and would read CAP), and Q4402964 russian-encoded (and would read SAR phonetically)... Q4402964 points to Q4035560 as "see also, latin alphabet"... --Hsarrazin (talk) 23:13, 27 March 2015 (UTC)
I've chnaged ru-links between these items according to phonetical similarity. --Infovarius (talk) 06:11, 30 March 2015 (UTC)

Weblink validation

Hello everyone, there is an ongoing discussion at Wikidata:Forum#Weblinkwartung?! (@Queryzo, @Pasleim) about webblink validation. Is there a bot or a tool which checks routinely whether websites (reference URL (P854), official website (P856)) are reachable (HTTP 200)? If not I could set up such a program maybe as an OAuth-application. --T.seppelt (talk) 15:00, 30 March 2015 (UTC)

List of persons

On it.wiki we have a lot of "List of persons" (persons by name, by surname, by year of Born/Death, for activity etc...). I have create Wikimedia list of persons (Q19692233) and use it with instance of (P31). Is correct or there is a more correct item to use? --ValterVB (talk) 13:09, 29 March 2015 (UTC)

What is the difference compared with is a list of (P360):Person? -- Innocent bystander (talk) 14:54, 29 March 2015 (UTC)
I asked here to know :) I thinked that to have instance of (P31) in these items is good, so I have created Wikimedia list of persons (Q19692233) with subclass of (P279) = Wikimedia list article (Q13406463) then we can add is a list of (P360)=family name (Q101352) or given name (Q202444) date of birth (Q2389905)--ValterVB (talk) 15:45, 29 March 2015 (UTC)
is a list of (P360) is used on many different types of items: on category items, e.g. Category:Malaysian people (Q5912379), on disambiguation pages, e.g. Robert Rose (Q3263), on regular articles, e.g. Counts of Villafranca (Q3951152) and on list articles, e.g. list of architects (Q754679). So having both statements is a list of (P360)=human (Q5) and instance of (P31)=Wikimedia list article (Q13406463) (or a subclass thereof) is okay, as the first statement does not imply the second one. --Pasleim (talk) 17:16, 29 March 2015 (UTC)
Wikimedia list of persons (Q19692233) should be deleted and our standard is a list of (P360)=human (Q5) and instance of (P31)=Wikimedia list article (Q13406463) used. --Izno (talk) 20:37, 29 March 2015 (UTC)
@Izno: I want to recall that instance of (P31) and subclass of (P279) have properties, and that if list of person is a subclass of human list and defined such that it is implied that a list of person instance has also a claim is a list of <human>, then everything is OK in the web ontologies world. We can partly use this property with a query like (see the source code of the page for a more readable form :
either instance of list of humans or with a list of human statement
TomT0m (talk) 15:23, 30 March 2015 (UTC)
List of personhumans isn't a subclass of human. It's a subset, perhaps, but not a subclass, as the properties of a human do not inform us about a group of arbitrary humans. --Izno (talk) 18:52, 30 March 2015 (UTC)
@Izno: my bad, I meant list, the query does not make the error ... TomT0m (talk) 10:16, 31 March 2015 (UTC)
Most wikipedia list articles are, in ontological terms, a class (i.e. a 'set'). As such we should do as follows:
  1. the name of the wikidata item should be editted to remove the words"List of" from the wikidata item name though "List of" should be preserved in an alias and in the wikipedia article name.
  2. add a "subclass of" statement
At least that is my opinion. Filceolaire (talk) 20:28, 30 March 2015 (UTC)

Sitelinks for meta?

Any idea when we can expect sitelinks for meta to be implemented? --SamB (talk) 23:47, 30 March 2015 (UTC)

Watch Meta. ;) An official ETA is not yet available as far as I know. --George (Talk · Contribs · CentralAuth · Log) 06:00, 31 March 2015 (UTC)

Declension/inflection of items

In some languages, an item's label is its "canonical name", which usually is the name in the masculine nominative singular case. How do we add other cases to it? I couldn't find it anywhere in the help.

That's something we need to start preferring Wikidata's data in Spanish Wikipedia's infoboxes. For example, the occupation property should be inflected according to the subject's gender, etc... --Angus (talk) 13:11, 31 March 2015 (UTC)

That is a very language- and culture-specific question, I do not know an answer to and I do not see any simple solution.
On svwp there are almost daily edit wars about if a gender-neutral form should be used, and how a gender-neutral form looks like. It may differ with time (teachers in 19th century had a gender-specific education) and if the gender directly affects the kind of work is done within a specific occupation. (actor/dancer) It may also be affected if the feminine and masculine versions of a word changes the meaning of it. (nurse female=somatic care/nurse male=mental care) There are also many cases where the female versions of some occupations/titles mainly apply to the wife of somebody with a certain occupation/title. If the person instead would be a woman, a gender-free version is used. (doctor/doctoress) -- Innocent bystander (talk) 13:56, 31 March 2015 (UTC)
I think that has to wait for adding Wiktionary data to Wikidata. Remind the dev team that this is needed and asked for :) --Denny (talk) 15:12, 31 March 2015 (UTC)

It's a good think that Wiktionary declension data is going to be added, but how? It is something that pertains to the label, not the item per se.

I think the localized labels could be items themselves. This way, every language's label can have a model ("Spanish label", etc.) with the necessary properties, from which the concrete labels would inherit. Every language/culture would have the set of properties that makes more sense to it. This allows not only for grammatical declensions, but also for other considerations, like the ones Innocent bystander talks about. Of course, all labels would ultimately descend from an "Item label" item with the three properties labels have now, so that data will stay.

What do you all think?

See Wikidata:Development_plan#Wiktionary_support. Wiktionary's item won't be the items about the notion, they will be items about the word (the label). Then declanation will be possible to add with properties we will create I guess. And meaning will link concept items to term items, I think TomT0m (talk) 18:57, 31 March 2015 (UTC)

Qualifying properties of places

Just a quick question, because I haven't seen it anywhere. Places are supposed to be qualified by period, right?

For example, Prague has property "Country: Czech Republic". That should be qualified with "Since: 1993", and there should be another country, "Czechoslovakia", qualified for the period 1918-1993, and so on.

Is all that information just missing everywhere, or is it supposed not to be added? --Angus (talk) 19:57, 31 March 2015 (UTC)

I suppose it's missing here, Wikidata can indeed show historical information. Country data is added by simple imports only reflecting the current state. I hope it will soon be imported to Wikidata, so we will be able to show in an infobox that someone was born in Prague, Czechoslovakia, according to their birth data. Matěj Suchánek (talk) 20:13, 31 March 2015 (UTC)
Retrieved from "https://www.wikidata.org/w/index.php?title=Wikidata:Project_chat/Archive/2015/03&oldid=398493706"