Wikidata:Project chat/Archive/2014/05

From Wikidata
Jump to: navigation, search

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

Contents

How do I indicate the representative of an electoral district

On the person who is elected, you want to say that he represents a district. On the district you want to say that the person is the incumbent.. How do I do this.

This is for a project of members of the Lok Sabha.. Thanks, GerardM (talk) 11:09, 24 April 2014 (UTC)

For the item about the person, you can add:
position held (P39): "Member of Lok Sabha 200X-200Y"
electoral district (P768): "Name of district"
start time (P580): Whatever
There are proposals for propertys for "Representing" for such things as political partys etc. (The political party-property does not always work, since people can represent groups they are not personally members of.)
There are discussions about how elections should be described, we have not fully agreed about that yet, mainly since I guess elections can be made in so many ways.
I have also been aware of that not all persons are directly elected. Many in the Swedish parliments are people who are replacing others who are retired from the politics or are temporary out of duty. Such persons also have to be described, but since they are not directly elected, they cannot simply be mentioned in the election-item. -- Innocent bystander (talk) (The user previously known as Lavallen) 11:51, 24 April 2014 (UTC)
I think the property electoral district (P768) should have a tandem item. There are position with name of place that have representatives in numerical district. For example electoral district (P768) = Cebu but it is for the 5th district of Cebu. Cebu is an item but the ordinal district is just a text. --Napoleon.tan (talk) 15:42, 24 April 2014 (UTC)
Why not create an item for "Cebu 5"? See Alabama's 7th congressional district (Q4705162) -- Innocent bystander (talk) (The user previously known as Lavallen) 16:02, 24 April 2014 (UTC)
I know that is an option but it would require creating a lot of entity that is not that much notable. Why not created district ordinal property? --Napoleon.tan (talk) 16:39, 24 April 2014 (UTC)
Feel free to propose one, but I consider them as very notable.
Another Q is how we connect the electoral districts with the administrative entities they belong to?
"Cebu 5": applies to jurisdiction (P1001) "Cebu"
"Cebu": electoral district (P768): "Cebu 1", "Cebu 2" etc...
?
But then you have to define which election the "Cebu X":"Cebu"-relation is related to.
The Swedish municipality of Kramfors for example belong to the "electoral district of Västernorrland county" in the national election, it belongs to the "electoral district of Sweden" in the EU-election, the "Ådalen electoral district" in the county counsil-election and there is only one electoral district in the municipal elections. -- Innocent bystander (talk) (The user previously known as Lavallen) 18:41, 24 April 2014 (UTC)
I agree that there should be an item for "Cebu 5". It has a start date, boundaries etc. At a more local level I believe that creating items for local council electoral wards is only necessary if you want to record the results for each ward. If we are just recording the overall result - total votes and seats for each party - then we only need an item for the overall council. In my opinion. Filceolaire (talk) 22:56, 24 April 2014 (UTC)
Okay, I guess creating district would be ok but the reason I said this is electoral districts changes drastically too and is not much documented in my country. That is why I would just select province and then electoral district ordinal/sub district name. --Napoleon.tan (talk) 00:56, 25 April 2014 (UTC)
I think that is the case in most countries. Here in Sweden, they do not change very often, but the number of seats related to each district changes every election. -- Innocent bystander (talk) (The user previously known as Lavallen) 08:21, 1 May 2014 (UTC)

Bug?

I recently created Q16740876 with mistakenly added Gujarati title. When I tried to remove it, the first letter can not be removed. I added English title Vinod Bhatt then. Can anyone look at it? I think its something like bug. Regards -Nizil Shah (talk) 09:03, 1 May 2014 (UTC)

I had no problems. Needed to press the backspace button twice to remove the letter. --Stryn (talk) 09:06, 1 May 2014 (UTC)

Gadget development

Hi, I'm starting to modify a little gadget. I know not a lot about it, so I'll share my experience here and start a helpage on gadget coding help page Help:Gadget Coding in which I'll detail my progresses, my usecase and compile the important documentation uris. I'll continue posting on this topic and might ask questions to more experienced coders. TomT0m (talk) 13:35, 1 May 2014 (UTC)

'als'/'gsw'? 'de-formal'?

i already asked on MediaWiki talk:Gadget-autoEdit.js, but it seems to be a bigger issue: the script sets language for 'alswiki' to 'gsw', 'be_x_oldwiki' to 'be-tarask' and 'simplewiki' to 'en'. where is this mapping documented? somewhere on meta? why are there lots of labels and descriptions in 'als'? i set my babel to als-0 ans gsw-0 and it seems to be pretty the same. but labels and descriptions can differ. have there been any discussion on this? similar with 'de-formal': i remember a decision not to use it in item-namespace, but it is used, and the text sometimes differ from 'de' (db-report). should de-formal-terms be fully removed? --Akkakk 00:50, 23 April 2014 (UTC)

According to Wikidata:Requests for comment/Labels and descriptions in language variants 'de-formal' should be removed. And from [1] I conclude that 'als' labels should be replaced with 'gsw' labels. But if so, the 'als' babel should be immediately removed so that it is no longer possible to add 'als' labels through the UI. --Pasleim (talk) 14:18, 23 April 2014 (UTC)
Wikidata:Requests for permissions/Bot/AkkakkBot 7. babel is an extention that is configured in LocalSettings.php, so we probably need to file a bug. --Akkakk 18:05, 23 April 2014 (UTC)
See bugzilla:49024 for the de-formal thing. John F. Lewis (talk) 18:13, 23 April 2014 (UTC)

what's with the other mappings? 'bat_smgwiki' => 'sgs', 'be_x_oldwiki' => 'be-tarask', fiu_vrowiki' => 'vro', 'roa_rupwiki' => 'rup' are all defined in [2] but not mapped in MediaWiki:Gadget-autoEdit.js --Akkakk 19:07, 23 April 2014 (UTC)

@Jitrixis: Is it a bug? --Pasleim (talk) 16:43, 24 April 2014 (UTC)
For the general bug in Wikidata I created bugzilla:57706. For Gadget-autoEdit.js I implemented User:Fomafix/MediaWiki:Gadget-autoEdit.js. --Fomafix (talk) 12:34, 28 April 2014 (UTC)
@Fomafix: Your script looks good. Could you merge the changes from User:Fomafix/MediaWiki:Gadget-autoEdit.js to MediaWiki:Gadget-autoEdit.js? --Pasleim (talk) 20:18, 28 April 2014 (UTC)

i created this table to get an overview: (edit table)

wiki wrong lc correct lc reason autoedit labels autoedit descriptions languages table task 5 task 7 Ladsgroup wrong babel users wrong labels wrong descriptions wrong aliases conflicts
none de-formal de rfc ignored✓ OK none✓ OK no entry✓ OK ignored✓ OK ✓ OK no me✓ OK 0 0 1 0✓ OK
als: als gsw config ✓ OK ✓ OK ✓ OK ✓ OK ✓ OK ✓ OK 0✓ OK 388 0 0 389
crh: crh crh-latn ✓ OK ✓ OK ✓ OK ✓ OK ✓ OK ✓ OK 0✓ OK 0 0 0 0✓ OK
no: no nb ✓ OK ✓ OK ✓ OK ✓ OK ✓ OK ✓ OK 2✓ OK 3,720 357 12 4,054
simple: simple en ✓ OK ✓ OK ✓ OK ✓ OK ✓ OK ✓ OK 9 informed 25,xxx 4,0xx 0 29,686
bat-smg: bat-smg sgs ✓ OK ✓ OK ✓ OK ✓ OK ✓ OK ✓ OK 0✓ OK 0 0 0 0✓ OK
be-x-old: be-x-old be-tarask ✓ OK ✓ OK ✓ OK ✓ OK ✓ OK ✓ OK 1 informed 1,2xx 270 0 1,499
fiu-vro: fiu-vro vro ✓ OK ✓ OK ✓ OK ✓ OK ✓ OK ✓ OK 0✓ OK 0 0 0 0✓ OK
roa-rup: roa-rup rup ✓ OK ✓ OK ✓ OK ✓ OK ✓ OK ✓ OK 0✓ OK 0 0 0 0✓ OK
zh-classical: zh-classical lzh ✓ OK ✓ OK ✓ OK ✓ OK ✓ OK ✓ OK 0✓ OK 13 0 0 13
zh-min-nan: zh-min-nan nan ✓ OK ✓ OK ✓ OK ✓ OK ✓ OK ✓ OK 0✓ OK 414 0 0 413
zh-yue: zh-yue yue ✓ OK ✓ OK ✓ OK ✓ OK ✓ OK ✓ OK 0✓ OK 171 22 0 190
none nl-informal nl no consensus to change 0✓ OK 121 8 0 0✓ OK

feel free to edit. if no one opposes i would alter my bot to include this mapping in tasks 5 and 7. --Akkakk 15:16, 28 April 2014 (UTC)

i created lists of conflicts betwen labels and descriptions with wrong and correct language codes. these can't be fixed by bot. --Akkakk 21:07, 28 April 2014 (UTC)

I reported two new bugs concerning this topic: bugzilla:64649 and bugzilla:64650. --Pasleim (talk) 09:56, 30 April 2014 (UTC)

There are the pairs ido/io (at least when using the template:babel), da/dk (was a WMF redirect some time / some years ago). לערי ריינהארט (talk) 23:29, 30 April 2014 (UTC)
there are no ido or dk terms, dk doesn't have a babel. --Akkakk 22:13, 3 May 2014 (UTC)

Name for an item: literary/literature form/format?

novel (Q8261) is said to be an instance of literary genre (Q223393), but isn't it (along with poems, plays, short stories and novellas) rather a literary/literature form/format? I'd like to create an item for this, but want input on what to call such an item:

  • literary form
  • literary format
  • literature form
  • literature format
  • ... something else?

--Bensin (talk) 13:58, 26 April 2014 (UTC)

To answer this we need definitions for form and genre. By the way there is also a generic Property:P136 property. TomT0m (talk) 14:03, 26 April 2014 (UTC)

User:Zolo
Jane023 (talk) 08:50, 30 May 2013 (UTC)
User:Vincent Steenberg
User:Kippelboy
User:Shonagon
Marsupium (talk) 13:46, 18 October 2013 (UTC)
GautierPoupeau (talk) 16:55, 9 January 2014 (UTC)
Multichill (talk) 19:13, 8 July 2014 (UTC)
Susannaanas (talk) 11:32, 12 August 2014 (UTC) I want to synchronize the handling of maps with this initiative
Mushroom (talk) 00:10, 24 August 2014 (UTC)
Jheald (talk) 17:09, 9 September 2014 (UTC)
Spinster (talk) 15:16, 12 September 2014 (UTC)
PKM (talk) 21:16, 8 October 2014 (UTC)
Vladimir Alexiev (talk) 17:12, 7 January 2015‎ (UTC)
Ham II (talk) 09:24, 31 October 2015 (UTC)
Sic19 (talk) 21:12, 19 February 2016 (UTC)
Wittylama (talk) 13:13, 22 February 2017 (UTC)
Armineaghayan (talk) 08:40, 10 March 2017 (UTC)
Hannolans (talk) 18:36, 16 April 2017 (UTC)
Pictogram voting comment.svg Notified participants of WikiProject Visual arts might be interested, would be cool if we had generic arts guidelines. TomT0m (talk) 14:09, 26 April 2014 (UTC)

I agree a novel is not a genre but more a form or format. Chicklit is a genre, and this is usually in the form of romance novels. Jane023 (talk) 14:22, 26 April 2014 (UTC)
I'd say romance is a genre too. As is fantasy, sci-fi, horror and perhaps drama and comedy. I'd say genre relates to the content of the text whereas the length and format is related to the form/format. Horror novels/plays/poems are all in the horror genre but differ in form/format. --Bensin (talk) 16:55, 26 April 2014 (UTC)
I'd say we don't need specialized properties for litterature then, as this seems a generality for all arts TomT0m (talk) 17:05, 26 April 2014 (UTC)
We're not talking properties here, "literary form" would be a simple item. I changed the discussion title to make it clear. Mushroom (talk) 17:15, 26 April 2014 (UTC)
Often properties are doubled by a class item. Globally the two subjects are closely related, if there is an item, it's pretty sure he will be used in some statements, by a generic property like instance of (P31), the class could be the domain of some property and so on. A question on this is an opportunity to talk of that. The next step is how do we say a specific work has a form and a genre. We can use instance of (P31), of course, like
< Star Wars > instance of (P31) View with SQID < Science fiction Hollywood film >
, with
< Science fiction Hollywood film > subclass of (P279) View with SQID < Hollywood film >
< Science fiction Hollywood film > subclass of (P279) View with SQID < Science fiction Work >
, but I think we also need a form of artwork property (which would subsume the literary work form one). TomT0m (talk) 17:30, 26 April 2014 (UTC)
Interesting point, I hadn't thought about that. I would probably oppose the new property since it would duplicate the information from instance of (P31). But then again, someone could come up and say: why not delete genre (P136) and create items like "horror book" and "fantasy book" to be used with instance of (P31)? Doesn't seem such a good idea to me. So I'm conflicted, maybe you're right. Mushroom (talk) 19:47, 26 April 2014 (UTC)
I would probably oppose the new property since it would duplicate the information from instance of (P31) That's because you are not aware of the class definition in a standard like Web Ontology Language (Q826165) View with Reasonator View with SQID. In those languages, instance of (P31) actually implies the presence on certain properties and maybe certain values of those properties in items who are instance of that class. Redundancy is a feature in those cases, I think we should not avoid it at any costs. TomT0m (talk) 20:57, 26 April 2014 (UTC)
I like "literary form". The English Wikipedia has a Category:Literature by form that includes: novel (Q8261), play (Q25379), poem (Q5185279), serial fiction (Q1347298), screenplay (Q103076), short story (Q49084), teleplay (Q356055), and literary series (no item, but we have book series (Q277759) which is a subclass). Mushroom (talk) 17:15, 26 April 2014 (UTC)
It seems that the situation is more complex for texts than for painting (where in practice, it seems relatively straightforward to use {{p|31]}for the type of object + genre (P136)). Yes it seems that "genre" means something like "type of content, general tone", while "form" would be based on simpler formal criteria, but the distinction is not that clear. In France at least, novel is often considered a genre (you can googlebook the "genre romanesque" vs "forme romanesque"). And that seems to make sense: you don't have clear formal criteria for defining a novel. What differentiate novels from, say, essays (fictional, narrative, etc.) may be more readily considered part of the definition of "genre" rather than of the definition of "literary form". --Zolo (talk) 17:58, 26 April 2014 (UTC)
I agree, that's an issue and not just for novels. Is poetry poetry just because of its form? Surely not. I like Bensin's idea of a clear distinction between form and content, but what should we do when that distinction is just too blurry? Mushroom (talk) 19:47, 26 April 2014 (UTC)
novel (Q8261) is a class. It is a subclass of (P279)=literary work (Q7725634). I suppose you could say it is also an instance of (P31)='literary form' (i.e. 'literary form' is the special type of class that it is) but I don't think that gives you much additional information. Filceolaire (talk) 19:58, 26 April 2014 (UTC)
It gives a lot of information if you want to retrieve all the known literary forms. A subclass of the works with a certain work might not be a literary form itself. TomT0m (talk) 20:57, 26 April 2014 (UTC)
I think we can put them in this order: novel (Q8261) is a subclass of literary form, which itself is a subclass of literary work (Q7725634)Wylve (talk) 21:35, 26 April 2014 (UTC)
You are wrong. If
< novel (Q8261) View with Reasonator View with SQID > subclass of (P279) View with SQID < literary form >
, then as , by definition of subclass of (P279), this would imply that
< Les Misérables (Q180736) View with Reasonator View with SQID > instance of (P31) View with SQID < literary form >
, as any instance of a class is also an instance of its superclass. Except Les Misérables (Q180736) is absolutely not a literary form. TomT0m (talk) 21:41, 26 April 2014 (UTC)
What TomT0m said.
  • Les Miserables is an instance of a Novel which is a subclass of literary work which is a subclass of creative work.
  • Novel is an instance of literary form. Literary form is a subclass of 'class'. Filceolaire (talk) 00:01, 2 May 2014 (UTC)

Wikidata Toolkit v0.1.0 released: Working with Wikidata in Java

I am happy to announce the very first release of Wikidata Toolkit, the Java library for programming with Wikidata and Wikibase. This initial release can download and parse Wikidata dump files for you, so as to process all Wikidata content in a streaming fashion. An example program is provided. The libary can also be used with MediaWiki dumps generated by other Wikibase installations (if you happen to work in EAGLE ;-).

Maven users can get the library directly from Maven Central (see the documentation); this is the preferred method of installation. There is also an all-in-one JAR at github (you will need to install the dependencies manually if not using Maven) and of course the sources.

Version 0.1.0 is of course alpha, but the code that we have is already well-tested and well-documented. Improvements that are planned for the next release include:

  • Faster and more robust loading of Wikibase dumps
  • Support for various serialization formats, such as JSON and RDF
  • Initial support for Wikibase API access

Nevertheless, you can already give it a try now. In later releases, it is also planned to support more advanced processing after loading, especially for storing and querying the data.

Feedback is welcome. Developers are also invited to contribute via github.  – The preceding unsigned comment was added by Markus Krötzsch (talk • contribs) at 16:04, 2 apr 2014 (UTC).

(Adding comment to get this archived at some point.) --Lydia Pintscher (WMDE) (talk) 10:59, 4 May 2014 (UTC)

Adding point in time (P585) to date items?

Hi everyone, I noticed that date items like May 5 (Q2550) and 2014 (Q1999) didn't contain point in time (P585). Is anything keeping us from (mass) adding point in time (P585)? On a sidenote, can anyone explain me the difference between time interval (Q186081) and no label (Q14646051) at Talk:Q186081#Merge?? Multichill (talk) 10:53, 3 May 2014 (UTC)

Time interval must be a duration : one second, two seconds, not localised anywhere in time, whether temporal interval must be "beetween the firth of this month and the 15 on the next". Or the opposite /o\ TomT0m (talk) 11:08, 3 May 2014 (UTC)
The "time interval" property should be deleted and replaced by the datatype with temporal units, since it is more appropriate. Regarding items like "May 5th", some time ago there was this property proposal for reoccurring dates in machine readable format, because they happen every year, but there was not enough support back then, maybe you could reopen it? For year items yes, point in time (P585) with precision year seems adequate.--Micru (talk) 11:39, 3 May 2014 (UTC)
@Micru: We're not talking about properties here, but about items (so classes here). TomT0m (talk) 12:02, 3 May 2014 (UTC)
TomT0m, yes the item acts as the class, but if you add May 5 (Q2550)<point in time (P585)> "May 5", that represents "the month of May of the year 5", and not all "May 5ths of all years".--Micru (talk) 12:47, 3 May 2014 (UTC)
You don't use the right property. A date used in that sense is a point in time:
I would say for recurring dates we could create properties like month in the Gregorian Calendar and day in month. Then we would have:
< 14th of july > month search < july >
day in month search < 14 >
(for example). TomT0m (talk) 13:04, 3 May 2014 (UTC)
Thanks for the explanation, I didn't understand well what you were saying, but it makes sense to create the properties that you suggest to represent reocurring dates. I just hope that there are no items for each day of each specific year, because that would be perhaps too much.--Micru (talk) 13:16, 3 May 2014 (UTC)
Interesting, so on May 5 (Q2550) I added "00000000005-05-01T00:00:00Z" what means May of the year 5. I would like to add something like "*-05-05T00:00:00Z" (like cron syntax), but I doubt the datatype supports that.
At least the non recurring items are correct, right? Please review 2014 (Q1999) api and January 2014 (Q2164864) api. Both now contain start time (P580), end time (P582) & point in time (P585). I wonder if we need to add "after" to point in time (P585). Multichill (talk) 15:20, 3 May 2014 (UTC)
All the dates that have a year (or that were entered thinking that the number was a year) are correct. The "before" and "after" parameters are not necessary in this case because you know the exact date. Mmm, and no, cron format dates are not supported, but we can just re-interpret them as items/classes as TomT0m suggested. Just need to propose the needed properties.--Micru (talk) 17:15, 3 May 2014 (UTC)
I would oppose adding point in time (P585) to 2014 (Q1999) and January 2014 (Q2164864) because they have a duration with a start time and an end time. There may other items where we use point in time (P585) for items with a short duration (e.g. a race/competition) and I think we can let those slide but these items have a very specific meaning related to time so I would want to be much more strict and give the start times and end times down to the second (including leap second (Q194230) where these apply). When will the UI allow us to add time of day?
When we have 'numbers with dimension' datatype then we should add a 'duration' property (in seconds). Filceolaire (talk) 18:50, 4 May 2014 (UTC)

Are contributions safe?

Since some months I have the feeling that not all edits are stored properly. Today during some de-capitalizations of nouns I experienced this at more items. Normaly I make a "purge" after all edits. However in some places after a half of hour I could see that not all my edits. They where not stored properly. Regards gangLeri לערי ריינהארט (talk) 15:31, 3 May 2014 (UTC)

Without links to specific examples and what you are missing it is really hard to help. Possibly someone else reverted your edits but it is impossible to tell without links. --Lydia Pintscher (WMDE) (talk) 11:05, 4 May 2014 (UTC)

Broken gadgets (bis)

I've just noticed that Q numbers are no longer displayed next to values in statement. Do you have the same issue ? — Ayack (talk) 11:56, 30 April 2014 (UTC)

Ayack: Can you provide a screenshot? I don't know exactly what are you referring to....--Micru (talk) 07:17, 1 May 2014 (UTC)
@Micru, Lydia Pintscher (WMDE): As I no longer see them, it will be difficult to provide you a screenshot. After looking at Category:Wikidata screenshots, it seems that it was not the default UI... But this raises another issue: most of the "standard" gadgets I opted in, no longer work. Especially the most useful ones: MediaWiki:Gadget-AuthorityControl.js, MediaWiki:Gadget-Descriptions.js, MediaWiki:Gadget-CommonsMedia.js and the one, unidentified, which display the Q number of the statement next to the value link. It's the second time they are suddendly broken (see Wikidata:Project chat/Archive/2014/04#Broken gadgets). — Ayack (talk) 09:06, 1 May 2014 (UTC)

No one else is having these issues? Or nobody uses these gadgets? — Ayack (talk) 14:32, 2 May 2014 (UTC)

@Ayack: I can hear the crickets... I have all these problems as well. The tool that used to display the Q-number beside statement items is User:Magnus Manske/wikidata useful.js (function showQ). I really wonder why no one else sees these malfunctions. LaddΩ chat ;) 00:09, 3 May 2014 (UTC)
I also have this problem. Since a few weeks ago MediaWiki:Gadget-AuthorityControl.js has been working on and off, now it is definitely out of service and it is very anoying. Bug opened as Bugzilla64842, notifying Lydia--Micru (talk) 18:52, 4 May 2014 (UTC)
@Micru: Thanks! — Ayack (talk) 08:32, 5 May 2014 (UTC)

Country of citizenship

With country of citizenship (P27), I've been using the period-appropriate country item rather than the current-equivalent country item (e.g. Kingdom of England (Q179876) rather than United Kingdom (Q145); or Russian Empire (Q34266) rather than Russia (Q159)). I've noticed some edits recently where the current country has also been added to the statement along with the period country. Should I be doing this too? Should I even be using the period country at all? - AdamBMorgan (talk) 17:28, 1 May 2014 (UTC)

IMHO We should use only the reference to the country at the date when the person was living and not the current one. This is more complex but more correct. Snipre (talk) 19:14, 1 May 2014 (UTC)
In some cases the country item is used for multiple versions of the country (e.g. no new items for changes in name or minor boundary changes) but wherever there are different items then the country of citizenship should refer to the item that the person was actually a citizen of. e.g. Isaac Asimov (Q34981) was born a citizen of Russian Soviet Federative Socialist Republic (Q2184) - not the Russian empire (that died two year earlier) nor the USSR (that wasn't formed till two years later). Filceolaire (talk) 23:16, 1 May 2014 (UTC)
What to do when some Wikipedias don't have distinct articles for the "nation" and the "state"? For instance in People's Republic of China (Q148) vs. China (Q29520) some of the articles on the PRC are just called "China". It often happens that the article on a country/nation also contains the information about its current state/form of government.
Also, when is a change small enough to be just considered a "change in name" as you say and what to do when there isn't a constitution establishing a specific name? Interesting example I found: Ancient Rome (Q1747689) is said to has part (P527) the various names/periods/specific organisations. --Nemo 06:23, 2 May 2014 (UTC)
Nemo: I am seeing the same problem everywhere, and this could be a partial solution: Bugzilla52564.--Micru (talk) 06:56, 2 May 2014 (UTC)
Maybe, but it's a tangent for the problem at hand. Consumers of country of citizenship (P27) will need to be able to express it linguistically; if a concept is only a redirect in my language, I'm unlikely to be able to actually use it (e.g. as adjective). --Nemo 08:44, 2 May 2014 (UTC)
Thats a different problem. What should a wikipedia do if a wikidata statement links to an item which doesn't have an article on that wikipedia. This will happen a lot because Wikipedia tends to group similar concepts in the same article while wikidata insists they have separate items - The 'Bonnie and Clyde problem'. One solution is to leave the link as a red link - with a 'translation required' tag if necessary. Another solution is to link to a redirect (see the Bugzilla bug listed above). A third solution is to link to a Reasonator page. Filceolaire (talk) 09:29, 2 May 2014 (UTC)
Wikipedias get to decide how their articles are organised and Wikidata gets to decide how it's items are organised and Wikidata has decided that if a concept can be described using statements then that concept needs a separate item to put those statements in. See my comment above re what wikipedia can do about this.
A change is small enough that it doesn't need a separate item if the both the before and after can be described in the same item using statements (with start date/end date qualifiers if needed).
When there isn't a constitution specifying a specific name then the name can be given in the label and aliases and no 'Official name (property pending)' statement is needed.
Hope that helps. Filceolaire (talk) 09:42, 2 May 2014 (UTC)
Filceolaire: continuing with what Nemo said, a software solution would be much better than just relying on redirects, like treating items connected with "consists of/part of" as a whole on request. For instance, see Nocturnes, Op. 9 (Q1140459) and its parts like Nocturne No. 2 in E-flat major, Op. 9, No. 2 (Q16739831) can be considered as one, or at least it can be signaled visually somehow. The interlanguage list could be enhanced by adding some structure to it, or it could be "flattened out", like it has been suggested for Wikisource work/edition pairs. We should give it some thought and collect edge cases, and perhaps even consider other properties to describe closer item relationships ("integral part of"?).--Micru (talk) 10:27, 2 May 2014 (UTC)
Be careful of trying to fix a wikipedia problem by messing with the wikidata ontology. I'm not saying don't do it; I'm saying be careful. Wikipedia has an existing mechanism (= software solution) for linking parts to a general article and that mechanism is the redirect. Where there is a semantic need for a link from an item for a part to the item for the whole then wikidata should have statements to fulfil that semantic need. The edge case is, as you said, the sitelinks - these are part one thing and part the other. Allowing links to redirects will mean articles on 'Bonnie Parker ' can link to articles on 'Bonnie and Clyde ' but for links in the other direction you do probably need some sort of special wiki-only version of has part (P527). Filceolaire (talk) 15:39, 2 May 2014 (UTC).
See the Wikidata:Property_proposal/Generic#facet_of discussion. Filceolaire (talk) 15:41, 2 May 2014 (UTC)
The property "facet of" is to link items that deal on the same topic but from a different perspective (homo sapiens<facet of>human, water data<facet of>water), it is not the same case. The problem with the redirects is that it only solves a part of the problem and it can even make things worse if you want to solve it completely. For any item that "integrally consists of" two (or more) sub-items, where do you place the redirects? On each sub-item to the missing interwikis that are in the main item? On the main item to the missing language interlinks from each sub-item? On both? And manually? It would be much better to keep the (1-depth) structure and display it like that on the side bar for each one of the items, but who knows if it would create more problems that it would solve... suddendly editing sitelinks would not be that easy any more.--Micru (talk) 18:24, 2 May 2014 (UTC)


To return to the original question, should Dante Alighieri (Q1067) have country of citizenship (P27) -> Italy (Q38) as he does now, or should it be Republic of Florence (Q148540)?--Underlying lk (talk) 03:12, 3 May 2014 (UTC)

Maybe none, since citizenship didn't exist in the 13th century? -- Innocent bystander (talk) (The user previously known as Lavallen) 04:32, 3 May 2014 (UTC)
Same as Lavallen, there was no cititzenship, and it could be argued that countries didn't exist in the same way as they do today. Use only "place of birth" and then it will be possible to check the property "is in the administrative-territorial entity" for that date. The cititzenship property is quite useless, it could be removed and just use "significant event:cititzenship" when there are changes or deviations from the expected value.--Micru (talk) 08:28, 3 May 2014 (UTC)
That makes sense, thank you.--Underlying lk (talk) 08:57, 3 May 2014 (UTC)
If Dante Alighieri (Q1067) has any citizenship statement then (in my opinion) it should be country of citizenship (P27) -> Republic of Florence (Q148540).
Province of Florence (Q16172) needs to be linked to Republic of Florence (Q148540) with some 'preceded by' statements. Filceolaire (talk) 19:27, 4 May 2014 (UTC)
There isn't any relationship whatsoever between Province of Florence (Q16172) and Republic of Florence (Q148540). --Nemo 14:23, 5 May 2014 (UTC)

significant event (P793) and migration

Changing country is a pretty significant event in most peoples lives and I just came up with a way to describe it using significant event (P793) that I have applied to Lila Tretikov (Q16741739). Make it two events!

OK? Filceolaire (talk) 20:10, 2 May 2014 (UTC)

Is there any way of incorporating the date in that, or would it need to be separate? Ajraddatz (talk) 05:32, 3 May 2014 (UTC)
A good example of the use of significant event (P793). I also support adding point in time (P585) as a qualifier since it is normal for significant event (P793). /ℇsquilo 05:48, 3 May 2014 (UTC)
See Lila Tretikov (Q16741739) where I have used point in time (P585) to say when it happened. Filceolaire (talk) 19:04, 4 May 2014 (UTC)

Commons links - to categories, pages, or both?

Hi all. An issue I've been worrying about recently is whether the "Wikimedia Commons page linked to this item" should be to individual pages or to the categories. Having a wikidata entry brings a distinct benefit to the page/category now since the interwiki links to the Wikipedia articles are automatically included in the sidebar. There's a long-running debate on commons about whether pages or categories are better, which I would rather not get into (since categories are clearly superior to pages. ;-) ), but it's not clear to me whether the single field available in this section should link to the page or the category. A practical example of this might be Q16741739, where I linked to the commons category, but the link was changed to point towards a new commons page, which I subsequently reverted so that the interwikis were still available on the category page - I'm not sure whether that was something I should have reverted or not. Perhaps it might be worth expanding the Commons page links to include both a page and a category link? (On a related topic, I don't understand why a commons category link exists separate from the commons project link, don't both of these serve the same purpose but the lesser-used project link gives extra functionality? Also, it would be nice to see a good replacement to commons links on Wikipedia that rely on wikidata...) Thanks. Mike Peel (talk) 19:49, 3 May 2014 (UTC)

Don't make cross namespace links, Commons:Lila Tretikov wins. Multichill (talk) 20:48, 3 May 2014 (UTC)
Except that Commons:Category:Lila Tretikov has twice as many images as Commons:Lila Tretikov and always will be because it is much easier for unskilled editors to add a photo to a category than to a gallery. There may be a good reason not to add cross namespace links but @Multichill: hasn't given it here. Filceolaire (talk) 17:35, 4 May 2014 (UTC)
I seen in this case, it's been changed back to the page rather than the category, which means that a load of interwiki links have been added to the category. I thought that wikidata was supposed to make manually listing interwikis a thing of the past? Thanks. Mike Peel (talk) 08:43, 5 May 2014 (UTC)
That is one of the biggest missunderstandings about Wikidata. Interwiki is not the main purpose and manual interwiki with [[bar: will remain in many pages. -- Innocent bystander (talk) (The user previously known as Lavallen) 08:51, 5 May 2014 (UTC)
Wikidata:Requests for comment/Commons links--Ymblanter (talk) 12:53, 5 May 2014 (UTC)

How to handle such links?

en:Wikipedia:Bot owners' noticeboards:it:Discussioni progetto:Botit:Discussioni Wikipedia:Bot, but en:Wikipedia talk:Botss:it:Discussioni Wikisource:Botit:Discussioni Wikipedia:Bot. --Ricordisamoa 06:22, 4 May 2014 (UTC)

it's an interwiki conflict --Akkakk 08:53, 4 May 2014 (UTC)
It's rather talk_pages included. Keep the local interwiki! -- Innocent bystander (talk) (The user previously known as Lavallen) 10:01, 5 May 2014 (UTC)

Remaining languagelinks

Hi, is there any up-to-date list of pages in dewiki that have languagelinks? Regards, IW 12:28, 1 May 2014 (UTC)

IW: You can query it with WDQ, for instance all items with a sitelink to dewiki: http://wdq.wmflabs.org/api?q=link[dewiki] --Micru (talk) 19:01, 4 May 2014 (UTC)
And then you also have these stats.--Micru (talk) 19:04, 4 May 2014 (UTC)
@Micru: I think you misunderstood. I'm looking for a list of pages on dewiki that have local languagelinks. But thanks for your answer. Regards, IW 11:39, 5 May 2014 (UTC)

@Addshore: This is dead. Can it be repaired? @Legoktm: This is gone. Should it be removed from Wikidata:Database reports? --YMS (talk) 11:46, 5 May 2014 (UTC)

Yes both of the above links are dead :) over the next few weeks I will look at generating a new list of remaining interwiki links. It should be a lot easier this time around as there are far far far far fewer links and managing the list should be a lot easier!
If I don't seem to do anything within a week or so please poke me again as its likely fallen off my todo list (again) ·addshore· talk to me! 12:12, 7 May 2014 (UTC)
That would be really great, thank you :) IW 17:27, 7 May 2014 (UTC)

800+ items to merge

Hi, could somebody help to merge 800+ items? Thanks. — Ivan A. Krestinin (talk) 07:54, 5 May 2014 (UTC)

Thanks all for help. I generalize search algorithm and it find 10000+ items now. — Ivan A. Krestinin (talk) 21:17, 5 May 2014 (UTC)
Thanks to you. --ValterVB (talk) 22:36, 5 May 2014 (UTC)
@Ivan A. Krestinin: Wow, tremendous work, thanks! I suggested an improvement at User_talk:Ivan_A._Krestinin/To_merge. - LaddΩ chat ;) 12:58, 6 May 2014 (UTC)
If can help [3] is @Byrial: outdate project but the algorithm is very good (if some one with C skill can update it) --Rippitippi (talk) 22:22, 6 May 2014 (UTC)

Mass message

Hi, do you know what happened here? --Stryn (talk) 12:48, 5 May 2014 (UTC)

Maybe something in that tech news update triggered an abusefilter or something? --GeorgeBarnick (talk) 15:15, 5 May 2014 (UTC)
More than likely an autoblock. Which has expired now. John F. Lewis (talk) 18:13, 6 May 2014 (UTC)
I re-opened bugzilla:58237 for this. In the future if you notice anything weird with MassMessage, ping me or drop a note on m:Talk:MassMessage. Thanks! Legoktm (talk) 01:21, 9 May 2014 (UTC)

How to change the order of a list ?

How to change the order of a list, so as to respect the chronological order ? As in Paris Q90, The head of (the actual) government (Hidalgo) should be on top of the the list but not at the bottom. Thank tou very much. --Cquoi (talk) 06:24, 8 May 2014 (UTC)

Cquoi, when you click on the edit link next to the statement, you will see tiny triangles appearing to the right of the statement. With those arrows you can move a statement up or down inside a group, or you can move the whole statement up (if you are editing the first statement of the group) or down (if you are editing the last one). Qualifiers can also be sorted. Please report if it is working fine when you reload the page because to me sometimes it went back to the previous order.--Micru (talk) 08:20, 8 May 2014 (UTC)
Thank you very much Micru. I could manage to put Hidalgo on top of the list but as her name disappeared ! So I had to remove my edit. Regards --Cquoi (talk) 15:55, 8 May 2014 (UTC)
You're edit was correct, but the software temporarily made her name disappear for some reason (must be a bug, I also encountered it while making the edit). It should be good now. The Anonymouse [talk] 16:07, 8 May 2014 (UTC)
That said, I think the list should be in direct chronological order rather than the other way around as it makes it easier to create timeline-templates. The current mayor should be mark as preferred (click edit and then on the three squares to the left of the value). This way this shold be the only one returned by default on Wikipedia. --Zolo (talk) 16:53, 8 May 2014 (UTC)
Long lists are pretty painful to fix by hand. We really need a bot that does these kind of fixes regularly. Similar to the disambig-bot on Wikipedia it could also alert users and say "You added a newer thing. Do you want to set this thing as the preferred entry?". Tobias1984 (talk) 21:49, 8 May 2014 (UTC)

Wikimedia disambiguation page (Q4167410)

Hi! https://www.wikidata.org/?diff=123670458&oldid=123669605#top is an example where an anonymous user added a Wikimedia disambiguation page as a link to another item using the qualifier imported from (P143).
The value Q15732187 for Englisch (Q15732187) View with Reasonator View with SQID can not be found in the history of the page.
I suggest to avoid / disable linking to disambiguation pages by Javascript code. Disambiguation pages are special elements in an ontological system. It is a mess that any user and any bot can add links to this "taboo (Q171180) View with Reasonator View with SQID" elements. gangLeri לערי ריינהארט (talk) 06:12, 9 May 2014 (UTC)

new start/end date qualifiers

How do I request new qualifiers for the "start date" and "end date" properties? Specifically, I want to record both when construction started and when services started, for example at Edinburgh Trams (Q2299624). Many other things will have this sort of distinction as well - e.g. many buildings have a construction start date, construction end date, opening date and official opening date. Former railways may have construction start, service start, service end, demolition date. Or am I going about this the wrong way? Thryduulf (talk: local | en.wp | en.wikt) 15:34, 2 May 2014 (UTC)

significant event (P793) is the way to do this. Use this property to links to generic items for construction (with start date and end date qualifiers) and start of service and official opening (each with point in time qualifiers). You may have to create some new items to link to.
There has been some comment that this property is a bit of a cheat - a way to create loads of properties informally - but I like to think it is a way for infoboxes to find the significant events without having to guess what properties are used with each item.
If we ever get a 'subproperty' mechanism then this might be changed but, for now, this is what we have. Filceolaire (talk) 15:49, 2 May 2014 (UTC)
In contrast to emigration/immigration below, I do not think this is a good use of significant event (P793). /ℇsquilo 05:52, 3 May 2014 (UTC)
If I remember correctly this was exactly the kind of use that was described when significant event (P793) was proposed. @Esquilo: how would you suggest Thryduulf should describe these dates? Filceolaire (talk) 19:11, 4 May 2014 (UTC)
It was proposed for uses like this, but it has not managed to gain acceptance. /ℇsquilo 16:47, 9 May 2014 (UTC)

Bug or feature?

If I add a claim to an item, Wikidata does not chek if that exact claim is already in the database. What is this good for? While the interface is a pain in the ass with its inability to strip spaces, carriage returns etc. from pasted text while meticulously calling those inputs "malformed", it doesn't even display a warning when I do this. Example: https://www.wikidata.org/w/index.php?title=Q2419755&diff=127232936&oldid=126118947 Bug or feature? --AndreasPraefcke (talk) 11:21, 9 May 2014 (UTC)

I guess you can call it a feature. Before it was throwing an error, which according to Bugzilla64431 had to be fixed. Maybe it is fixed now? Personally I think you should be allowed to enter twice the same value because you can add different qualifiers to each one of them, but yes, perhaps a warning would be nice.--Micru (talk) 11:47, 9 May 2014 (UTC)
Thanks for the answer. Yes, considering the qualifiers it may make some sense. (In all the cases when I made this mistake, though, there wasn't a qualifier in the original statment, and none in the new, so this makes the edit still wholly undesirable). With Wikidata items growing, and an interface that is the most wasteful in its screen space vs. data ratio that I have ever encountered, plus with a sorting of statements that seems random at best, it will become increasingly difficiult to see if an statement is already in the database, and I guess double statements will grow in number. I just hope there's a bot around that will delete those in a timely manner. --AndreasPraefcke (talk) 12:25, 9 May 2014 (UTC)
Statements are regrouped by property, so unless there is a lot of statements with the same property it won't be that hard. And the Wikidata user interface is beeing reworked by the developpers right now, so we'll see :) I remember asking about this already somewhere like on wd:dev. TomT0m (talk) 12:29, 9 May 2014 (UTC)
Sorry, I meant: sorting of properties that seems random. --AndreasPraefcke (talk) 13:41, 9 May 2014 (UTC)

Q1326938

Hi! @Lymantria: The items is used at special:WhatLinksHere/Q1326938. For a reason not known to me iit is llisted also at tree via 31,279,361 for fundamental interaction (Q104934). gangLeri לערי ריינהארט (talk) 18:33, 3 May 2014 (UTC)

Are you sure you wanted to post this here instead of on User talk:Lymantria so that User:Lymantria would actually notice it? Multichill (talk) 19:30, 3 May 2014 (UTC)
a) The item was merged and deleted the 9th of April 2014 and nobody complained. b) Lymantria did not edit the 3th of May. c) There is a bug in the link above, a link to a deleted item which only developers may detect. d) "weak interaction" (Q11418) another "fundamental interaction" (Q104934) was an orphaned (linking to nowhere / without an instance of / subclass of statement) item. It seems that we can live without it ... gangLeri לערי ריינהארט (talk) 01:10, 4 May 2014 (UTC)
Any news on this? The involved pages are translation pages and include Q1326938 which is not listed / included in the English source page. Why the value is displayed in the translations? לערי ריינהארט (talk) 23:01, 9 May 2014 (UTC)

Natural substances

Hello. Natural substances have composition and properties within a range. For instance, Mohs hardness (Property:P1088) of ferro-actinolite is within a range: 5-6. Is there a standard procedure or do we need two new properties ('Mohs hardness higher than' and 'Mohs hardness lower than') ??? Regards --Chris.urs-o (talk) 07:50, 5 May 2014 (UTC)

You can enter the value 5.5±0.5. Snipre (talk) 11:24, 5 May 2014 (UTC)
@Snipre, Chris.urs-o: The solution with (mean value) ± (standard deviation or whatever this is supposed to be) is just a temporary solution to this problem. The Mohs' hardness of minerals is not always an average value with some minerals being harder than others (The values then straying around the mean). It is often a true range of values that is a product of natural variations in the chemistry of the minerals. I think we already discussed that for some minerals we need to add crystal face and scratching direction as qualifiers, because they sometimes have big differences in hardness. Tobias1984 (talk) 21:06, 6 May 2014 (UTC)
We need to allow at least two formats for numerical values: 5.5±0.5 and 5-6. --Chris.urs-o (talk) 02:12, 7 May 2014 (UTC)
There is nothing in the notation x±y that implies normal distribution (Q133871). /ℇsquilo 09:39, 10 May 2014 (UTC)

Top of the subclass tree about physical entities

Hi, want some input about the top of the tree around physical entities. It seems like uncomplete and like a mess to me. We should be able to express things and relationship between class of entities like manufactured objects, constructions, natural constructions, animals, plants, civil engineering construction (Q1411945), architectures, architectural structures ... and there might be a lot of cultural differences in the modeling. So it would be great if we could open a discussion about that. This is an open subject, so please add your thoughts about the problems you encountered if you already tried to add statements on those kind of items or what can be found of the subclass tree at that point. TomT0m (talk) 12:37, 9 May 2014 (UTC)

This is what upper ontologies are for. On Classifying Material Entities in Basic Formal Ontology discusses this precise topic in great detail. Though slightly broader, the discussion Top of the subclass tree from December 2013 is also relevant here. Emw (talk) 11:36, 10 May 2014 (UTC)

Bug?

Twenty minutes ago I deleted no label (Q16188434) as duplicate of Cesar de Brabander (Q11963392). I always wondered how duplicates could exist, so I tried item by title. Hmmm, no result. Is there a bug behind this causing duplicate entries? Kind regards, Lymantria (talk) 11:47, 10 May 2014 (UTC)

Item by title can find the page at Q16188434 (I tried to restore). I think this is related to Wikidata:True duplicates. --Stryn (talk) 12:32, 10 May 2014 (UTC)
Indeed it is related to Wikidata:True duplicates. But I think the not deleted item should be assigned to the corresponding site on nowiki... I see there is Bugzilla65130. Lymantria (talk) 12:45, 10 May 2014 (UTC)
Solved. Seems I was impatient. Lymantria (talk) 12:59, 10 May 2014 (UTC)

Military units

What is the best way to list a soldier's (or sailor's/airman's/etc) military units? I thought of military branch (P241) but all the examples I looked at seem to be at a higher level (e.g. Royal Navy or United States Army). Would employer (P108) be more appropriate? - AdamBMorgan (talk) 16:28, 1 May 2014 (UTC)

Perhaps a "military unit" property could be proposed. Branch is the elemental level or career path depending on the country. Ajraddatz (talk) 16:31, 1 May 2014 (UTC)
part of (P361) is the best existing property. member of (P463) seems to be limited to organizations and clubs. You could propose a more specific property for this purpose. Is there any other context in which such a property could be used? Filceolaire (talk) 23:21, 1 May 2014 (UTC)
No, it would only be used for military members to denote which specific sub-organization (not element) they are a part of. This makes some sense; in unified forces, such as the Canadian Armed Forces, a person can be in what is referred to as a "purple" unit - one which allows members of any element (navy, army, air). If a branch property could be justified, then a unit one could be as well IMO Ajraddatz (talk) 05:31, 3 May 2014 (UTC)
I would not use part of (P361) for any human (Q5). I think the best way of expressing the connection to a military unit is to use it as a qualifier to position held (P39). Example: position held (P39) => executive officer (Q1383594) qualifier of (P642) => Life Regiment Hussars (Q10568222). /ℇsquilo 05:43, 3 May 2014 (UTC)
That's actually a very good way of doing it, killing two birds with one stone! Allows the position to be specified and the unit. Ajraddatz (talk) 05:46, 3 May 2014 (UTC)
I'm not sure I like this. Is it more important to state the organisation the person is a part of or its position / function in it ? If we will soon want to query this later, the query to now if a person was a member of a human organisation (any type) would be "has the person held any position [of this precise organsisation]", where if we use "part of" it would be "has this person been a part of this organisation". But there is mainly (too my knowledge) several of participation to an organisation : beeing an employee, a client, an adherent to an association ... We should take everything into account and build a consistent way of saying all this before making choices. TomT0m (talk) 11:04, 3 May 2014 (UTC)
"member of" makes a lot of sense, but we've given it a very precise definition. "Part of" is not really appropriate - an organisation is part of a larger organisation, but a person is a member of an organisation. Wouldn't it be simpler to redefine "member" to include military units as well as voluntary organisations? (The discussion on the talkpage for P463 suggests that that's broadly appropriate)
"Position held" seems tempting, but in the military context this will often be several different positions/ranks within a single unit, and so probably overcomplicating things. Andrew Gray (talk) 18:40, 4 May 2014 (UTC)

After a bit of thought (and adding a lot of conflict (P607)/military branch (P241) entries...), here's an outline of how I'd suggest we use military properties for people:

@AdamBMorgan, Ajraddatz, Filceolaire, Esquilo, TomT0m: - thoughts? Am I missing anything obvious? Andrew Gray (talk) 11:34, 10 May 2014 (UTC)

Those properties all seem reasonable to use as you suggest, Andrew. A few others from Wikidata:List_of_properties/Person jump out:
For military persons, it seems like a niche property "military casualty classification" would be useful to precisely capture information like killed in action (KIA), missing in action (MIA), wounded in action (WIA), etc. Emw (talk) 20:00, 10 May 2014 (UTC)
Wouldn't it be simpler (and possibly clearer) to define manner of death (P1196) to include "killed in action" and "missing in action" rather that treating these as subclasses of homicide/unknown? (For one thing, a lot of KIA is not known with enough detail to define homicide vs. accident). We'd need a different way to capture "wounded in action", though. Andrew Gray (talk) 12:13, 11 May 2014 (UTC)
Andrew, manner of death (P1196) is mapped to the 'manner of death' field on the US Standard Death Certificate, which doesn't include 'killed in action'. It turns out I was wrong about KIA -- at least as defined by the US military, it is only "applicable to a hostile casualty", which seems to be a subclass of homicide. Deaths of a military person due to accident or suicide are classified as a "non-hostile casualty" instead of KIA. Emw (talk) 18:31, 11 May 2014 (UTC)
Hmm. It's taken me a bit of mental juggling, but I think I see what they're getting at with the restriction - the problem is, of course, that in many cases what we know is simply "KIA". Would it be a suitable value for cause of death (P509), though? Andrew Gray (talk) 19:26, 11 May 2014 (UTC)
Andrew, cause of death (P509) is for more granular claims, ideally an item mapped to an ICD disease or injury term (e.g. http://www.icd10data.com/ICD10CM/EIndex/W/War_operations). KIA strikes me as somewhere between 'cause of death' (e.g. 'combat using blunt or piercing object') and 'manner of death' (e.g. 'homicide') in granularity, and part of a controlled vocabulary that includes relevant terms outside the scope of those properties (e.g. WIA, DOW, POW, MIA). I opened a property proposal so that vocabulary can be expressed in a cohesive bunch: Wikidata:Property_proposal/Person#military_casualty_classification. Emw (talk) 03:04, 12 May 2014 (UTC)

National Library of Israel

Has anyone experience with the catalogs of the National Library of Israel (Q188915)? I've checked National Library of Israel ID (P949).

Examples:

It looks like the system numbers are not stable. --Kolja21 (talk) 22:48, 11 May 2014 (UTC)

What are the numbers telling us? I think they are identifying mentions of a person in books, so if Albert Einstein is mentioned in a book it gets a unique number. What they should have is the entry for a book should list a unique identifier for each person mentioned in the book. --Richard Arthur Norton (1958- ) (talk) 22:16, 11 August 2016 (UTC)

Interwiki links screwed up

All the links to Hesperus and Phosphorus (morning star) were mixed up. I canceled many but not all wrong links but I don't succeed to reintroduce the correct links. I am afraid that the software isn't user friendly enough to prevent such mistake and to take care of it. Pinea (talk) 21:43, 9 May 2014 (UTC)

@Pinea: Can you explain further? --AmaryllisGardener talk 23:12, 9 May 2014 (UTSC)

A lot of pages related to Hesperus were linked to Phosphorus and viceversa. So the problem was that I had to exchange links and I was not allowed by the system. I had to destroy half of the links before modifying the other half. After several attempts I succeded to fix most of them but not all. For instance I don't know how to handle the eswiki Hespero, still connected to Phosphoros. Moreover I destroyed the links of dewiki hesperos, fiwiki hesperos, eowiki hesperos and nowiki fosforos, and I am not able to link them to the correct page anymore(I really tried only for dewiki). I also don't read chinese, japanese, hindi, persian or arabic titles so I don't have the slightest idea whether they are connected to the proper pages. There are several questions. First why some wiki behave differently than others. Even in wikidata table some links were initially impossible to remove (an error flag was popping out), Gd only knows why. Second the difficulties probably were due to the fact that the software cannot handle easily the replacement of a wrong link with another one. Rather than simply telling that I cannot make the link because another page from the same wiki is already linked, it should suggest me what to do, for instance could allow the replacement provided I also suggest a new link for the other page previously linked there. Pinea (talk) 08:57, 12 May 2014 (UTC)

Completely useless request

Hello.

Q13, Q42, and Q666 respectively point to triskaidekaphobia, Douglas Adams, and the number of the Beast. So why not exchange Q69 and Q2349 ? :-) --Agatino Catarella (talk) 07:46, 10 May 2014 (UTC)

 :-) Thanks. --Agatino Catarella (talk) 17:34, 10 May 2014 (UTC)
I totally support! I would also like to exchange Maria Walliser (Q1984) and Nineteen Eighty-Four (Q208460). /ℇsquilo 08:58, 12 May 2014 (UTC)

Fixing import including coordinates of place of burial

User:KrattBot added coordinate location (P625) to ~722 records of human (Q5) causing a bunch of constraint violations. I had a plan to work with Ladsgroup to have a bot make the following changes…

add place of interment (P119) and use the value cemetery, burial plot, or tomb (Q16859021) as a placeholder item with the qualifier of coordinate location (P625) to hold the former coordinate value.

Does this sound reasonable or does someone have a better solution?

You can see Eugene Bullard (Q1367535) for an example

Jared Zimmerman (talk) 16:31, 11 May 2014 (UTC)

Sound very reasonable to me. /ℇsquilo 08:49, 12 May 2014 (UTC)
I also noticed these coordinates, imported from enwiki or fawiki, but i am afraid that is not always clear, to what are these coordinates related (place of death? place of burial? place of birth?). So the result can be incorrect claims. --Jklamo (talk) 14:14, 12 May 2014 (UTC)

Butterkäse

How to link de:Butterkäse. Thanks Mike Coppolano (talk) 06:46, 12 May 2014 (UTC)

job posting: looking for a PHP backend developer

Hey everyone,

I am looking for a great PHP developer to join the software development department at Wikimedia Germany. We have a lot of backend tasks to do around Wikidata especially for the support of Wikimedia Commons. More details about the position are available at https://wikimedia.de/wiki/PHP_Backend_Developer_(f/m) If this is you please do apply. If you know someone who fits please send them them the link. This is a great chance to make a difference around Wikidata's software.

Cheers --Lydia Pintscher (WMDE) (talk) 14:54, 12 May 2014 (UTC)

2 new cool tools: Edit Item Suggestions and Maps

Hey :)

2 new cool tools that have not been mentioned here yet are an edit item suggester and a map with WP-geocoordinates filtered by using Wikidata classes. More details about them in these two mailing list posts:

Cheers --Lydia Pintscher (WMDE) (talk) 15:03, 12 May 2014 (UTC)

Strange item not really well defined

I run into dick (Q15333) View with Reasonator View with SQID . It seems to be an item we should try to figure out. If it was an article, I would say that it's about the popular ways to speak about the male sexual organ in the different languages. Would that be a correct definition ? TomT0m (talk) 16:42, 10 May 2014 (UTC)

It is a difficult case because each language has a different number of synonyms that might have separate Wikipedia and for sure different wiktionary pages. I haven't had the time to read the wiktionary guidlines, but they are here: Wikidata:Wiktionary.Tobias1984 (talk) 09:11, 11 May 2014 (UTC)
w:Dick (slang) was recently created, so I have added it as a sitelink.
One way to model it would be to say dick and mat (Q2096843) are said to be the same as penis (Q58).
Another way, which I have done, is to say this word is main subject (P921) => penis (Q58). John Vandenberg (talk) 06:32, 13 May 2014 (UTC)

Page view statistics

Is there any page view statistics available? If no, is it planned? I see it has been requested a few times here and elsewhere. John Vandenberg (talk) 06:23, 13 May 2014 (UTC)

It would be nice if the community would finally get better access to meta-data without having to finish an informatics degree first. Page-view statistics should be available through a point and click interface. This should allow comparing wikidata-item views, or the attached Wikipedia pages. For example it should be easy to access in which language of Wikipedia the article about corruption is visited most often. Are these people working on this?: https://www.mediawiki.org/wiki/Analytics - Another good thing would be real time data. This would allow admins to quickly filter out items that are getting unusual traffic. These items could then be protected and improved. -Tobias1984 (talk) 10:14, 13 May 2014 (UTC)

Reasonator as a default gadget?

Can we make the ReasonatorTools script a default gadget? Its effect is to add a portlet link under the 'tools' to view the current item in Reasonator.--Underlying lk (talk) 15:56, 6 May 2014 (UTC)

We can but community consensus (very recently discussed) is not strong for it. That means no for people who feel I am being ambiguous. John F. Lewis (talk) 17:11, 6 May 2014 (UTC)
Without beeing iconoclast, What we call community here is a very small number of people. Even Underlying Lk's opinion (I don't think he was) if he was alone, would make a difference. And the person who could be the most interested in that gadget might be person who don't discuss here. TomT0m (talk) 20:45, 6 May 2014 (UTC)
Last time it was proposed the discussion was muddled by talks of replacing {{Q}} with {{Q'}}, but there was definitely support for the gadget.--Underlying lk (talk) 08:37, 7 May 2014 (UTC)
There was for the gadget, not for enabling it by default for all. John F. Lewis (talk) 22:39, 7 May 2014 (UTC)
I don't see any reason why not to, once someone l10n-ifies it. Legoktm (talk) 01:23, 9 May 2014 (UTC)
@John F. Lewis: True, but that is because I didn't know about the possibility of setting a gadget to default for all users when I created the previous discussion, hence the new proposal.--Underlying lk (talk) 07:28, 9 May 2014 (UTC)
You didn't but the community did and it was opposed against. John F. Lewis (talk) 15:46, 9 May 2014 (UTC)
@John F. Lewis: Not at all, I could count many people in support of the gadget and only one explicitly against making it a default gadget, the other votes were in opposition to {{Q'}}, which is not part of this proposal.--Underlying lk (talk) 13:41, 11 May 2014 (UTC)
I see at least 4 opposes for the default gadget. Assuming the lack of discussion there anyway apart from a reasonator cabal, I don't see community consensus as I stated in the close. John F. Lewis (talk) 16:58, 11 May 2014 (UTC)
  • Symbol support vote.svg Support because those who votes and are aware it exists are not those who would need it the most. TomT0m (talk) 16:03, 9 May 2014 (UTC)
  • Symbol support vote.svg Support It provides language fall back.. it is what makes Wikidata usable in the small languages. I also approve of the argument by TomT0m. GerardM (talk) 18:15, 11 May 2014 (UTC)
  • Symbol support vote.svg Support How much count the vote of a visitor that doesn't vote? I don't know, but I vote on their behalf.--Micru (talk) 11:20, 12 May 2014 (UTC)
  • Symbol oppose vote.svg Oppose It is not so useful for enabling it for all users. — Ivan A. Krestinin (talk) 12:19, 12 May 2014 (UTC)
  • Symbol support vote.svg Support It's one small link in the Wikidata navigation bar. Who doesn't want to use it may simply ignore it, just like he/she probably ignores most of the links there. All the others get easy access to a more appealing and more intelligent viewing interface for all Wikidata items, not only if someone set a Q' link somewhere. --YMS (talk) 12:31, 12 May 2014 (UTC)
  • Wikidata user interface is overloaded already. — Ivan A. Krestinin (talk) 12:38, 12 May 2014 (UTC)
    • Is it? The "Tools" section is collapsible and has only seven links by default. I added nine for myself and still never felt like it's being overloaded at all. And if it was, I would rather remove some default links that are useless in Wikidata in my point of view (e.g. "Printable version" and especially "Cite this page"). --YMS (talk) 13:24, 12 May 2014 (UTC)
  • Symbol oppose vote.svg Oppose, I don't know is it a good idea to add a link to external site. Maybe add some gadget which you can enable if you want, but not enabled as default. --Stryn (talk) 13:00, 12 May 2014 (UTC)
    I think it is a bad argument. Wikidata is a generic tool which can handle all kind of datas, and it won't change. Which means there always be external from Wikidata tools who are better at doing things that Wikidata itself. Wikidata is designed to have clients and datas are supposed to be used by external tools, it is doomed to that. I prefer to see Wikidata as the backend of an ecosystem of tools who uses, help the editing, uses the data, builds cool visualisations, helps structuring the datas, calculates constraint violations, ... We already have Wikipages inside the website to point to them. Some are important in our thinking on how to structure the datas, think of the constraint violation bots. So ... is really Reasonator on wmflabs just an external tool or more a part of the ecosystem ? Compared to that I don't think ... is an opinion that needs an argumentation :) TomT0m (talk) 15:48, 12 May 2014 (UTC)
Symbol oppose vote.svg Oppose I think a gadget is a tool which makes it possible to edit and improve thinks within a project. Reasonator provides a nice outside view to our data, nothing else. In other words: Reasonator is simply a viewer to our data, whereas eg WDQ is a really powerfull tool.--Succu (talk) 17:57, 12 May 2014 (UTC)
Nope, Reasonator can add labels in your language to items he diplays that are not, so it's not totally true. I think good visualisation is really important, don't you ? A newbie can understand the powerfulness of Wikidata thanks to that, whereas a mere backend is not really hot. And it's really easy to go back to Wikidata from Reasonator if you want to edit stuff. Reasonator is also hackable. TomT0m (talk) 18:21, 12 May 2014 (UTC)
Viewing a fitting representation of data (eg. an excel diagram) helps understanding the relationsships between data. Having a handy interface to correct them is another theme. --Succu (talk) 18:53, 12 May 2014 (UTC)
Sorry, I don't understand where you want to go and how that's relevant to the discussion. This is not really an argument against reasonator as a default gadget. TomT0m (talk) 19:00, 12 May 2014 (UTC)
It's related to your remark „I think good visualisation is really important, don't you ?” Excel has both. Thats all. --Succu (talk) 19:10, 12 May 2014 (UTC)
Oh, ok I get it. I don't think Wikidata has a generic graph drawing tool planned, though we will be able to define domain specific extraction tool by the generic query engine, which will be the equivalent of Excel charts and formulas. But eventually why not, I guess a timeline is also a generic way to place dated items or claims on a visualisation. We're far from having such tools yet unfortunately. Reasonator is still the best option for item data vizualisation as far as I know. That and beeing an open project. TomT0m (talk) 19:49, 12 May 2014 (UTC)
You really think you got it? Think twice about my initial comment. Excel was only (as I think) a wellknown for data representation.
BTW: Yes this is an open project and feel free to write a really editable version of Magnus Reasonator or something comparable. --Succu (talk) 20:32, 12 May 2014 (UTC)
Mmm OK, if you say so. If you challenge me to think twice I sometime think there is sometimes nothing to understand in your posts /o\ In particular I fail to see a justification for your opposition. TomT0m (talk) 21:06, 12 May 2014 (UTC)
I simply delivered my arguments TomT0m. --Succu (talk) 21:16, 12 May 2014 (UTC)
  • Symbol neutral vote.svg Neutral I think the gadget is very useful, but I'm not sure about the external link (as said by Stryn). --AmaryllisGardener talk 19:40, 12 May 2014 (UTC)
  • Symbol support vote.svg Support It's one more link between others. It doesn't hurt. If anybody (probably a trusted, serious and long-standing contributor) feels awkward seeing this fancy Reasonator thing, surely he will know how to disable it in Preferences. I guess newcomers will be more grateful for this stuff (if they find it, of course). And if by any means this add results into an unlikely situation of damage to the project and inasumible chaos we could talk about it again here and disable it to all users. Nothing is irreversible. Well, at least this one.Totemkin (talk) 14:30, 13 May 2014 (UTC)
  • Symbol support vote.svg Support per Totemkin --Pasleim (talk) 15:42, 13 May 2014 (UTC)
  • Symbol oppose vote.svg Oppose I don't really see advantages in Reasonator (just one more badly designed interface) and those who want to use it, are indeed using it, so they have no problem. What we don't need is extra buttons, making this interface more difficult to handle for all users. - Brya (talk) 16:43, 13 May 2014 (UTC)
  • Symbol support vote.svg Support לערי ריינהארט (talk) 01:08, 14 May 2014 (UTC)
  • Symbol oppose vote.svg Oppose Reasonator is a nice tool, but not so essential to be a default gadget. Also it is external tool, so it may disappear, stop working or start being buggy anytime (i still miss some tools from old toolserver). Last but not least wikidata UI is already overloaded. --Jklamo (talk) 09:00, 14 May 2014 (UTC)

omegawiki contributors please help!

@Bigbossfarin @GerardM Hi! It is quite late. While linking WD items to omegawiki I found more examples where the oposite links are

linking to disambiguation pages as [6] linking to Persian (Q543213) View with Reasonator View with SQID
items at omegawiki without links to WD as [7] and flerovium (Q1302) View with Reasonator View with SQID

There are more examples. Please let me know a stable url at WD whhere to post such issues. It would be nice to know who is contributing at both projects. Today there are only a half of dozen of professions at omegawiki.
There would be a lot of such items where aliases have to be cleaned at the field of this profession items. See the history for "ethnology" (Q43455) [8] and "ethnologist" (Q1371378) [9] . Thanks gangLeri לערי ריינהארט (talk) 23:36, 9 May 2014 (UTC)

Try posting them at http://www.omegawiki.org ----- Jura 16:54, 10 May 2014 (UTC)
@Jura1 I can not log in at omegawiki for some weeks. לערי ריינהארט (talk) 22:33, 10 May 2014 (UTC)

DefinedMeaning:7001 - Bambara see Bambara (Q33243) View with Reasonator View with SQID and Bambara people (Q504685) View with Reasonator View with SQID . @Bigbossfarin @GerardM "Please let me know a stable url at WD whhere to post such issues. It would be nice to know who is contributing at both projects." gangLeri לערי ריינהארט (talk) 22:17, 11 May 2014 (UTC)

FYI: items with DDC = Dewey Decimal Classification (P1036) but without OmegaWiki Defined Meaning (P1245) לערי ריינהארט (talk) 16:34, 13 May 2014 (UTC)

@Kipcool both DefinedMeaning:7610 - Navajo and DefinedMeaning:7611 - Navajo link to Navajo people (Q108266) View with Reasonator View with SQID. Please note Navajo (Q13310) View with Reasonator View with SQID. Can you please create a peer page at Wikidata? Thanks in advance! gangLeri לערי ריינהארט (talk) 18:52, 14 May 2014 (UTC)

Split death by burning (Q468455)?

Skimming through the properties specified for the "one of" constraint of cause of death (P509), I noticed that death by burning (Q468455) should probably be split - however, I am not sure what is the best way to proceed. While the English item describes an actual cause of death ("death by burning"), the German item is about the pile of wood that is used to burn someone and not about the event of causing death. In any case, it is odd to have, in German, the pile of wood specified as cause of death - piles do not kill, being burned on them does. Random knowledge donator (talk) 09:29, 14 May 2014 (UTC)

Yes that needs to be cleaned up. The stake (german: Scheiterhaufen) is in my opinion similar to the guillotine a tool or a method. There are a lot of semantic mixups with cause of death (P509). Some of the discussion is here: Property_talk:P509#Values. Also looking at this table shows that a lot of things are thrown together Wikidata:Database_reports/Constraint_violations/P509#Values_statistics. -Tobias1984 (talk) 11:13, 14 May 2014 (UTC)

Mining accident

Need some help finding all the mining accident (Q1550225) on Wikidata. There is also a list item list of mining accidents (Q1406882) View with Reasonator View with SQID. Especially help from somebody who knows zh would be appreciated: https://www.wikidata.org/w/index.php?search=%E7%9F%BF%E9%9A%BE&title=Special%3ASearch&go=Go - Please translate into English and set instance of (P31) = mining accident (Q1550225) -Tobias1984 (talk) 11:58, 14 May 2014 (UTC)

Thank you User:Wylve for the zh-translations. Just this item still need translation 1981 Shihuijing Second Coal Mine disaster (Q10843963). It helped me find a few more articles about accidents in turkey. Can somebody that knows Ukrainian and Russian still check this list: de:Liste_von_Unglücken_im_Bergbau#Ab_2001. In recent time especially the Ukraine saw some terrible accidents. -Tobias1984 (talk) 13:28, 14 May 2014 (UTC)
I've translated 1981 Shihuijing Second Coal Mine disaster (Q10843963). —Wylve (talk) 15:13, 14 May 2014 (UTC)

chemical element (Q11344) instance of (P31), subclass of (P279) and part of (P361)

Hi! I worked at the de-capitalization of titles linked to chemical element (Q11344) WLH.
I expected to see the (all) elements at tree via instance of (P31) and the classes at tree via subclass of (P279).
hydrogen (Q556) is in the first "tree" but is also a subclass of chemical element (Q11344) and nonmetal (Q19600) .
What would be the appropriate usage? @Danmichaelo, @Baronnet, @Styrke, @Nima_Farid, @Miss_Manzana, @国家宗主, @ব্যা_করণ, @Szczureq, @Svavar_Kjarrval‎ Best regards gangLeri לערי ריינהארט (talk) 09:28, 30 April 2014 (UTC)

I thing one should follow the systematics of the periodic table (Q10693) WLH. in ~15 min tree via part of (P361) should be available. לערי ריינהארט (talk) 10:29, 30 April 2014 (UTC)
All chemical elementtis are subclass of chemical element (Q11344) because all elements are class of isotopes. See Wikidata:WikiProject_Chemistry/Tools. Thanks. Snipre (talk) 11:50, 30 April 2014 (UTC)
And there are discussions about using "part of" for classification like nonmetal (Q19600), group, block or period. To clarify. Snipre (talk) 12:11, 30 April 2014 (UTC)
Nope, elements are
< Oxygen > subclass of (P279) View with SQID < atom >
, an any oxygen atom is an atom.
< Oxygen > instance of (P31) View with SQID < Chemichal Element >
(Oxygen itself is a type of atom, who is labelled Chemichal element because all the atoms of the class have the same atomic number. The atomic number is the defining property of this class of atoms. Similarly Template:Isotope is a type of atoms. isotope of hydrogen (Q466603) is a subclass of isotopes. Tritium is an instance of isotope, an instance of Hydrogen isotope, and a subclass of atom. TomT0m (talk) 12:31, 30 April 2014 (UTC)
@TomT0m No, you can't have Tritium is an instance of isotope AND an instance of Hydrogen isotope. Either Tritium is an instance of isotope AND an instance of Hydrogen or only Tritium is an instance of Hydrogen isotope. Because Hydrogen isotope is an instance/subclass of isotope. Snipre (talk) 15:41, 30 April 2014 (UTC)
Yes you can, take the set of all iotopes, who will countain all germanium isoptopes (as sets of atom), tritium ... take the set of hydrogen isotopes. It's a subset of it. Tritium is an instance of both, and all Hydrogen isotopes are isotopes. It's the definition of a subclass. Tritium is an isotop. An isotope is a set of atoms instances who have all the same composition of their nucleus. Maybe a drawing would help. TomT0m (talk) 15:50, 30 April 2014 (UTC)
@TomT0m. Ok you are right, you can but this should be avoided: add a new parameter like radioactive isotope. So you can have tritium is an instance of isotope, radioactive isotope, isotope of hydrogen, radioactive isotope of hydrogen, radioactive hydrogen, radioactive atom,... Add a fourth criterion like artificial isotope and I let you imagine the possible number of combinaisons. This problem occured for WP categories with the "vegeterian russian speaking english biologist" case instead of the four individual categories allowing increasing possible matching scenarios through cross queries. WD is a database meaning we should aim to avoid any data agregation to allow the largest possible queries in the future. Snipre (talk) 14:29, 1 May 2014 (UTC)
Wikipedia Categories are not exactly a system in which this complexityé could be managed :) With Wikidata this is another story. In OWL for example a class can be defined by something who looks lot like a query. So a reasoner engine can infer an item belongs in a class without stating it explicitely, which means a bot can complete the graph. A tool like Wikidata Query by magnus can generate a subclass tree and find the instances of the class and the instances of the subclasses in the instance list (if the link is explicit, Wikdata query is not a complete reasoner, just a query engine.) Think also of places classification. So we don't have to make all the combinations of course, but classes like musician or women or 'elements are still interesting and meaningful in real life, so if there is no technological problems, there is interest to explicit them (and this probably would make querying more efficient). That said while queries are not implemented that's still speculations :). This is also really interesting for NPOV as we can use without problems a lot of classification systems and tag their classes as instances of classes defined in this classification. TomT0m (talk) 14:56, 1 May 2014 (UTC)
Please see Wikidata:Property proposal/Generic#element_of (and Wikidata:Property proposal/Generic#context). In set theory there are classical relations subset (one equivalent is part of (P361)) and another is element of (the new priority). At the periodical table the chemical elements are elements (they are subsystems i.e. a subset in chemical physics (Q2001702) ). Regards gangLeri לערי ריינהארט (talk) 13:33, 30 April 2014 (UTC)
copied from the proposal: copper (Q753)is in element of relations with group 11 (Q185870) and period 4 (Q239825) and d-block (Q214503)
לערי ריינהארט (talk) 13:50, 30 April 2014 (UTC)
(edit conflict) Seems mostly redundant with instance of (P31) and subclass of (P279). They have the same logical properties in OWL, except OWL is more powerful because OWL allows to define the class intentionally. See this link. I voted oppose with this argument. TomT0m (talk) 13:52, 30 April 2014 (UTC)
@TomT0m: So where on Wikidata is a discussion about chemical element (Q11344)? When do you expect to have a proper mapping? gangLeri לערי ריינהארט (talk) 14:12, 30 April 2014 (UTC)
I don't remember where we discussed but Wikidata:WikiProject Chemistry seems a good place to start, although it's not easy to find. Wikidata: showcase items is the community agreed place to see example. Property mapping is not really a thing we discussed yet, as it might not be a 1-1 mapping but might more be something like a property-query mapping. TomT0m (talk) 14:27, 30 April 2014 (UTC)
@לערי ריינהארט: The big problem is the lack of definition for instance of (P31) and subclass of (P279).
@TomT0m: I don't see the difference between atom and chemical element: both have the same level and both can be defined with the atomic number. Perhaps it's time to define the use of these two properties in a clear manner. For me it'is simple: if a term can be divided into suclasses, this is a class or a subclass. If the term can't be divided further this is an instance. So chemical element can be divided into each element, and each element can be divided into isotope so they are classes and subclasses. Isotopes can't be divided so they are instance. Without a better definition which is approved by the community, the choice between subclasses and instance stay in the hand the contributors and the most used approach or the selected one by a project is the only one which is applied. Snipre (talk) 14:58, 30 April 2014 (UTC)
@Snipre: : Parts of the answer and several consistent ways (but only partials)
  • The (non rigorous) maths view : a class is a set, and the instances are elements. Subclass is the subset relation, so it is transitive : a subclass Z of Y if Y is a subclass of X is a subclass of X. So an element of Z is an element of X.
  • The class/token relationship : tokens are object of the real world, and classes regroup thoses objects. This is a useful way of seeing things, but we must take care that transitivity is not violated, for example to class administrative divisions it has often be used the class as a WIkimedia (sub)category, this implied that "département" was a subclass of "french administrative division", which does make sense, but "french administrative division" was a subclass of "administrative divisions by country", which does not make sense.
  • To take care of this problem, it is useful to relax the view used by Emw of the strict class/token with an object beeing either one or the other but not both in some ways. This has been called [Punning] in the world of web standards. A class can also be viewed as a token : for example, we imagine that the <departement> class has an equivalent real world object. Then we can regroup it with other pseudo tokens like <commune in france> in a class <french administrative division>. This also works for ship classes : a ship belongs to some kind of ship class, the ship class itself is usually classified irl. We match the conventions used elsewhere with punning, which is really good.
  • More math again : we got a set of objects with properties like "age : 28" or "Sex:male". A class is defined like some kind of query. Take all the objects in the database with properties "age>10" "age<20" "instance of:human". Call that collection of object a class, name it "teenagers", and call the query the extension definition of the class. Call the matching item "instances". TomT0m (talk) 15:19, 30 April 2014 (UTC)
Applied to chemical elements and their classification :
Atom classes.svg


@TomT0m Even if the presentation is nice, I am sorry I don't understand the principles which lead to that classification. Which criterion put a terme in the "type" class or in the "type class" class. Snipre (talk) 11:32, 1 May 2014 (UTC)
@Snipre:, it's actually a quite litteral transcription on the definition of chemichal element displayed in frech Wikipedia (I take it litteraly and did not make it : 'Un élément chimique désigne l'ensemble des atomes caractérisés par un nombre défini de protons dans leur noyau atomique." (a chemical element is a set all all atoms with the same number of protons in their atomic number). It fits exactly : the tokens of this classification are the atoms. They are regroup in classes, the second columns, which are sets of atoms. The biggest class of atoms is the atom class, who is the subset of all real world object we name atom (a poor definition but it's enough :p). Hydrogen and Germanium are also a classes (of atoms), they regroup atoms with a given atomic number (this number is class dependant). Both are subset of atoms, so they are subclasses of <atom>. This is for the first column and the pure class/token relationship.
Now a question : can we say the closest hydrogen atom of my finger who is in my coffee cup at this precise moment is an element ? not according to the Wikipedia definition. It is a token, not a class. So <element> does not fit in the second columns, in which there is only classes of atom tokens. So what can we call an element ? By definition, an element is a class of atoms with the same characteristics ... So
< <Hydrogen> > instance of (P31) View with SQID < <class of atoms> >
and <element> is a type of atom classes. All of them regroups atoms, what make them different (so its their characterization property), is the value of the atomic number of the atoms they regroup. Does that make any sense to you ? TomT0m (talk) 12:23, 1 May 2014 (UTC)

I used the instance of (P31) relation in all three places. Most items where "orphaned". Many languages are de-capitalized. gangLeri לערי ריינהארט (talk) 18:13, 30 April 2014 (UTC)

End of the day status on "meta"-objects:

P.S. Finaly Reasonator for periodic table (Q10693) with its relation to the extended periodic table (Q428887) .
A few items do not have FreeBase statements. Where is a place to collect such items and who can add them? gangLeri לערי ריינהארט (talk) 18:31, 30 April 2014 (UTC)

It's simpler to use {{Q'}} who generates a reasonator links :) group (Q83306) View with Reasonator View with SQID period (Q101843) View with Reasonator View with SQID periodic table block (Q193099) View with Reasonator View with SQID Looks not bad to me. TomT0m (talk) 18:43, 30 April 2014 (UTC)

Thanks for mentioning template:Q'. The actual classification (systematic) is redundant for groups (because of the main group). Feel free to modify it to something more understandable.
I like reasonator because it is a graphical user interface and I want to take advantage of the "tree" output for demonstration to unskilled people. I do not care if tree via instance of (P31) or tree via subclass of (P279) lists all chemical elements. But if you decide what makes more sense the elements should be listed only at one of these graphs. I think it is just a matter of preference and one can be used to either one. If the second graph is the one you prefer then the additional lexicographical terms which should be moved away with whatever trick. לערי ריינהארט (talk) 19:57, 30 April 2014 (UTC)

Unfortunately &depth=1 does not work at [10].
I managed to link the groups to some categories from languages other then English and added English labels for the categories. This might help third languages to find the categories easier.
For the elements I did not de-capitalize all aliases; neither in English nor in other languages.
The work is just a start, an exercise of dilettantism, as a subcalss of no label (Q1558663) ;-) . Thanks for any help! לערי ריינהארט (talk) 23:43, 30 April 2014 (UTC)


Pictogram voting comment.svg Comment: This seems to be a reiteration of the germanium subclass tree discussion from a few months ago, which in turn bears strong resemblance to an even earlier discussion on a proposed definition of "instance". Like others there, I find "atom type class" -- referred to as "atom kind class" in the linked discussion -- problematic. What about when we classify "atom type class" as a particular type of some more general entity, for example "physical object"? Does "physical object" become "physical object type kind class"?

I encourage others to review existing physics ontologies for perspective on this modeling question. For example, the Dumontier lab at Stanford has OWL ontologies for atoms, among other concepts in natural science. See http://dumontierlab.com/?page=ontologies. It seems like those classes might even be tied into the widely-used BFO upper ontology (outlined here). One thing is clear: like the vast majority of other OWL ontologies, the Dumontier atom ontology represents atoms as classes -- with owl:Class and rdfs:subClassOf, i.e. subclass of (P279). The same is done with the CHEBI ontology, which is somewhat of a reference ontology in the OBO Foundry. Emw (talk) 13:00, 1 May 2014 (UTC)

Correct me if I'm wrong, but these ontologies does not uses the <element> notion ? It does'nt fit in your scheme. Or it is assimilated to <atom>. Yet we have articles for <atom> and for <element> We also have tools to link them soundly using proerties. We are discussing a real usecase which is a really common patterns (ship and commonly used ship classes are exactly in the same case for example). TomT0m (talk) 13:17, 1 May 2014 (UTC)
We need to develop our own onthology which should be more simplified than the specialized one which can't be used by the occasional contributors. But we need an easy way to define what is an instance and what is a class at WD level to ensure a common use of the classification in WD. Snipre (talk) 13:49, 1 May 2014 (UTC)
The class/token as two levels is not so hard to understand I think. then to go from a token to a class we use instance of (P31). Subset is subclass (a subclass is a collection of objects of the same kind than the parent classes), so if we have <rock band> as a class, <Mick Jager> is not an instance of it, nor a subclass. A subclass of <rock band> is a set of band, such as <metal rock band>. Now below class and tokens we can add a third entity : token/classes/sets of classes. To go fron classes to set of classes we need a instance of (P31) statement. I think we got a lot confused because we rejected the idea a class cannot be a token at first, but this makes things a lot easier and avoid mistakes. You use subclass of (P279) when : the item has instances, and when everything we can say about its instances as a whole is also true for the instances of the parent class. If you want to link a stuff with a set of stuffs of this kind you use instance of (P31). TomT0m (talk) 14:18, 1 May 2014 (UTC)
TomT0m, my opposition to instituting a special ontological layer for sets of classes is (surprise!) rooted in the fact that it has no basis in Semantic Web standards or conventions. Your idea would be a fundamental and unique change in knowledge representation on Wikidata, but those qualities make the idea suspect.
The Dumontier lab's ontologies include both 'element' and 'atom' classes in the same ontology. From http://ontology.dumontierlab.com/chemistry-complex:
  • Atom
<owl:Class rdf:about="&dl;Atom">
  <rdfs:subClassOf rdf:resource="&snap;Object"/>
</owl:Class>

  • Element
<owl:Class rdf:about="&dl;Element">
  <rdfs:subClassOf rdf:resource="&dl;PureSubstance"/>
</owl:Class>

Those class definitions are simple and don't account for important things like isotopes, nor how an element has a restriction that it is composed of one type of atom. But Dumontier's classes are an example of how these basic concepts can be modeled in OWL. Emw (talk) 15:49, 3 May 2014 (UTC)
@Emw: Of course there is a theorical basis to that in the web semantic standards, it's called punning :) . It does not also says that the element class is composed of O, not of O2 for example. They use the 2002 definition of the ontology, which means they could not use punning. But that view is consistent with the ontology I proposed, we can merge the too without contradictions. TomT0m (talk) 16:03, 3 May 2014 (UTC)
TomT0m, isn't your unconventional "type class" construct essentially trying to hard-code a type relation that is conventionally solved at query- or modeling-time by reasoning engines via punning? Why not just use (hydrogen, subclass of, chemical element) like the rest of the Semantic Web? This conveniently also entails that hydrogen (Q556) (why were all P279 claims removed from chemical elements, by the way?) is about the set of things known as "hydrogen", and not any particular instance of hydrogen. Emw (talk) 16:18, 3 May 2014 (UTC)
I don't find the statement
< hydrogen > subclass of (P279) View with SQID < chemical element >
anywhere in the ontology you just linked. This would entail some inconsistencies, which is really bad for entailing. We are OK that hydrogen is a class of atoms (this is in the ontology I proposed. There is a lot of hydrogen instances in a glass of water. By transitivity of subclassing, this means that any of this instances is chemical element. Wow, but I though a chemical element was a pure substance, a single atom is not a pure substance by itself ? see their atom ontology. Hydrogen is a subclass of atoms.
<owl:Class rdf:about="&dl;HydrogenAtom">
<rdfs:subClassOf>
<owl:Class rdf:about="&dl;Atom"/>
</rdfs:subClassOf>
</owl:Class>
If it is also a subclass of Element, then ask yourself : is not ontologically the atom concept the same as the element concept ? A reasoner has no reason not to entail that. Yet we have two Wikipedia articles, with two different definitions. That's because ontologically an element is a set of atoms. Which means the class element is a set of set of atoms. This is just maths, this follows from the instance of (P31)/subclass of (P279) equivalent to set membership/subset relations. Individual atoms are members of the set of all atoms. Individual elements are subsets of this, not members. The set of all elements is a subset of the powerset of all atoms. And OWL is built on axiomatic. TomT0m (talk) 17:10, 3 May 2014 (UTC)
TomT0m, good point regarding (hydrogen, subclass of, chemical element); I see that the Dumontier atom ontology and ChEBI (here) both state (hydrogen atom, subclass, atom), and do not explicitly state (hydrogen, subclass of, chemical element). Interestingly, neither ontology models "hydrogen"; the subject in both is "hydrogen atom". So let me rephrase the question to which you were replying: in Q556, why are we using (hydrogen, instance of, chemical element), like noone else in the Semantic Web?

Because it's the correct thing to do and it is the precise transcription of the definition of chemical elements.
"Because it's the correct thing"? Tautological answers are tautological, TomT0m. Also, just to get this out of the way, your second rationale is bizarre. "Hydrogen is an instance of a chemical element" is obviously not, in any way whatsoever, a precise transcription of the definition of chemical elements, i.e. "A chemical element is a pure chemical substance consisting of a single type of atom distinguished by its atomic number."
Perhaps you meant "because 'hydrogen is an instance of chemical element' is a precise transcription of the definition of hydrogen." If so, that is not a bizarre rationale, but it is still incorrect. The question at hand is not "why are we using (hydrogen, is a, chemical element)?". If it were, then you would have a plausible argument since the English Wikipedia article on the subject begins: "Hydrogen is a chemical element".
But that was not my question. Again, my question is: "why are we using (hydrogen, instance of, chemical element)?" Note that 'is a' does not equal 'instance of'. In fact, in many major ontologies -- like those of the Open Biomedical Ontology (see here) -- 'is a' maps to 'subclass of', i.e. rdfs:subClassOf.
In summary, "because it's correct" is not a legitimate answer and your second rationale is incorrect. Emw (talk) 06:04, 4 May 2014 (UTC)
In your story about hydrogen in the glass of water, you seem to suggest that the lack of an explicit statement on the relationship between element and atom in the Dumontier atom ontology entails an inconsistency. Surely the lack of an explicit statement (atom, ?, element) might make that ontology incomplete, but that does not mean it is inconsistent.
No, if there is no statement it's because they can't make the statement because of the lack of punning in the OWL version they are using. TomT0m (talk)
What are you replying "no" to? Your second sentence is a non sequitur relative to you're replying to. Emw (talk) 06:04, 4 May 2014 (UTC)
"If (hydrogen) is (a subclass of atom and) also a subclass of Element, then ask yourself : is not ontologically the atom concept the same as the element concept?"
No. Holding both (hydrogen, subclass, element) and (hydrogen, subclass, atom) does not entail (atom, same as, element). A reasoner would not infer that. Perhaps you're confusing this with implicitly establishing (atom, same as, element) through the statements (atom, subclass, element), (element, subclass, atom). Emw (talk) 18:55, 3 May 2014 (UTC)
You did not answer the question. What do you think is the difference beetween an atom and an element ? Please remember my quote on the definition on element earlier. TomT0m (talk) 19:40, 3 May 2014 (UTC)
I obviously answered the question. You asked a yes or no question, I answered "no". I then how explained your logic was incorrect, which you have yet to address or acknowledge. Before you reply to this, please specifically address that rebuttal. Regarding your new, different question, one difference between an atom and an element is that an element can be composed of multiple atoms, but an atom cannot be composed of multiple elements. Emw (talk) 06:04, 4 May 2014 (UTC)

2014-05-02: Based on sleipnir.js I found a link tree via 31,279,361 which today might be quite usefull.
I will try to make a general overview based on extended periodic table (Q428887) .
I think that the property titles (and some language descriptions) for instance of (P31) and subclass of (P279) are causing misunderstandings. In Esperantoo the main meaning "subaro de" is "subset" and not "subclass". Regards gangLeri לערי ריינהארט (talk) 11:05, 2 May 2014 (UTC)

gangLeri, P31 and P279 are labeled "instance of" and "subclass of" because those are the closest to the names of the core Semantic Web properties upon which they are based. P31 is based on rdf:type. P279 is based on rdfs:subClassOf. The new RDF primer has a good introduction to these basic constructs.
You say that the meaning of "subaro de" in Esperanto is "set" and not "subclass". In the Semantic Web, "subclass" means essentially "subset". Indeed, rdf:type and rdfs:subClassOf are denoted as ϵ (element of) and ⊆ (subset of) in description logic, the formal knowledge representation system from which Semantic Web languages like RDF and OWL are derived. A deeper discussion on the nature of 'instance of' and 'subclass of' might also interest you.
In summary, P31 is labeled "instance of" and P279 is labeled "subclass of" because of convention in the Semantic Web. Emw (talk) 02:12, 3 May 2014 (UTC)

Reboot

@Emw: I think we are getting to hypercritical method (Q3333663) View with Reasonator View with SQID without wanting it, so after a good night of sleep let's recap without the noise.

So is a is ambiguous, yes. In common language we often says things like "911 Porsche is a car" when we should say it's a car model.

  • Hydrogen is a chemical elements
Two viewpoints :
  1. hydrogen is a chemical element, like the definition said
  2. Hydrogen is the set of all hydrogen tokens
Let's review the second. As Hydrogen is a set of token, this is a class. We get
< Hydrogen > subclass of (P279) View with SQID < Atom >
, as all the tokens are atoms. No problems about that.
Let's review the first. It's a bit more difficult because this does not fit the token/class as easily. Let's say Hydrogen is a token. Then Chemical element is the class it belongs two.

Problem : the two are inconsistent (trivially) if you consider the strict class/token distinction there is in OWL1 and in description logic in general : an item can be an individual or a class, but not both. So in a description logic a solution might be to make too items for hydrogen : one for the class of atom, one for the chemical element. This is what the ontology you link to does.

Now OWL2 introduces punning, because that make sense to go further. Hydrogen IRL can be both when we need one or the other definition. This is convenient for us in Wikidata as we don't really have one or the other viewpoint to push, we need to be NPOV. There is a translation for that in description logics where the items are finally split.

Advantage : "Un élément chimique désigne l'ensemble des atomes caractérisés par un nombre défini de protons dans leur noyau atomique." translates naturally in a consistent definition in this framework. This is not exactly the definition you showed for chemical elements, but it seems a little weird, the notion of substance seems not really useful and even cumbersome, it is not used in the rest of the english article if I read well. Except for the notion of pure substance, defined in french article by

Une substance pure constituée d'atomes du même élément chimique est appelée corps simple (A pure substance made of atoms of the same elements is a simple corp) which seems to me the definition of an element in english ... TomT0m (talk) 10:16, 4 May 2014 (UTC)
Pictogram voting comment.svg Comment: TomT0m, I was not being hypercritical in my replies to you. Your comments had blatant logical errors, and seemed somewhat flippant and accusatory. Furthermore, your three comments largely avoided actually responding to the matter at hand in each. I asked you to "please specifically address that rebuttal" in your reply, which you have not done. I will wait for you to do that before replying to the content of your post above. In other words, do you agree that the seeming claim of yours in question -- (hydrogen, subclass, element) and (hydrogen, subclass, atom), therefore (atom, owl:sameAs, element) -- is incorrect? If not, what is your rationale? Please include a "yes" or "no" in your reply, as this is partly a yes-no question.
Also, please do take greater care in your writing. Click "Prévisualiser" before saving, or read over what you write after you save and fix major errors. Your first sentence is currently unreadable to anyone without French set as their default language; it reads "I think we are getting to (no label) (Q3333663) without wanting it". This is clearly a major problem for the target audience on an English forum. You also write "We get
< Hydrogen > subclass of (P279) View with SQID < Atom >
, as all the tokens are atoms". Obviously, something is wrong there. Presumably you meant 'Subtype' (and why is Template:Subtype named 'Subtype'? Not a good name.) This is a wiki, but expecting others to fix these kinds of errors in your comments is unreasonable. These are not small details; I am not being hypercritical here. These are basic elements of consideration for your readers. Emw (talk) 18:20, 4 May 2014 (UTC)
@Emw: I did not want to answer to those question because we are losing ourselves into non relevant details. Anyway : maybe it's not enough for the reasoner to infer the equality of those classes by itself, it could be that those classes are disjoint. But what's for sure is that it imply that any hydrogen instance is both an (instance of) element and an atom. Which imply that, if <element> is a substance, any of its instance is a single molecule made of only one atom of hydrogen. And here we got the inconsistency to the french definition.
I'll add a sentence of the french Wikipedia article who is a confirmation L'oxygène est un élément chimique, mais le gaz appelé couramment « oxygène » est un corps simple dont le nom exact est dioxygène, de formule O2, pour le distinguer de l'ozone, de formule O3, qui est également un corps simple (Oxygen is the chemical element, but the gaz called usually oxygen is a simple corp whose exact name is di-oxygen, 02, ...). Looks like the writer of this introduction is a teacher and knows exactly where the confusions (or the cultural disctinctions) are :).
I apologize, but beeing who I am, I can't promise this won't happen again. And I'm beginning to be tired of that endless discussion. All of this for a definition mismatch ... Plus I often only work with the diffs, which does not help. TomT0m (talk) 18:55, 4 May 2014 (UTC)

──────────────────────────────────────────────────────────────────────────────────────────────────── Summarily dismissing a set of rebuttals as "irrelevant details" is a recurrent pattern in discussions with you, TomT0m. Who decides which points are irrelevant -- you, when you don't feel like replying to direct rebuttals of your central points? How convenient. The appropriate thing to do in such cases is to quickly acknowledge the validity of the rebuttals, or explain why they are incorrect, and then move on to related matters. Simply "rebooting" the conversation to ignore inconvenient rebuttals that directly contradict the point of yours at hand is rude and dismissive. Nevertheless, I will move on to points in your most recent post.

You said "it imply that any hydrogen instance is both an (instance of) element and an atom. Which imply that, if <element> is a substance, any of its instance is a single molecule made of only one atom of hydrogen." It's not clear what the "it" you're referring to is -- please be more specific and explicit. Also, since this is necessarily somewhat of a technical discussion, please note that "a single molecule made of only one atom of hydrogen" is nonsensical. A molecule consists of two or more atoms; there is no such thing as a molecule made of only one atom of hydrogen. That actually does seem irrelevant to your point; I simply note the error for the sake of precision. If an ontology does indeed make a cardinality restriction of 1 on the number of atoms in a "chemical element", then it is incomplete, and possibly but not necessarily also inconsistent. (It ignores the fact that H2, O2, O3, a 1cm cube of pure gold, etc. can be validly considered chemical elements.) But the Dumontier ontology under discussion does not do that.

For reference, below is how the Dumontier class hierarchy models carbon. You can see it for yourself by installing Protege, going to File -> Open from URL..., and pasting in http://ontology.dumontierlab.com/chemistry-complex. Then click on the 'Classes' tab and expand Thing -> entity -> continuant -> independent_continuant. The 'object' class referenced below is a subclass of 'independent continuant', which is part of the Basic Formal Ontology (BFO 1.0). I've chosen to outline carbon simply because the Dumontier ontology has more a complete representation of carbon than hydrogen.

Dumontier chemistry ontology (complex). Some classes omitted for brevity.

  • ...
    • ...
    • object
      • atom
        • carbon atom
      • chemical substance
        • pure substance
          • element
            • carbon
          • allotrope
            • carbon allotrope
              • diamond
              • graphite
      • mixture ...
      • molecule
      • organic group ...
      • ring ...
      • source
    • ...

Here are the ontology's comments for relevant classes:

  • Atom: a particle that retains all the chemical properties of a given element
  • Chemical substance: any material with definite chemical composition
  • Pure substance: a chemical substance that cannot be separated into other substances by any mechanical process
  • Element: a pure substance that cannot be decomposed or transformed into other chemical substances by ordinary chemical processes. An element is made up of atoms with the same atomic number
  • Allotrope: a structural variant of an element
  • Diamond: a carbon allotrope in which each carbon atom in diamond is covalently bonded to four other carbons in a tetrahedron. These tetrahedrons together form a 3-dimensional network of puckered six-membered rings of atoms.
  • Graphite: allotrope of carbon which is a conductor, and is the most stable form of solid carbon.

You said "maybe it's not enough for the reasoner to infer the equality of those classes by itself, it could be that those classes are disjoint." Here again, it is unclear what that critical first "it" is: what is not enough? It seems worth noting that, in the Dumontier ontology, neither 'element' nor any of its ancestor are disjoint with 'atom', and 'element' is not disjoint with 'allotrope'. At the same time, there is no 'owl:sameAs' nor 'owl:equivalentClass' assertion equating 'element' and 'atom'. This presumably accounts for the fact that an element can be composed of A) one atom or B) multiple atoms with the same atomic number.

A notable aspect about the ontology is that 'carbon' and 'carbon atom' and 'hydrogen' and 'hydrogen atom' are all modeled as classes by default, via rdfs:subClassOf (i.e., subclass of (P279)) -- not rdf:type (i.e., instance of (P31)). I do not think that the lack of punning in OWL 1 DL caused that. Precise labeling was probably more of the reason. In other words, like O2 is often colloquially called "oxygen" (e.g. "Humans breathe oxygen") and setting the label of 'O2' to 'dioxygen' is more precise, so does distinguishing in labels between 'hydrogen' and 'hydrogen atom' make precise the distinction between any of the various elemental forms of hydrogen (e.g. H, H2) and hydrogen as the class of single H atoms. Similarly, having different labels for the various allotropes of carbon makes the concepts more intuitively precise, e.g. 'diamond' is an elemental form of carbon.

Not only does Dumontier's consistent and reasonable modeling not require punning, it also does not need a special item labeled 'atom type class'. I am actually most interested in discussing this "type class" construct. It strikes me as the most severe mistake you have pushed (and pushed, and pushed) to date. It seems completely unnecessary and unprecedented in the Semantic Web, and would be a colossal source of confusion for users.

In summary, the Dumontier ontology does not imply that elements must consist of one atom. It also does not imply that one atom is not an element. It seems to model chemical elements and related scientific concepts fairly comprehensively and precisely, and does not seem to contort itself to deal with the lack of punning in OWL 1 DL. Also, adding a specially designated 'type class' construct to the basic constructs of 'instance' and 'class' seems like an especially bad idea. Emw (talk) 03:40, 5 May 2014 (UTC)

I hear you. I do not fully agree but I hear you. Yet in french the Mendeleiev table if called the preriodic table of elements, and is made of types of atoms. Those atoms are called elements by definition. Elements in this sense are sets of classes of atoms. Punning allows us to use the same item for both, which is really convenient. Like I said they uses too items probably because they can't.
One question about their ontology : what (do you think) are the tokens of their Elements and Hydrogen classes ?
@Emw: It seems completely unnecessary and unprecedented in the Semantic Web, and would be a colossal source of confusion for users.

I absolutely do not agree, It's very useful in a number of usecases, avoid unecessary items, and allows to make sets of classes, which is a huge thing for Wikidata. I'll highlight all the usecases I pushed and pushed and pushed, for which you can't have a valid solution because you can't make the difference beetween a token class and a class of class. The latter beeing what we usually call a classification. You can't say

< French region > subclass of (P279) View with SQID < type of administative division >

because that would be inconsistent per transitivity of the membership, just

< French region > subclass of (P279) View with SQID < French administrative division >

. But

< French region with more than 1 million inhabitant > subclass of (P279) View with SQID < French administrative division >

is also true (although we might of course might not want this precise class in Wikidata, of course, just an example). In those cases, it's pretty convenient to say

< French region > instance of (P31) View with SQID < type of administative division >

but not

< French region with more than 1 million inhabitant > subclass of (P279) View with SQID < type of french administative division >

to say the latter is not a member of the classes used by french administrative division

About the {{Type}} and {{SubType}} templates, OK I use this words and this can be confusing. If X is an instance of Y, let's call Y the type of X. A type T' (a class of individual) is said to ba a subtype of T if we can always say that if an instance I has for type T', it also has T has a type. TomT0m (talk) 10:31, 5 May 2014 (UTC)
TomT0m, the obvious formatting errors in your reply -- internally inconsistent paragraph indentation and yet another currently redlinked template -- indicate that you are still not giving due consideration to your readers by simply clicking "Prévisualiser" ("Preview"), glancing for five seconds at what you've written, and fixing mistakes. Again, please take greater care in your writing.
You ask "One question about their ontology : what (do you think) are the tokens of their Elements and Hydrogen classes ?" The Dumontier ontology contains no tokens, i.e. it has no instances or individuals. It's all classes; "is-a" is rdfs:subClassOf. See Understanding and using the meaning of statements in a bio-ontology: recasting the Gene Ontology in OWL by Aranguren et al., 2007. This is consistent with how most ontologies in physics, chemistry and biology model the world. Can you show me any ontologies that use your modeling proposal?
You also say "I'll highlight all the usecases I pushed and pushed and pushed, for which you can't have a valid solution because you can't make the difference beetween a token class and a class of class." The solution here is to generally avoid so-called "type classes". Your use cases include a 'type of administrative division' item, yet Wikidata currently has no such item. This leads me to believe that 'type of administrative division' is extraneous, like 'atom type class'. The use cases you have cited seem like a solution in search of a problem.
Returning to your question "what do you think are the tokens of their 'element' and 'hydrogen' classes?", I would answer, of course, that those tokens are actual concrete instances of hydrogen with unique locations in space and time -- a particular cloud of hydrogen gas, for example. Unlike the class hydrogen, a particular cloud of hydrogen gas would not conventionally be considered a class. The cloud of hydrogen is a set of hydrogen atoms, yes, but it is a set in which the proper membership relation is part of (P361), not instance of (P31). (The cloud of hydrogen is an object aggregate of hydrogen atom objects; see pages 4 through 10 here.) In the Dumontier ontology, that particular cloud of hydrogen would be an instance of the hydrogen class, and thus also an instance of element, since (hydrogen, subclass of (P279), element) as outlined in the hierarchy from my previous post. Emw (talk) 04:05, 7 May 2014 (UTC)
So it's an instance of pure substance, isn't it ? TomT0m (talk) 09:37, 7 May 2014 (UTC). OK, this is for the english definition. How would you go for the french definition ? An element is a subclass of atom, then. We got an interwiki conflict here, but let's suppose we got an item <French definition chemical element>. TomT0m (talk) 11:05, 7 May 2014 (UTC)
to add a cultural argument and make my question more contextful, an actual newspaper blog article about the hypothetical discovery of a new element, in french (sorry, but it's appropriate). The matter in question is the new element, which probably will exist as a single atom, not as a pure substance. Just a single one atom existing, and yet they are speaking to add a cell in the periodic table of the elements. It's not a pure rethorical or theorical question, it's a real concept actually widely used in france at least, I don't know in the rest of the world.TomT0m (talk) 12:54, 7 May 2014 (UTC)

──────────────────────────────────────────────────────────────────────────────────────────────────── TomT0m, here are the definitions of chemical element on the English and French Wikipedias, with translations from Google Translate of each:

  • Chemical element: "A chemical element is a pure chemical substance consisting of a single type of atom distinguished by its atomic number, which is the number of protons in its atomic nucleus."
  • French translation: "Un élément chimique est une substance chimique pure constituée d'un seul type d'atome qui se distingue par son numéro atomique, qui est le nombre de protons dans le noyau atomique."
  • Élément chimique: "Un élément chimique désigne l'ensemble des atomes caractérisés par un nombre défini de protons dans leur noyau atomique."
  • English translation: "A chemical element designates the set of atoms characterized by a defined number of protons in their nucleus."

I see the difference in perspective you mention, but I do not think there is some fundamental disagreement between French- and English-speaking chemists on the ontological nature of chemical elements. I think the difference is a matter of copyediting, not philosophy. In any case, the Dumontier ontology does not state or entail that element is a subclass of atom, nor does it state or entail that atom is a subclass of element.

However, there seems to be a glaring inconsistency between the Dumontier ontology and English Wikipedia on the definition of "chemical compound". Here is how English article defines the concept: "A chemical compound is a pure chemical substance consisting of two or more different chemical elements[1][2][3] that can be separated into simpler substances by chemical reactions."

This means that English Wikipedia -- presumably accurately summarizing reliable sources in references 1, 2 and 3 -- considers the statement (chemical compound, subclass of, pure chemical substance) true. The Dumontier ontology, however, states (pure substance, disjoint with, compound) and (pure substance, subclass of, chemical substance) and (compound, subclass of, chemical substance). Surely 'pure substance' is the same as 'pure chemical substance' and 'compound' is the same as 'chemical compound'. Simply put, English Wikipedia says all chemical compounds are pure substances, while the Dumontier ontology says no chemical compounds are pure substances. Which is correct?

The ChEBI ontology seems generally better than the Dumontier ontology. It is vastly more comprehensive, more frequently updated, part of large applications, and has institutional backing (from EBI). It can be explored through the ChEBI web-based ontology explorer, or through the ChEBI ontology web interface for specific concepts or by opening the URL ftp://ftp.ebi.ac.uk/pub/databases/chebi/ontology/chebi.owl (big file!) in Protege.

Interestingly, the ChEBI ontology sets 'element' as a synonym of 'atom', and 'hydrogen' as a synonym of 'hydrogen atom'. It has a separate item for 'hydrogen molecular entity'.

Below is an abbreviated subclass hierarchy from ChEBI for 'hydrogen atom' and some other concepts.

ChEBI ontology (big 91 MB source). Many classes omitted for brevity.

  • Thing
    • chemical entity
      • atom (synonym: element)
        • main group chemical atom ...
        • metal atom ...
        • nonmetal atom ...
        • s-block element atom ...
          • alkali metal atom ...
          • alkaline earth metal atom ...
          • helium atom ...
          • hydrogen atom (synonym: hydrogen)
            • deuterium atom
            • protium atom
            • tritium atom
      • chemical substance
      • group ...
      • molecular entity
        • elemental molecular entity
          • elemental carbon
            • diamond
            • graphite ...
            • graphene
            • monoatomic carbon
              • carbide(1-)
              • carbide(4-)
              • carbon(1+)
          • elemental hydrogen
            • dihydrogen ...
            • dihydrogen(.1+) ...
            • monoatomic hydrogen
              • hydride ...
              • hydrogen(.) ...
              • hydron
        • ion
          • anion ...
          • cation ...
        • polyatomic entity
          • heteroatomic molecular entity
          • macromolecule
            • biomacromolecule
              • information biomacromolecule
                • polynucleotide
                  • nucleic acid
                    • deoxyribonucleic acid
                      • double-stranded DNA
                      • single-stranded DNA
                    • ribonucleic acid
                      • dsRNA
                      • messenger RNA ...
                      • ribosomal DNA
                      • small nuclear RNA
                      • transfer RNA ...
                • protein
                  • amyloid fibril
                  • lipoprotein
                    • high-density lipoprotein
                    • low-density lipoprotein
                    • oxidised LDL
                    • very-low-density lipoprotein
                  • metalloprotein
                    • iron protein
                      • hemoprotein
                        • cytochrome ...
                        • globin
                          • hemoglobin ...
                          • myoglobin
                  • histone
                  • simple protein
                    • apoprotein
                      • apolipoprotein
              • polysaccharide ...
            • polypeptide
              • beta-amyloid ...
              • insulin (human)
          • molecule ...
    • subatomic particle
      • boson
        • meson ...
        • photon
      • fermion
        • baryon ...
        • lepton
          • electron
        • quark ...
      • nuclear particle
        • nucleon
          • proton
          • neutron
    • unclassifieds

What are your thoughts on the ChEBI ontology? I would be interested in looking into it further. Emw (talk) 03:59, 9 May 2014 (UTC)

  • I like this ontology, it is consistent with itself and pretty much everything we said in this discussion, plus more. Yet because of the lack of punning they can't express the french definition of elements : in their ontology, it is a synonym for atom. I'm cool with this in a model without punning as long as it's self consistent and expressive, but I think the french definition is really useful, for two reasons : first, big trees like this are not so easy to organize and navigate. Adding additional informations like type of classes is useful for that.
Subclass trees are not difficult to organize or navigate, TomT0m. Want to see the structure of a tree and the items in it? Use Wikidata Generic Tree. Want to re-organize the tree? We have a great UI for that. Notably, this is the way the rest of the Semantic Web organizes this kind information. Want to know what's really difficult to organize and navigate? A completely unconventional modeling system that noone else in the world uses. That is what you are proposing. Emw (talk) 12:57, 10 May 2014 (UTC)
  • Another point : Wikidata's application are not their application, they are comprehensive in their domain, and they probably want to use the ontology to work with actual molecules or so. In Wikidata, the primary objectives are different, we want to build and model a Mendeleiev table before we actually can express the whole chemistry field :) Then it's convenient to have a reified concept for hydrogen as an individual usable as rows of the periodic table using the same item we use to represent the set of hydrogen atom. hence we can express both (THE hydrogen atom in my cup of tea yes, he is unique and I'm very fond of him, instance of (P31), hydrogen) and (hydrogen, part of (P361), Mendeleiev table) (I don't seek to be entirely accurate) without creating a new item.
EBI, the maintainers of ChEBI, is a bioinformatics institute. They generally don't work with actual molecules, to my understanding. ChEBI: a database and ontology for chemical entities of biological interest is worth reading to get a better understanding of the goals of that project. Those goals overlap with Wikidata's much more than not.
There is simply no need to explicitly have claims like (hydrogen, instance of, chemical element) to build a periodic table of elements. We can use the claim the rest of Semantic Web uses -- (hydrogen, subclass of, chemical element) -- and model 'hydrogen' as an instance with punning at run time. Emw (talk) 12:57, 10 May 2014 (UTC)
  • a last point : this ontology seems entirely consistent with the model I proposed earlier (If we got two different concepts for elements). So compared to this, we just add informations, tag the tree, and add the power to express definitions of concepts as we use them in the real world, without losing any consistency, which seems convenient to me. Overall we just gain in expressive power and just have a simple concept to explain to newbies, but I don't think anyone will be shocked by the fact te hydrogen concept has several facets and that the member/subset relationship is out of reach of anybody if explained clearly. (@Emw:) TomT0m (talk) 12:19, 9 May 2014 (UTC)
The ChEBI ontology is not consistent with a central aspect of your model. It has no explicit statement (hydrogen, instance of, chemical element). The ChEBI ontology has no explicitly-declared individuals or instances. It is all classes, like the Dumontier chemistry ontology: it uses only rdfs:subClassOf and not rdf:type (nor both) as you propose. Emw (talk) 12:57, 10 May 2014 (UTC)
@Emw: It's a matter of definitions only, Wikidata supports different definitions if they are associated with different items. So adding an item with the french definition, maybe not labelled chemical elements, can play this role without any problem. Without introducing any inconsistencies in particular (I think). TomT0m (talk) 20:10, 10 May 2014 (UTC)

I don't like the idea that because punning is not used in the world on ontologies, it seems like an evidence to me, so I just googled a little bit about punning. see http://wiki.opensemanticframework.org/index.php/Ontology_Best_Practices for example : I quote : Treat many concepts via "punning" as both classes and instances (that is, as either sets or members, depending on context). The "punning" technique enables "metamodeling," such as treating something via its IRI as a set of members (such as Mammal being a set of all mammals) or as an instance (such as Endangered Mammal) when it is the object of a different contextual assertion. Use of "metamodeling" is often helpful to describe the overall conceptual structure of a domain. See endnote[18] for more discussion on this topic .

So it seems the best practices are not what you think they should be for everyone in the world. Not using punning seems to me like a headeach because there is many many objects in the world that can be conceptually describing both a class and can be classified themselves as instance of something, take the example of a car model. There is many instances of a car model, some of which may be notable, yet it's convenient to be able to say

< Porsche 911 > instance of (P31) View with SQID < car model >

. Since there is no theorical problems about that, think of the complication it would be not to use this. Less items, less properties, same expression power. Seems a really good example of Ockham rasor in the model due to adding expressive power. This is what parcimony is about : don't introduce new concept if you can do the same without them, but introduce them if this in fact reduces complexity or is necessary. TomT0m (talk) 13:39, 10 May 2014 (UTC)

TomT0m, I am aware of punning and the canonical Harry the Eagle example cited in endnote 18. I have not said that punning itself is a bad idea. What I have said is that modeling things like chemical element with the explicit statement (hydrogen, instance of, chemical element) is unconventional in Semantic Web ontologies, and that your proposal to add statements like (s, instance of, atom type class) seems extraneous and unhelpful.
Looking further into the ChEBI ontology in Protege reveals that while it has no individuals or instances declared explicitly via rdf:type (i.e., instance of), many of its classes do have object properties, like 'has part' -- e.g. (atom, subclass of, chemical entity), (atom, has part, atomic nucleus), (atom, has part, electron). Object properties can only be applied to individuals. Protege infers that 'hydrogen atom' has parts 'atomic nucleus' and 'electron' because it is a subclass of 'atom'. Thus, ChEBI uses punning but achieves it without explicit rdf:type statements, notably without statements like (hydrogen, instance of, chemical element). Emw (talk) 02:58, 13 May 2014 (UTC)
The statement is just the translation of the definition of the definition given in the french article. This is just a matter of NPOV. There is no real good argument here not to include it. We just will not tag the class as a member of the ChEBI ontology, if it has a compatible licence. I'll add some argument : it makes explicit something that is implicit in the CheEBI ontology, so it's pretty non high skilled human friendly. Not everyone is a Protege Ontologist :) Through proper labelling of the item and add information that are redundant in a reasoning context we help the humans, Wikidata is for both. TomT0m (talk) 10:53, 13 May 2014 (UTC)
TomT0m, you're waffling. In one post you espouse "parsimony" and "Ockham's razor" and "best practices", then when I demonstrate that one of the world's most highly-referenced, reliable chemistry ontologies does all those things without using statements like (hydrogen, instance of, chemical element), you suddenly do a 180 and contradict yourself by talking about how including redundant information -- the opposite of parsimony and Ockham's razar -- is actually a good thing.
You then resurrect an assertion about translation that I rebutted over a week ago before you decided to "reboot" a discussion that was inconvenient for your position. Finally, your argument that it is better for human reasoning to make explicit redundant statements like (hydrogen, instance of, chemical element) and (hydrogen, subclass of, chemical element) when the former is implied is very strange. I can confidently tell you that people will not fuss about there being a lack of an explicit 'instance of' claim even though an item uses object properties that can only be applied to individuals. Omitting the extraneous statement (hydrogen, instance of, chemical element) would eliminate a major source of confusion among users, be in line with Semantic Web conventions, and not violate NPOV -- which is probably the most esoteric invocation of that concept on Wikimedia, ever. Emw (talk) 13:00, 13 May 2014 (UTC)
We're going on personal arguments and not on discussion again. About the definition, I don't see how you can seriously justify there are several one who are just a matter of viewpoint but does not translate differently, where it's an exact transcription. The choices of Wikimedia projects in those cases is to state all the definition. All the rest is shoehorsing something that is really not a problem. If we need one item labelled class of elements with the same atomic number in english and chemical elements in french and you want to label the atom class element as an alias or whatever, I'm fine. But I don't see how you made a convincing case not to state that definition. The words used are not the same, and we can pretty much handle it fine in Wikidata. The Ockham Rasor was used for the modeling tools we used : we can trade another modelling concept with more compact models. Which allow to add details and annotation for a human to explore the model without more cost in the number of items, so it's clearly beneficial. After that, I don't think minimising the model makes it more readable to human, which means parsimony does not apply : we lost something if we do that. Parsimony implies we can minimise the model if the software can make the inferences itself. But Wikibase does not make any inferences, yet it is supposed to be explored by human. I'm fully consistent in my argumentation, no contradiction :) TomT0m (talk) 15:24, 13 May 2014 (UTC)
TomT0m, asserting that statements are contradictory and that rebuttals have been ignored is not a personal argument. Again, and this is not personal, please take more care in your writing -- read over what you write before clicking "Enregistrer" ("Save"). Your second sentence is so syntactically malformed and assumes so much hidden and ambiguous context (which definition(s)? what is the "exact transcription"?) that it's impossible to understand what you're saying. My guess is that you're asserting that the statement "hydrogen is an instance of a chemical element" is an exact translation of the French Wikipedia article's definition of the subject in élément chimique or hydrogène. It clearly is not; your assertion is effectively the same as the one I rebutted over a week ago. Your comment on the subject here is simply reiterating your premise; still ignoring, still not addressing the rebuttal of that premise. If you dispute the rebuttal, please read the linked rebuttal, perhaps even quote which part you're disputing, and specifically explain how it is incorrect.
The way you shift your portrayal of Ockham's razor and parsimonious modeling when rebutted is incredible. Let's review the point at hand. We have an item, hydrogen (Q556), which currently has the claims (hydrogen, subclass of, chemical element) and (hydrogen, instance of, chemical element). I have demonstrated that the second statement is inferred and unnecessary for metamodeling. To be clear, the entire point of punning -- which you have cited repeatedly as a reason to retain (hydrogen, instance of, chemical element) -- is to enable inference by machines, not humans. Humans have no problems interpreting a class as an instance, and the current lack of native automated inferencing is irrelevant, since arguing about punning is only relevant in the context of automated inferencing. The fact that the explicit claim (hydrogen, instance of, chemical element) is unnecessary to achieve desired automated inferencing means that keeping that explicit claim is the opposite of Ockham's razor and parsimony.
But wait! We need the redundant claim so that hydrogen (Q556) is "more readable to human" and because it is "supposed to be explored by human", you say. That's absurd. There would be no decrease whatsoever in the readability or explorability of the 'hydrogen' item if only (hydrogen, subclass of, chemical element) were used, as compared the having both claims (hydrogen, subclass of, chemical element) and (hydrogen, instance of, chemical element) on the item. Quite the contrary: having both claims is redundant and more confusing for users. No French users will come along and thank us for having both explicit claims on 'hydrogen'. They would be perplexed about what the difference between (hydrogen, subclass of, chemical element) and (hydrogen, instance of, chemical element) is and why both statements are used, be compelled to read up on this needless esoteric distinction, grumble about our overcomplicated data modeling, and be more likely to bid Wikidata adieu. Emw (talk) 03:32, 15 May 2014 (UTC)
@Emw: This should not happen. The items should be labelled differently in the same language. In french:
In english:
Let's study the set of <Hydrogen> <Helium>, ... classes. What do they have in common ? All of the atoms of the member of those classes, respectively, are identical. Let's take the characteristic property of Hydrogen. Let atoms(CLASS) be the function that returns the set of atoms of the instances of this class, maybe decomposing the molecules if there are some.
Then, for hydrogen, we got for all atom in atoms(<hydrogen>) atomic_number(atom)=1 for Helium we got for all atom in atoms(<Helium>) atomic_number(atom)=2. Now lets call the class Element the class of all classes C who have a characteristic property exactly There exist a number at such that for all atom in atoms(<C>) atomic_number(atom) = at.
Then we get a transcription of element into the french definition. Now the question, is this useful ? I think yes. There is other kind of atom classes, such as isotope. If we want to query the atom classes who are elements, we can use this definition. The same if we want to query all isotopes. In another sense, think of it as a reification of the query you would build to make to return exactly those classes. Just query the direct subclasses of atoms might not be enough: think of a class atoms with a specific kind of neutron in it, it's not an element class, yet it might be a direct subclass of atom. It's a matter of robustness of the changes that can be done in Wikidata, which we can't really see long term. TomT0m (talk) 10:30, 15 May 2014 (UTC)

Object aggregation in other fields

Emw, thanks for your comment because I think it is applicable to other fields as well. For instance now I'm working on the classification of Chopin works. I had a ontological dilemma when figuring out the role of Polonaises (Q3492430) and its composing parts. It is not an object per se, but an aggregation of objects. At the beginning I thought of considering it a class, but it cannot be a class when it is just an object aggregation. So now I just labeled as such: Polonaises (Q3492430) <instance of> object aggregation (Q16773055). In this case I assume that it inherits the values "instance of" of its composing parts, do you think it is necessary to be explicit about this inheritance (by actually creating the statement Polonaises (Q3492430)<instance of>polonaise (Q203478)), or it can be just assumed? I'm asking because in this case it is easy, but sometimes the composing parts can have different values of "instance of", thus making maintenance complicated (unless there was a bot checking for such things).--Micru (talk) 06:46, 7 May 2014 (UTC)

Micru I guess if you take the class of all Chopins works, then <Polonaises> is a subclass of it. If you want to state that they are all polonaise (Q203478) View with Reasonator View with SQID , then they also are a subclass of (P279) them.
< Polonaises (Q3492430) View with Reasonator View with SQID  > instance of (P31) View with SQID < object aggregation >
is redundant with the fact that it is a class, as a class is a set of object and one of our basic unit of classification. Adding a statement for this class to be an instance of something is useful if you want to retrieve all the classes that class Polonaise wrt. their author, for example, in which case you could add
< chopin's polonaise > instance of (P31) View with SQID < polonaises by author >
, as the Wikipedia category might exist. TomT0m (talk) 11:00, 7 May 2014 (UTC)
TomT0m, the class "works by chopin" does not exist (unless you consider Category:Compositions by Frédéric Chopin (Q7132896) to be it), and it doesn't make sense to create a class for each author works when it can just be queried. What exists is the List of compositions by Frédéric Chopin by genre (Q1785783), which is an instance of the class Wikimedia list article (Q13406463). This instance "consists of" several parts, and these parts are in turn aggregated objects of other parts. Please forget about class classification for an instant (which is also correct) and let's focus on the parts of an instance. What are the properties of the "intermediate building blocks"? --Micru (talk) 11:43, 7 May 2014 (UTC)
It does not make real sense to say that the Chopin polonaise article in Wikipedia is about a list of Chopin Work. It's about some CHopin's Work who have something in common, which in turns lead the makers of the catalogue to use that to sort the datas and build the catalogue. A musical album consists of a list of musical pieces, no question about that, but it make more sense to model the subjsct article for what it is. An the subject is neither a catalogue nor an album. It's some Chopin's work in the Polonaise style. TomT0m (talk) 12:01, 7 May 2014 (UTC)
It will not make sense to you because you want to see classes everywhere even when it is cumbersome to do so. In this particular case the wikipedia pages primarily represent a sub-list of works. I hope you don't want to consider lists as classes...--Micru (talk) 12:39, 7 May 2014 (UTC)
Well, in a fair number of cases, lists are just arbitraly ordered collection. Mathematically, the difference beetween a list and a set is that the order of the elements does not matter in the case of sets, and that an element belongs to a set only once, where in a list an element can appear any number of time and the order in which the elements appears is meaningful (any element has a rank in the list). So yes, in a number of cases where the order in the wikipedia articles is just alphabelically sorted or other kind of not so meaningful sorting, I would say the article are there to list all the elements of a class (sometimes it's even exaustive). So I would even ask you a further question : why would you add in the composed of statements the whole set of these works where, like the class item we should not create, could be queried (malicious smiley :)) TomT0m (talk) 12:49, 7 May 2014 (UTC)
I know it is not necessary but before using the "tree view" I was getting lost in the forest of music works, so I just used "consists of" as a scaffolding to build the highrise :) --Micru (talk) 13:00, 7 May 2014 (UTC)
OK, now imagine we call that a "class item", and that to each class we create we associate a query (namespace Q) that list the instances ? Does this now seems redundant to create the class and the subclasses items ? TomT0m (talk) 13:17, 7 May 2014 (UTC)
As soon as you consider it a class you will be flattening the structure, and I want to keep it as it is now, as a nested collection of (sorted) lists. --Micru (talk) 14:20, 7 May 2014 (UTC)
Micru Seems like you're trying to reflect the Wikipedia organisation of Wikipédia in Wikidata. What I did for that a while back for Unicode block (Q3512806) View with Reasonator View with SQID (by the way, is this article a list in the Wikipedia sense ?) was something like
< unicode characters from 0000 to FFFF > part of (P361) View with SQID < unicode characters >
,
< unicode characters from 0000 to 1111 > part of (P361) View with SQID < unicode characters from 0000 to FFFF >
, ... and the corresponding caracters. Thar was for some cases like this I wanted to create the subsequence of and in list properties, because I as unsatisfied of part of as part of does not have a sequence semantics, only a composition one. But it's a little bit different as the list of unicode characters is naturally ordered by the natural order on their codepoints. With the works of Chopin, you don't really order them by type, you class them ... TomT0m (talk) 15:20, 7 May 2014 (UTC)
Last thing, why do you say a class hierarchy is flat ?????? TomT0m (talk)
Part of can have (simple) sequence semantics, you just need to use "followed by" as qualifier and point to the next part. You can order whatever you want, that is why such classifications exist. Regarding the flat structure, I said that because if you consider "Polonaises by Chopin" as a query of instances, then you are missing the structures of composition (part of) that are in between.--Micru (talk) 17:10, 7 May 2014 (UTC)
I don't follow you, in beetween the Polonaise ? TomT0m (talk) 19:34, 7 May 2014 (UTC)
What I meant is that you would get as result a list of instances, like Polonaise, Op. 26, No. 1 in C-sharp minor (Q11339374) or Polonaise, Op. 26, No. 2 in E-flat minor (Q16766398), but not that they form a group called Polonaises, Op. 26 (Q3071076).--Micru (talk) 05:58, 8 May 2014 (UTC)
As a result of what, Succu ? I don't follow you. If you take the subclass tree of Chopin's work, and then take the leaf's of the tree and lists their instances, isn't it exactly what you want ? The subclass tree is the organisation structure and can be explored, and the instances are the works. You can even query the instances if the classes inside the tree, which are larger groups, to get the larger group of instances. TomT0m (talk) 11:03, 8 May 2014 (UTC)
No, that is not what I want, I want a tree of *groups of instances*, whenever they exist, not of *classes*. You are of the opinion that *everything* is either a class or an instance, but that view is making you fail to understand that there is also object composition (Q390066) or aggregation, and that it also should be accounted in the classification. So please, go read the wikipedia article and make sure that you understand that principle, because otherwise I see no point in continuing this conversation.--Micru (talk) 11:43, 8 May 2014 (UTC)
@Micru: I'm perfectly fine with with statements like
< Human body > composed of search < heart; skin ... >
, and I created a bunch of these statements. For aggregation though, I only encourntered that term in UML modeling, which is pretty different from RDF or OWL modeleling. So as the class classification system works really fine for me here and seems to achieve everything we want to do without losing any way to use the datas for doing whatever you want to do, I don't see the usefulness to introduce this. A class can be viewed by a set of object (or by the properties of this object). Here we can perfectly extract this set by saying we want all the Polonaises made by Chopin. Per that definition it's a class. What does an aggregation adds (it's even not well defined in UML ... my UML teacher from a few years ago told us students that they where still trying to figure this out writing thesis on the subject.) ? TomT0m (talk) 11:57, 8 May 2014 (UTC)
Let's try another way... which class represents for you Polonaises, Op. 26 (Q3071076)?--Micru (talk) 12:50, 8 May 2014 (UTC)
@Micru: Well, En musique, noté « op. », il permet de situer un morceau de musique dans la production d'un compositeur ; il est alors appelé numéro d'opus, et suit l'ordre de publication. (in music, it allows to position a musical piece in the production of a composer. So opus. 26 is a rank order in a sequence, we just have to know which one, is it in the sequence of the polonaise by Chopin ordered by their writing number ? Or all it's musical production ? I don't know, maybe you do.
It's also a <Polonaise> and a <Musical piece>. So
< Polonaises, Op. 26 (Q3071076) View with Reasonator View with SQID > instance of (P31) View with SQID < Polonaise >
if <Polonaise> is the set of all musical pieces in this style. (and
< <Polonaise> > subclass of (P279) View with SQID < musical pieces >
.
If it's a class, let's take all the times musician played it. Then they are instances of this class (which are obviously tokens of the real world).
So the only not totally settled question is how we represent sequences in Wikidata. I think we just need a common way. TomT0m (talk) 13:08, 8 May 2014 (UTC)
Actually it is a set of 2 polonaises. Or if you prefer to put it in simpler way, if I have an orange and I put it together to an apple, then I have a group of fruit, not an instance of a fruit. In the context of "group of instances" you can represent a sequence, but if you don't acknowledge its existence of a group beyond classes then it is pointless to speak about sequences.--Micru (talk) 13:23, 8 May 2014 (UTC)

──────────────────────────────────────────────────────────────────────────────────────────────────── Fair point. Then it's a subclass of Polonaise with exactly two instances, we can express that without problem using classes. I actually created the group of humans (Q16334295) View with Reasonator View with SQID which has for example duo (Q10648343) View with Reasonator View with SQID as a subclass, (using classes), so I think I should agree with you, yet I'm not sure I unerstand where you are going. But I'm not sure I follow you. I don't know why this Polonaises are grouped together, but we probably could use the same approach. TomT0m (talk) 13:47, 8 May 2014 (UTC)

I'm glad that we are doing some progress. Now we are going one level up, back to the tree, where there are all kinds of groups. Groups of mazurkas, of bourrées, of nocturnes, even a group of miscellaneous compositions. The groups are very diverse and theoretically there should be a class for each one of them, but I'm too lazy to create one class for each one of these groups. So, what if we just have a generic class that represents a group of diverse instances?--Micru (talk) 14:27, 8 May 2014 (UTC)
@Micru: What's the meaning of this groups in musical studies ? If we get good definitions, then we can discuss on how to model that. But I think it's better to come back from the field before trying to go generic, as there is probably precise definition for opus, ... and so on, in classical music. Then we can think of how regroup the members of these Opus in a group. Is it like a musical album ? Is an opus supposed to be the set of pieces in a standard music concerts ? TomT0m (talk) 14:38, 8 May 2014 (UTC)

Modulo interwiki conflics, the relevant item about opuses seems to be opus number (Q385271). I would define it as a set of musical work of a classical music composer published together used in work identification. Is that correct ? We should add that that opuses are usually indexed by their number in the chronological order of their publications. And that each work blelonging to the opuses have an index number. TomT0m (talk) 15:13, 8 May 2014 (UTC)

I edited a little bit Études, Op. 25 (Q16741233) View with Reasonator View with SQID as an example, which could become a showcase item eventually. I also found another item magnum opus (Q491014) View with Reasonator View with SQID which seems to be

< magnum opus (Q491014) View with Reasonator View with SQID > subclass of (P279) View with SQID < opus number (Q385271) View with Reasonator View with SQID >

, but not sure yet as the cataloguing is not totally consistent, Mozart is not catalogued by Opus if I understood and could maybe have an Opus Magnus. TomT0m (talk) 15:19, 8 May 2014 (UTC)

I removed your edit because it makes no sense. It is like saying that a car is an instance of its plate.--Micru (talk) 21:37, 8 May 2014 (UTC)
If you go to Wikimania we can continue with our conversation there. I see no point in continuing it in written form.--Micru (talk) 21:39, 8 May 2014 (UTC)
@Micru: I don't go to Wikimania unfortunately. But it makes perfect sense to say an Opus is an Opus. I really don't follow you. TomT0m (talk) 22:08, 8 May 2014 (UTC)
An opus is an identifier.--Micru (talk) 06:13, 9 May 2014 (UTC)

Pictogram voting comment.svg Comment: Micru, apologies for the delayed reply. You asked "So now I just labeled as such: Polonaises (Q3492430) <instance of> object aggregation (Q16773055). In this case I assume that it inherits the values "instance of" of its composing parts, do you think it is necessary to be explicit about this inheritance (by actually creating the statement Polonaises (Q3492430)<instance of>polonaise (Q203478)), or it can be just assumed?"

When I noted that a cloud of hydrogen gas is an object aggregate of hydrogen objects, I was referring to the ontological concept 'object aggregate' from the Basic Formal Ontology (BFO), a widely-used upper ontology. Here is an introduction to the concept from page 9 of the document I referenced, On Classifying Material Entities in Basic Formal Ontology:

"An object aggregate is a material entity consisting exactly of a plurality of objects as continuant parts.
More formally:
If a is an object aggregate, then if a exists at t, there are objects o_1, .., o_n at t such that:
for all x (x part of a at t iff x overlaps some o_i at t)
An entity a is an object aggregate if and only if there is a mutually exhaustive and pairwise disjoint partition of a into objects [Bittner and Smith 2008]. Examples are: a symphony orchestra, the aggregate of bearings in a constant velocity axle joint, the nitrogen atoms in the atmosphere, a collection of cells in a blood biobank."

So, in BFO, for something to be an object aggregate it must be a material entity. BFO defines 'material entity' on page 9 of On Classifying as "continuants that include some portion of matter as part, where 'portion of matter' is understood in the broadest possible sense to include also undetached portions of matter such as your head and scattered portions of matter such as the Buffalo Symphony Orchestra. Every material entity is localized in space. Every material entity can move in space. Every entity which has a material entity as part is a material entity." Presumably, you do not intend to model of Chopin's polonaises as portions of matter. Thus, in the sense you intend to classify them, Chopin's polonaises are not object aggregates. Regarding your question about inheritance of type from a part to its whole ("...or it can be just assumed?"), I would say: no, a whole does not inherit the object of its parts' instance of (P31) (rdf:type) statements. For example, an orchestra can be said to be an have musicians as its parts, but an orchestra is not a musician.

I haven't found a BFO-based ontology that gives a fantastic treatment of musical works. The Information Artifact Ontology (IAO) might be the closest, but it's not really close enough. IAO can be explored most easily here by expanding 'information content entity' at left, etc.

It seems more work has gone into representing musical works with the FRBR model, which you're probably more familiar with. The Application of FRBR to Musical Works by Christopher Holden talks more about that. The FRBR-aligned Bibliographic Ontology (FaBiO) also seems relevant. Even after looking through all of those, though, it's unclear to me how the item 'polonaises by Chopin' would be classified. It's clearly a collection of works. In line with a position I explained in January, I think I would consider 'polonaises by Chopin' to be a subclass of polonaises -- more specifically, equivalent to the class of things that are a polonaise and where the creator is Chopin. Emw (talk) 13:00, 9 May 2014 (UTC)

@Emw:, thanks for your clarification. Regarding the FRBR, well, whatever distiction we make we are still moving at "work" level, so I guess we should use "instance of:[creative] work collection". We'll have to continue mapping ontologies at the "book task force" and see what are the missing parts.--Micru (talk) 19:23, 11 May 2014 (UTC)

Maybe another example helps. Recently Group (Q4150646) was tagged as subclass of (P279) of cultivar (Q4886). At first sight this seems to be OK to treat Group (Q4150646) as a specialization of cultivar (Q4886), but it's defined as an assemblage of cultivar (Q4886). So in my opinion Group (Q4150646) should be treated as a superclass of cultivar (Q4886). --Succu (talk) 19:26, 9 May 2014 (UTC)

Succu, is every instance of Group (Q4150646) an instance of cultivar (Q4886)? Is every cultivar (Q4886) an instance of Group (Q4150646)? If not, then subclass of (P279) is not the correct relation in either case, per rdfs:subClassOf. Emw (talk) 03:36, 10 May 2014 (UTC)

Population statistics?

Ar there bots who systematically add the population of localities here? In the Russian Wikivoyage, we use an extention which adds the population directly to the infobox, and we discovered that there are very few localities on Wikidata with the population (or we have been unlucky not to hit them). Is there any help needed from not bot-owners?--Ymblanter (talk) 19:29, 10 May 2014 (UTC)

It would be easy to mass-import population from infoboxes, but sourcing would be an issue, given that some people are against the use of imported from (P143). So while the import is possible, it would likely be controversial.--Underlying lk (talk) 13:44, 11 May 2014 (UTC)
It does not seem to make much sense import it from Wikipedia while in most case, we have much better, and usable, data from national statistics agencies and the like. Indeed, user:ValterVB has done a good job at importing this for Italy. The main issue is copyright (see for instance Wikidata:Requests for permissions/Bot/ValterVBot 12). --Zolo (talk) 14:05, 11 May 2014 (UTC)
Yes for now I have stopped for problem with copyright, isn't clear if I can import data for italian population. But if are available data with CC0 I can import them. --ValterVB (talk) 14:13, 11 May 2014 (UTC)
May be we should discuss it separately per country. For example, for Russia population data for all districts, towns, and urban-type settlements are available from the 1989, 2002, and 2010 Censuses. These data were manually added to the English Wikipedia (which does not have articles on all urban-type settlements, but has all towns and districts) and to the Russian Wikipedia (which has all the articles). Would it be an option to import them from Wikipedia and link to the censuses? I have done the same for Kazakhstan in the English Wikipedia (1999 and 2009) and now I am doing it for Ukraine (2013). Should I try to formulate separate countries as bot tasks?--Ymblanter (talk) 14:53, 11 May 2014 (UTC)
I think that for countries for which the license is ok we should discuss it country by country (or database by database), but for countries for which the issue is the license, we should try to discuss general solutions. Actually apart from the US, I do not know which countries have CC0-compatible data. There are lots of things that were imported to Wikipedia but that seem tricky to have on Wikidata license-wise (because it is a CC0 semantic database rather than a CC-by encyclopedia).--Zolo (talk) 15:41, 11 May 2014 (UTC)
Me neither, May be indeed asking the legal would help.--Ymblanter (talk) 16:31, 11 May 2014 (UTC)
we already asked legal (answer on meta here). The crucial quotes "In the absence of a license, copying all or a substantial part of a protected database should be avoided." but see the whole thing, or at least the 'Conclusion'. Filceolaire (talk) 21:05, 13 May 2014 (UTC)
Thanks, I see. For the population, there is obviously no creativity involved, so that copying it directly from the databases to Wikidata would be ok for the US and not ok for EU. (I would still like to know whether it is ok for the former USSR countries, particularly for Russia, Belarus, Ukraine, and Kazakhstan). Thus, for EU we have essentially two choices: To forget about the population data, or to copy it from Wikipedia citing Wikipedia as a source. I would certainly prefer the second option.--Ymblanter (talk) 06:40, 14 May 2014 (UTC)
Well, it does not help to cite Wikipedia, since I hope that data is from the Census organisations in the first place. And copying WP would in itself be a copyvio since we have a CC0-license. -- Innocent bystander (talk) (The user previously known as Lavallen) 07:46, 14 May 2014 (UTC)
Then we basically invite people to add this data unsourced.--Ymblanter (talk) 18:48, 14 May 2014 (UTC)
If they/we add only a fraction of the data in the databases, there is no problem. I have for example only added the latest data for Stockholm, nothing else. That is not a copyvio. The problem comes when users edit systematicly in a subject or with a bot. Random single edits is not a problem. -- Innocent bystander (talk) (The user previously known as Lavallen) 08:08, 15 May 2014 (UTC)

Request for comments: How to deal with open datasets?

During the Zürich Hackathon I met several people that looked for solutions about how to integrate external open datasets into our projects (mainly Wikipedia, Wikidata). Since Wikidata is not the right tool to manage them (reasons explained in the RFC as discussed during the Wikidata session), I have felt convenient to centralize the discussion about potential requirements, needs, and how to approach this new changing landscape that didn't exist a few years ago.

You will find more details here on meta: How to deal with open datasets. Your comments, thoughts and ideas are appreciated!--Micru (talk) 09:42, 15 May 2014 (UTC)

Mathematics

Efforts to bring structure to mathematics are now organized at Wikidata:WikiProject Mathematics. Are there more than 3 properties that currently exist and are within the scope of the project? Many more will need to be proposed. As usual all willing participants are welcome. -Tobias1984 (talk) 17:59, 15 May 2014 (UTC)

Announcement : new variant template of Q : {{Q''}}

wgArticleId:394#reasonator_template
Hi, I just created a template {{subst:Q''}}, who is intended to help in discussions. Some of you might have remarqued some users had the very good idea to, when they talk about a template, to insert the label of the template as a Wikitext comment, which for whose who visualise often diff of discussions, helps a lot. I was inspired and created a template that is intended to be substituted in discussions.

{{subst:Q''|Q1}}
produces in the wikitext
{{Q'|Q1}}  <!-- universe -->
an in the html

Universe (Q1) View with Reasonator View with SQID

Hope you enjoy PS: if some of those who already does this can gives a hint on how they do, it would be cool. Gadget ? manually ? TomT0m (talk) 18:10, 3 May 2014 (UTC)

Another thing : I also created recently a couple of templates related to properties : {{Pnumber}}, who prints a property number from an english (atm) label : {{Pnumber|instance of}} gives 31. It is used by {{P'}} : {{P'|instance of}} gives instance of (P31) View with SQID. The reverse mapping is manually created and can be completed by editing the lua module Properties.

{{C}} now can use these properties label : {{C|Children|subclass of|human}} gives

< Children > subclass of (P279) View with SQID < human >

TomT0m (talk) 20:40, 3 May 2014 (UTC)

example: quantum field theory (Q54505): quantum field theory (Q54505) View with Reasonator View with SQID @TomT0m Thanks for the template! Is it possible to add a browser item selector field for the language of your choice? This would save a lot of time.
Unfortunately Field theory (Q1262328) (the parent class) is messy: I found field theory (Q903820). לערי ריינהארט (talk) 02:47, 4 May 2014 (UTC)
the sq page is a disambiguation page, it should not be linked to a valid item
the en page is a link to a paragraph
This is not what semantically linking / mapping should look like. Regards gangLeri לערי ריינהארט (talk) 22:20, 3 May 2014 (UTC)
לערי ריינהארט It's not intended to, it's just meant to ease discussions and to make the wikitext more readable. What's wrong with that ? TomT0m (talk) 10:21, 4 May 2014 (UTC)
@TomT0m Thanks for the answer! I just search some (Javascript-) code to add optional language codes for pages as user:Gangleri/sandbox/index/extended periodic table. I worked on the el, bg, ru, sr, uk, af, ca, cs, da, es, et, eu, fi, fr, hr, hu, is, it, la, lad, lt, lv, mi, mt, nb, nn, nl, oc, pl, pt, rm, rmy, ro, rup, sk, sl, sq, sv, sw, tn, tr, zu, eo, ia, ie, io, nov, vo, en, pih, de, bar, gsw, ksh, pdc, pfl, lb WMF language codes. לערי ריינהארט (talk) 10:56, 4 May 2014 (UTC)
לערי ריינהארט I don't think any javascript code is needed, just create a links such as in esperanto this previous one in the page. Of course the user will have to change the language back but with a notice should be OK. TomT0m (talk) 12:21, 4 May 2014 (UTC)
@TomT0m: a) I know these parameters: "setlang will not work for anonymous users, only "uselang" will work for these visitors.
b) One could also use / add a template with the language codes at the top of the page. I prefer one with the languages sorted grouped by script-groups using ISO 15924 (Q16866) View with Reasonator View with SQID codes (see en:template:ISO 15924/footer). Do you know some candidates, maybe some templates with a "small" versions/options? Thanks for any help in advance.
c) The browser language selector field would be a helpful solution for any kind of "externals". לערי ריינהארט (talk) 12:45, 4 May 2014 (UTC)

@TomT0m I added some links to Category:User fr. Then I changed the language to french via the language selector at th topof the page. I expected to see the Reasonator in French. Can you please add a parameter WMFLCODE and change the link to the format

http://tools.wmflabs.org/reasonator/?lang={{WMFLCODE|TRICK}}&q=Q{{VALUE}}&live . This way my browser can manage same style bookmarks.

TRICK should be the "best" autodetect language value we can use today. Regards gangLeri לערי ריינהארט (talk) 22:31, 10 May 2014 (UTC)

@TomT0m any news on this? @PiRSquared17 do you have an idea how to implement an optional language parameter?
@לערי ריינהארט: Hello. :-) Could you please test User:PiRSquared17/Q' for this? Example: {{User:PiRSquared17/Q'|123}}->lang: en September (Q123) (view with reasonator)). It uses {{int:lang}} to get the user's language. If it is satisfactory, it can be added to Q'. You can also add a "lang" parameter, which consists of a language and an optional list of fallbacks, e.g. lang: fr,de,es no label (Q123) (view with reasonator)). That should be documented in the /doc subpage if it is added. πr2 (t • c) 16:51, 15 May 2014 (UTC)

@PiRSquared17 @TomT0m user:PiRSquared17/Q' has the features I needed. Here some validation examples:

I am not good with /doc issues. Please modify template:Q' and the relevant templates. Thanks in advance! gangLeri לערי ריינהארט (talk) 18:25, 15 May 2014 (UTC)

Thanks for helping. Do you think we should incorporate the fallback automatically with Template:GetFallback? Then we could have "bar" automatically generate "bar,de". Maybe this functionality would be better implemented in Reasonator itself. πr2 (t • c) 18:40, 15 May 2014 (UTC)
I support using "GetFallback"-functionality by default in all multilanguage sites, tools, templates etc. It is important to communicate this. People should know a central place where they can find similar topics. GetFallback should overwrite single language parameters. For language analysis task any combination should be available for users via the desired combination.
{{User:PiRSquared17/Q'|315|lang=el,fr}} generates: lang: el,fr no label (Q315) (view with reasonator))
Maybe the lang parameter value should be displayed as (Wikidata-Reasonator small logo.svg el,fr) .
gangLeri לערי ריינהארט (talk) 19:40, 15 May 2014 (UTC)
@PiRSquared17, לערי ריינהארט: I'm OK to this kind of changes, but I'd like the default to wikitext annotation for the labels to be english as most of the discussions are in english here. Hopefully the entity selector to insert in wikitext I prepare will be be ready soon. TomT0m (talk) 21:42, 15 May 2014 (UTC)
{{{lang|{{int:lang}}}}} defaults to the users language anyway. Links using other values should be labeled as (Wikidata-Reasonator small logo.svg lv,fi) for lang=lv,fi. Then the link can be changed if required manually into ru or whatever. לערי ריינהארט (talk) 05:21, 16 May 2014 (UTC)

victim of Stalinism

Are items like victim of Stalinism (Q16333842) (created by user:לערי ריינהארט) appropriate? They are essentially emulating the category Category:Victims of Stalinism (Q8285743), which if you read de:Kategorie:Opfer des Stalinismus is a very broad category. It is being used as e.g. Alexander Ilyich Yegorov (Q291310) instance of (P31) victim of Stalinism (Q16333842). John Vandenberg (talk) 06:05, 13 May 2014 (UTC)

Some of basic WD questions relate to a fundamental one: Is WD moving to an ontology? Can contributors emancipate from their local WP ideologisms? It is known that matter changes the geometry of space, same does ideology it perverts the perception of right and wrong. Another technical issue is if the set of axioms is complete, if we have the right (enough) priorities to integrate / map concepts from other interacting projects. WD does not include its maintenance statements into the set of properties. This reminds the paradox in the classical set theory. In the paste years I read words as "ultimate". If WD wants to be an ontology similar as a multiverse it needs to map concepts from other sites. Why librarians (DDC) have a sort order? Order is the first question asked by visitors. Is WD a place where things happen randomly and items can be found randomly?
We need to remind that truth and logical are not subjects to democracy or votes.
I have not read what WD is. There is a road map somewhere and there are some naive expectations. Such expectations can be found in fairy tales especially that success is the result of less or no work.
Regards gangLeri לערי ריינהארט (talk) 07:02, 13 May 2014 (UTC)
So much questions ! My Troll-level measuring tool just exploded. It has a while ago in fact. TomT0m (talk) 11:07, 13 May 2014 (UTC)
@John Vandenberg: - The statements you mention are problematic. There are rather undisputed facts like "Albert Einstein - {{p|31} - human", that can go without a source for the time being. But I don't think a highly political statement as this one should be allowed to exist without a source. A source adds, among other things, perspective. It is the old example of "terrorist" versus "freedom fighter" which are used by the opponents and supporters of the group. - The final wording usually assigned by the winner of the conflict. In Wikidata we have the chance to model both perspectives, which is much more compatible with modern science of conflicts: moving away from the blame-game and settling-the-score kind of history, to an understanding of how conflicts arise and how they are a part of human nature. In total we should try to keep Wikidata-statements about conflict neutral and I don't think we should be the moral authority of the past nor the present. So I think all these statements should be given a source within 24 hours or deleted. And the source has to refer to the person as a victim of Stalinism (Q16333842) and not just come close to that (e.g. victim of corrupt court, victim of his time, victim of "Great Purge"). -Tobias1984 (talk) 14:00, 13 May 2014 (UTC)
In my opinion we can't use "instance of" for this kind of description: victim of communism is not a characteristic of one person. This kind of information should by added in another way, for example in the description of the cause of the death. And as Tobias mentioned, this should be sourced. Snipre (talk) 14:24, 13 May 2014 (UTC)
I agree that we can't use instance of (P31) with victim of Stalinism (Q16333842) but items like this can be used with cause of death (P509) and killed by (P157). Filceolaire (talk) 20:52, 13 May 2014 (UTC)
Instead of cause of death (P509) it should be manner of death (P1196). But that doesn't work as long as victim of Stalinism (Q16333842) is subclass of (P279) of person. It seems like 2 different data models are colliding here. -Tobias1984 (talk) 21:53, 13 May 2014 (UTC)
"Instead of cause of death (P509) it should be manner of death (P1196)."
Tobias, not quite. Per the description of manner of death (P1196) and its creation discussion, 'manner of death' is mapped to the term as used in a controlled vocabulary from standardized death certificates. I agree that "victim of Stalinization" is not an appropriate object for 'cause of death' claims, either.
Our existing mortality properties don't capture certain kinds of context well, but I don't think we should alter their meaning from standard definitions to accommodate it. A qualifier on these claims, or a new property, would be better to capture this kind of context.
For example, consider a hypothetical person named Mikhail, who died of starvation as part of the Nazino affair, which we'd like to capture on Wikidata. Here's a possible way to structure data about the subject:
instance of (P31): human (Q5)
place of death (P20): Alexandrovsky District, Tomsk Oblast (Q2031378)
date of death (P570): May 1933
cause of death (P509): starvation (Q853930)
part of (P361): Nazino affair (Q1850440)
manner of death (P1196): homicide (Q149086)
instance of (P31): mass deportation (Qy)
part of (P361): Stalinism (Q179121)
Things are less straightforward for the actual person in question, Alexander Ilyich Yegorov (Q291310). According to his English Wikipedia article, "Yegorov died in prison" -- sparse information. That leaves 'cause of death' and 'manner of death' unknown -- was the cause of death 'tuberculosis' and manner of death 'natural causes', or perhaps 'stabbing' and 'homicide'? One thing is for sure: 'prison' is not a cause or manner of death. Emw (talk) 01:05, 14 May 2014 (UTC)
@Emw: Thanks for clearing that up. At least for the imprisonment we could have property "imprisoned in" and use time qualifiers? -Tobias1984 (talk) 07:41, 14 May 2014 (UTC)
Good to see that we can easily develop our database structure but I think this kind of information is far beyond the current state of WD. In my opinion most of details can't be modeled in a good way in a database and this should be described in a text in WP articles. If we can provide for each human the death date and the death place we will reach a good level. WD shouldn't focus on details but in a minimal set of data available in all items according to their classes. I prefer to be able to find a death date in every person item instead of finding a lot of details of the death for one person and no data in 10 other items about persons. Snipre (talk) 08:53, 14 May 2014 (UTC)
@Snipre: It does not work that way, we're in a collaborative project where everybody is free to develop the details he wants details. And contribution power can not be substituted : someone who works on the detail on an item might not be the one who will do mass editions. TomT0m (talk)
@Emw: I would say as Stalinism is a political view, not an event. TomT0m (talk) 09:10, 14 May 2014 (UTC)
TomT0m, that's yet another unreadable post. Again, use the "Prévisualiser" feature, please! To your point: perhaps. Emw (talk) 12:54, 14 May 2014 (UTC)
Sorry again, but believe me, you'll go faster correcting yourself than trying to reeducate me :) I don't use the button, I use ctrl+alt+P, yet waiting for a usable discussion system like flow, this will still not make the feature usable in long posts.
TomT0m, asking others to correct your frequently-mangled posts is unreasonable. Your readers are not your janitors. Please just do you what everyone else does and read over what you write before or after posting, and fix flagrant errors. Emw (talk) 03:53, 15 May 2014 (UTC)
Tobias, yes. Alternatively, I suppose we could also use residence (P551) to describe imprisonment, in addition to the comfier residences that property tends to be used for. Emw (talk) 12:54, 14 May 2014 (UTC)
Along with Occupation=>Political Prisoner (unless they were a Prison guard). Filceolaire (talk) 19:33, 15 May 2014 (UTC)
Agreed, probably. If they're a prisoner of war, Wikidata:Property_proposal/Person#military_casualty_classification: 'prisoner of war' would also make sense. Emw (talk) 00:45, 16 May 2014 (UTC)

Items with same sitelink

Playing around with the new Wikidata dupe finder, I noticed that at least for the Thai wikipedia there are several items which both point to the same Wikipedia page. At least some of them I created myself using the Widar items creator, which apparently has a bug - for example it created Nongsuawittayakhom School (Q16307415) even though no label (Q13027327) already existed. As there are a bit too many to put them manually into the requests for deletion, and the merge widget does not work for these either, can any bot user maybe help to clean up. Ahoerstemeier (talk) 11:23, 15 May 2014 (UTC)

see WD:true duplicates. afaik they can't be found easily, since the site-links of one of the items isn't in the database. --Akkakk 20:35, 15 May 2014 (UTC)
A script is being run right now on the server to find them. Sorry it took so long to write that. --Lydia Pintscher (WMDE) (talk) 20:39, 15 May 2014 (UTC)
Script run has finished, I've added the items to WD:true duplicates. Cheers, Hoo man (talk) 12:50, 16 May 2014 (UTC)

Invalid coordinates

user:Multichill has known that his bot user:BotMultichill is importing coordinates that are not recognised by the software for months (see bugzilla:62105), but doesnt stop. Two weeks ago I raised this at User_talk:BotMultichill#odd_error, and three days ago I ask him to skip importing invalid coordinates. He has not provided useful responses to my queries, or those of others who are reporting other problems.

Has the development team indicated that they will change the software to support the data as imported by user:BotMultichill? The comments on bugzilla:64887#c3 indicate that the bots invalid data were not the intention of the software, but maybe the devs have changed their mind?

Also it would be good to have an estimate of how soon the bot will finish its current data load? If it is nearly completed a set of data, it might be better to let it finish. But if it isnt nearly finish, I think the bot should stop and restart only once the devs and bot operators have resolved the issue, and corrected the existing data if necessary. John Vandenberg (talk) 04:14, 16 May 2014 (UTC)

We will try to tackle it in the next sprint. --Lydia Pintscher (WMDE) (talk) 10:17, 16 May 2014 (UTC)
@Lydia Pintscher (WMDE):, will Wikibase support all the coordinate data that the bot has added? If it isn't going to be fixed soon, one way or the other, could the error message be changed (toned down), as it currently is very annoying for this to appear on so many items when conducting training or demonstrating the functionality to potential contributors. John Vandenberg (talk) 10:29, 16 May 2014 (UTC)
As far as I understand it they should all be fine after our fix. The issue is that the UI can't properly display them at the moment. The data is fine as far as I can tell. --Lydia Pintscher (WMDE) (talk) 10:33, 16 May 2014 (UTC)
At the hackathon in Zurich it was mentioned that a large amount of coordinates that are associated with Wikipedia articles is available for inclusion in Wikidata.. Is it best to wait until this fix is in ? Thanks, GerardM (talk) 10:41, 16 May 2014 (UTC)

Wikidata gadget on clients

There is a Wikidata gadget enabled on some Wikipedias which shows the item details under the page title, created by User:Yair rand. It would be good to copy this, and translate it, to other Wikipedia and sister projects to increase visibility of Wikidata for people who wish to enable it. It is currently on six Wikipedias under the name 'WikidataInfo'. See google hits[11]. It might be on other Wikipedia using a different name.

It would be very cool if this gadget, or another one like it, enabled viewing & editing of the label in other languages (as specified via {{Babel}} on the userpage, like we have here on Wikidata) or maybe just English, directly from the Wikipedia page. i.e. if I am on Japanese Wikipedia, I would like to see the English label, and if it is blank I should be able to click on it to open a text box and enter a value that is submitted to Wikidata. John Vandenberg (talk) 12:43, 16 May 2014 (UTC)

Note I have added a request for it to become a gadget on en.wp: en:Wikipedia:Gadget/proposals#Wikidata items again. John Vandenberg (talk) 12:57, 16 May 2014 (UTC)

Legendary or Mythical

Is there a difference between Legendary (See legend (Q44342) and Mythical or Mythological (see mythology (Q9134)?

English Wikipedia makes the distinction that a Legend legend (Q44342) is located in the context of history and the subject could be at least based on an actual human while myth mythology (Q9134) is usually out of time so a mythical character is a subset of fictional character but a legendary character might not be.

Based on this distinction we get this conclusion:

Myths
  • Book of Genesis
  • Book of Revelation
  • Creation stories
  • Stories of Gods
Legends
  • The Gospels
  • Aeneid
  • Siege of Troy
  • Stories of Heroes

Anyone else have a thought? Filceolaire (talk) 00:56, 7 May 2014 (UTC)

Not easy, some claim that the Norse mythology is based on a group of people who imigrated to Scandinavia around 2000 years ago. That something like what happend to Eric IX of Sweden (Q310152) after his death 1160 also happend around 1000 years earlier with a man named Odin (Q43610). Eric became a saint, while Odin became a god.
I think it will be very difficult to find simple general rules for fictional/mythical characters. I, for example, read StarTrek-novells. They are considered non-canonical and often contradict each other. When did Thomas Riker die for example? In the last books I read, he was still alive. So there exist some StarTrek-Universes where Riker is dead, and others where he is alive, the mirror-universes uncounted. -- Innocent bystander (talk) (The user previously known as Lavallen) 17:55, 7 May 2014 (UTC)
The good thing is that we can collect different classification methods and let each Wikipedia decide. For instace we could have a "preferred by:English Wikipedia" qualifier, so they can select their own classification. In a way similar to "imported from" but with a different meaning. Same applies to parallel universes, you can collect all statements and let the reader sort them out according to the POV they want to explore.--Micru (talk) 18:06, 7 May 2014 (UTC)
I would be strongly against this Micru. I think we can achieve some consistency. Filceolaire (talk) 07:21, 13 May 2014 (UTC)
Filceolaire: there are two forces at play here. One is the need to centralize achieving consistency, and the other one is the need to keep diversity. If you focus on the first without guaranteeing the second, our project will face strong opposition in sensitive issues. I don't know if this is one of them.--Micru (talk) 11:01, 17 May 2014 (UTC)

About quality / duplicate items / creation of many more items

In an RFC about quality, the key point in the text is that in order to allow Wikidata to serve data to info-boxes, every article has to have an associated item.

The problem is that we do not agree on the need for having items for every article. Obviously, the creation of an item for every article means that many items will have unrecognised inter-language links. Arguably this is the case already but with a massive influx of new items this will be even more of an issue.

The situation we face is that we cannot have both. It is impossible to add items that will contain data that is harvested from articles when you first have to ensure that a subject has not been covered by another item.

As more data is available for an item, it becomes possible to identify items that are likely about the same subject. A human with the same name date of birth and date of death is likely one human. It becomes even more likely when other statements coincide as well. The point is that items with many statements have a high level of quality; they can serve as providers of information to a project and by the same token, they make it easier to identify duplicates.

An item with *four or more information items* is worthwhile to have. It provides basic usable information. Duplicate names can for instance be found in tools like Reasonator, it also shows the year of birth, death and gender.

An information item is a link, a statement or a label.

Given that 50% of our items have none or one statement, it is imho essential to bring more statements to the existing items. For humans for instance the most basic information are properties like a date of birth and death, gender.

The question is how do we move towards quality AND towards having Wikidata as the source that serves data to WMF projects. Thanks, GerardM (talk) 09:35, 16 May 2014 (UTC)

Your problem is an identification one meaning the main task is the labeling of the item in a sufficient number of languages. Here the main work is to list items with only one sitelink and with label in only one language and to find contributors ready to translate labels in English and perhaps in other languages. Even if the birth date and the death date correspond I think nobody will link an item if he can't read/translate the label in English. Snipre (talk) 10:49, 16 May 2014 (UTC)
The same birth data / death date can be very useful to group items, just listing 50% of the items might not be realistic. And there might be really few people who can do a Chinese / english translation (outside of China, but I don't remember seeing a lot of Chinese here.) What would be cool is that we have transliterations in pinyin though, as it is often available in Wikipedias articles. TomT0m (talk) 11:14, 16 May 2014 (UTC)
Having a label in Dutch, German, English and French will likely get a name that is spelled exactly the same. They are the first target of finding people who are the same. We use a gadget to transliterate Russian among others, it could also help. It is unlikely that we get them all in this way but when we get many it already helps. To do this we need statements.. GerardM (talk) 13:52, 16 May 2014 (UTC)
There are quite a lot of biographic articles in Wikipedia that have dates of birth/death. I think it would be most helpful to import these systematically into Wikidata. --- Jura 06:29, 17 May 2014 (UTC)

Getting the JobCode proposal for your home country ready (will take you 5 to 10 minutes maximum).

I've compiled a list of all existing JobCode systems in the world - from the UN - as part of the Occupations and professions task force. Those systems enable to identify an occupation precisely on a Job Offer, and get all kind of interesting statistics from national statistics organizations. I've created draft property proposals for each of them, and I need your help to finish filling them in. Based on my past experience, it should take you 5 to 10 minutes maximum. Here's what you need to do:

  1. Find your country (eg. Norway https://www.wikidata.org/wiki/Wikidata_talk:Occupations_and_professions_task_force#Norway-_STYRK-08)
  2. Click on the UN link (in Source:) to find the details about your national system
  3. Edit and fill the draft property proposal based on the UN page.
  4. Either submit the proposal yourself or tell me it's done so that I can do it.

--Teolemon (talk) 10:34, 17 May 2014 (UTC)

Teolemon, instead of creating so many properties, wouldn't be better to have a single generic property "job code" and use a qualifier "type of job code"? It would save everyone countless hours... I'm saying that also because some of the available job code properties are underused, so I see little point in creating new ones if the old ones are not alive.--Micru (talk) 10:54, 17 May 2014 (UTC)
Yes, that might be a better solution, we loose the validation rules, but it'd be more scalable. --Teolemon (talk) 13:38, 17 May 2014 (UTC)
This would work in a similar way to station code (P296). A generic qualifier property was proposed to use with this kind of property (Wikidata:Property_proposal/Archive/20#specifically) but not approved though catalog (P972) does exist. Filceolaire (talk) 20:45, 17 May 2014 (UTC)

Wikiquote Phase 2

Hey everyone,

Lydia is focusing on some Outreach tasks at the moment so I have volunteered to make this announcement. The development team are currently planning to enable Phase 2 for all language editions of Wikiquote on June 10th. For those who don't know, Phase 2 is enabling data access from Wikiquote to Wikidata and vice versa. Lydia also wants to thank all users who helped make the Phase 1 deployment a successful launch on all Wikiquote languages. Post any questions/comments either here or on the coordination page at Wikidata:Wikiquote. Thanks, John F. Lewis (talk) 15:57, 17 May 2014 (UTC)

Wikidata:Wikiquote should have been updated first... --Nemo 17:12, 17 May 2014 (UTC)
I would like to know what data, specifically, is proposed to be used in Wikiquote articles. I note that the linked Wikidata:Wikiquote page "explicitly excludes the usage of Wikidata to store the actual content of Wikiquote articles." [emphasis in original] I think it should be proposed and discussed in concrete terms at Wikiquote before any of this (non-content?) data is included in Wikiquote articles. ~ Ningauble (talk) 17:30, 17 May 2014 (UTC)
@Ningauble: It is up to Wikiquote to decide what data they want to include. If it was up to me I would suggest adding an infobox to wikiquote pages with basic data from wikidata but it isn't up to me - this discussion will have to happen on Wikiquote. Filceolaire (talk) 20:49, 17 May 2014 (UTC)

How to (properly) capture uncertainty?

I ran into two issues of uncertainty which I do not see a way for being able to capture those properly.
The first one was on Abraham von Freising (Q330885): According to the reference, the person may have died either on 7 June 993 or 7 June 994. This could be reflected by using a time range or a data type specific qualifier like "alternative date". But, actually, these are two discrete values, basically a list of dates. Eventually, I added both which, at first, seems reasonable and was done before. However, when querying for people having died in 993, one would receive Abraham von Freising (Q330885) without any hint that this information is not certain. Consequently, when querying for people having died in 993, one would assume that this person, in fact, died in 993 and uncertainty becomes fact.
Another example is Wolfgang Carl Briegel (Q1523127). According to one reference, the person may have been a student of Johann Erasmus Kindermann (Q466635). Qualified by the same time range, I added "unknown value" and Johann Erasmus Kindermann (Q466635) for student of (P1066). However, split into two separate statements, that does not really reflect what the reference expresses and applying both statements, backed by the same reference, seems even odder than backing different values for date of death (P570) with the same reference. Expressing that Johann Erasmus Kindermann (Q466635) may have taught Wolfgang Carl Briegel (Q1523127) using student (P802) on Johann Erasmus Kindermann (Q466635) seems kind of impossible without some weird qualifier expressing "may be false". One could argue to just drop that uncertain information and use "unknown value" exclusively, but, well, that would be a loss of information and I am sure such problems occur in other situations as well (an example of a more prominent topic may be to model something like "Roger Godberd (Q7358238) might have been Robin Hood (Q122634)").
Random knowledge donator (talk) 07:45, 15 May 2014 (UTC)

An important question. There are those who will state that this can be handled by assigning "preferred" / "normal" / "deprecated rank", but this feels more or less like a stopgap measure. - Brya (talk) 11:16, 15 May 2014 (UTC)
Thanks for your response. I agree with the ranks not being a proper solution since that would not be what they are intended for. Using ranks for uncertainty would overload their purpose. Random knowledge donator (talk) 10:49, 19 May 2014 (UTC)
You can also use statement disputed by (P1310) when there is a source opposing any of the statements, or you could propose another property to be used as qualifier, like "statement doubted by". I think it would make sense in this case.--Micru (talk) 11:07, 17 May 2014 (UTC)
Thanks for your response. The problem is that, in many cases, there is no other source which is opposing the statement - or one would need to specify the same source as opposing source via your suggested qualifier. Basically, a reference states that something is "likely" and/or lists a specific set of potential values. That is what would need to be reflected. If someone died either on 7 June 993 or 7 June 994 might very likely rest upon unreadable handwriting in a medieval document. Having a qualifier which expresses uncertainty might be a solution but it would need to be a qualifier without any kind of value just saying "is uncertain" (acting more like some kind of flag) since that is what the reference expresses - regardless of any other reference challenging the value. Random knowledge donator (talk) 10:49, 19 May 2014 (UTC)

My hope was to receive some more input on that subject. I wonder whether it is not really of interest or nobody really knows how to handle it. In the end, it might also be a technical limitation no one really thought about in the first place - maybe there should by another snak type "uncertain value" as opposed to "custom value", "no value" and "unknown value"? Or it simply is not considered Wikidata's obligation to capture uncertainty (which would be a pity from my point of view)... Random knowledge donator (talk) 10:49, 19 May 2014 (UTC)

two articles in Wikipedia - how to handle?

Item no label (Q5960203) is staying alone till the article in es:wp get merged with tavern (Q154250). How to handle this until the merge? Thank you, Conny (talk) 19:10, 18 May 2014 (UTC).

We can just wait when (and if) they do the merge. --Stryn (talk) 20:01, 18 May 2014 (UTC)

Q5235809

There's something wrong with the interlanguage links of Q5235809. In English, it's a biographical article; in French, it appears to be a how-to page of some sort. --Redrose64 (talk; at English Wikipedia) 14:19, 19 May 2014 (UTC)

French link removed now. It does not need an item on Wikidata. --Stryn (talk) 14:25, 19 May 2014 (UTC)

Template:Q

Hi,

This template exists on the English Wikipedia. Could we already import it in Commons or should we wait Phase 2 ? Pyb (talk) 16:29, 18 May 2014 (UTC)

We should wait phase 2. Phase 1 is only interlanguage links in the sidebar. Tpt (talk) 17:29, 18 May 2014 (UTC)
thx Pyb (talk) 18:38, 18 May 2014 (UTC)
It is useful to create and use a 'Q' template, carefully, especially in talk page discussions. It only injects label in local language into the cached page, which is usually going to remain stable 'enough' for discussion purposes. And it is re-rendered each time the page is changed. Also 'Q' may already be used by a different template, as it is very short, so it is useful to start talking with the local community about changing 'Q' to be the wikidata template. Iirc 'P' is currently used on en.wp. :( John Vandenberg (talk) 06:34, 20 May 2014 (UTC)

Property Suggester Evaluation

The PropertySuggester is still in review, but we would like to collect some feedback on the quality of the suggestions for our bachelor thesis. It would be great if you could take a few minutes to evaluate some suggestions on this site: http://suggester.wmflabs.org/wiki/Special:PropertySuggester

Thank you! – The preceding unsigned comment was added by ValueError (talk • contribs) at 09:40, 20 May 2014 (UTC).

Please sign your posts here, to make it possible for our bots to put each thread in the archive. -- Innocent bystander (talk) (The user previously known as Lavallen) 14:47, 20 May 2014 (UTC)

Birth dates and death dates from Wikipedia

Hello all, per request of Gerard I wrote a code to get BoD and DoD of English Wikipedia and Wikidata (by dumps) and compares them. the bot reports on differences or lack of data in Wikidata.

It's working now and by 5 AM UTC it'll be finished reporting. I like to move the reports to Wikidata project, Do you agree? Writing this code wasn't easy because of harvesting DoD and DoB from English Wikipedia isn't easy. BestAmir (talk) 22:38, 19 May 2014 (UTC)

The interesting thing is Jimbo's birthday isn't the same. Ask him which one is correct :D Amir (talk) 04:32, 20 May 2014 (UTC)
The major take away is that conflicts are as important as new data. We should typically report on conflicts and do nothing. We can report on missing statements in Wikidata, at this stage importing everything new makes sense to me. GerardM (talk) 06:14, 20 May 2014 (UTC)
Tried to look at some Bdr/NW, but have not found any problem there (date of birth was already on wikidata for all checked items). Maybe some error in code or old dumps. --Jklamo (talk) 12:00, 20 May 2014 (UTC)
I will check them, I think probably it's a bug in the dump reader of wikidata. Amir (talk) 13:46, 20 May 2014 (UTC)
It'll be good idea to check which birth/death dates should be converted to Julian calendar by comparing Russian Wikipedia and Wikidata. As well as correctness of such dates (I found several mistakes in Russian Wikipedia). --EugeneZelenko (talk) 14:15, 20 May 2014 (UTC)

@Jklamo: I made the most horrible mistake possible, I checked for Wikipedia instead of Wikipedia, I'm starting over @EugeneZelenko: Converting Julian to Gregorian and vice versa is pretty much easier than getting the data from Russian Wikipedia but I can't see the best algorithm for this checkAmir (talk) 06:54, 21 May 2014 (UTC)

Infoboxes in Russian Wikipedia contains dates on both calendar (when it's appropriate). If I'm not mistaken, first is Gregorian. Converting could not be done automatically, because Julian calendar dates make sense only to persons from countries where it was used. I could provide help with templates/parameters names. --EugeneZelenko (talk) 14:03, 21 May 2014 (UTC)

Just pointing out, that there is some code for BoD and DoD harvesting from wiki in User:Magnus Manske/wikidata useful.js. --Jklamo (talk) 14:20, 21 May 2014 (UTC)

Items without links to articles

Hi all, just a pair of quick questions for confirmation:

Are data items without any links kept or deleted? In my experience over the last two years, they get deleted. In this mail, it is claimed that there is 'official sanction' from Lydia of WMDE that they won't get deleted. But again, as per my experience, particularly, on IRC back in December 2012, a WMDE employee had stated that it is the community and not WMDE that makes such decisions.

Please confirm on the two things: 1. What happens to empty items? 2. Who makes the decisions?

Thanks, --Rsrikanth05 (talk) 10:02, 20 May 2014 (UTC)

An empty item is different from an item without sitelink. See WD:N for a view of the criteria. In short : there may be data statements about the item, it might be notable if it is clearly refering to an identifiable entity in the real world, and an absence of sitelink is nowhere sufficient to delete it. TomT0m (talk) 10:29, 20 May 2014 (UTC)
I read the notability criteria. It says: 'It contains at least one valid sitelink'. That sort of contradicts the statement that lack of sitelinks won't get an item deleted. --Rsrikanth05 (talk) 10:35, 20 May 2014 (UTC)
Reread then. This is only one of the alternatives in a It must satisfy at least one of the following criterion. It's sufficient, but not necessary. TomT0m (talk) 11:45, 20 May 2014 (UTC)
That is no official sanctioning... I just stated what the community decided already anyway. --Lydia Pintscher (WMDE) (talk) 11:59, 20 May 2014 (UTC)
In practice, administrators at RfD decide in each particular case.--Ymblanter (talk) 15:33, 20 May 2014 (UTC)
Rsrikanth05 If any other items have statements which link to the item with no sitelinks then you should assume that it should be kept under criterion 3 of WD:N. Use 'What links here?' to check. Wait a while after an item is deleted (I would say a day) before proposing deletion, just in case the creator hasn't finished creating these statements. Filceolaire (talk) 22:12, 20 May 2014 (UTC)
Thank you Filceolaire. Also, for the record, I'm not asking this to delete an item, but rather to safeguard against deletion. --Rsrikanth05 (talk) 05:59, 21 May 2014 (UTC)

Chemical and phisical properties

1) Why are so many properties proposals in pending state? What "Approved for creation; waiting for Number datatype" means?

2) It wouldn't be better to make subcategories of All Properties like natural properties -> chemical properties. Because this template it's not enough.

Eloy (talk) 10:37, 20 May 2014 (UTC)
Because the datatype required for these properties has not yet been lauched. /ℇsquilo 13:54, 20 May 2014 (UTC)
When will it be launched? --Jakob (talk) 15:13, 21 May 2014 (UTC)
That was answered during the office hour two days ago (see: full log)
[18:30] <JohnLewis> Lydia_WMDE: On that note; any updates with mono-lingual ior quantity + units datatypes or any more on the stove?
[18:30] <Lydia_WMDE> mono-lingual is being worked on right now
[18:31] <Lydia_WMDE> quantities with units not yet
[18:31] <Lydia_WMDE> i am hoping to find a student or two to work on that
[18:31] <Lydia_WMDE> if anyone is interested please let me know
--Micru (talk) 15:26, 21 May 2014 (UTC)
Expanding on the above; some datatypes are planned however have yet to actually be developed, similar to Wikidata's Phase 3. Wikidata is an in-active development project unlike most other projects which are built purely on MediaWiki or some other completed extension(s). John F. Lewis (talk) 18:43, 21 May 2014 (UTC)

Authority control template for editors' user pages

Please could somebody import the {{Authority control}} template from English Wikipedia, (or Wikimedia Commons, or another variant) so that editors may add en:ORCID and other identifiers, to their user pages? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:25, 16 May 2014 (UTC)

@Pigsonthewing: ✓ Done The Anonymouse [talk] 15:13, 19 May 2014 (UTC)
Thank you, The Anonymouse! Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 19:33, 21 May 2014 (UTC)

Bot with admin rights

Just to inform you: Ladsgroup got unrestricted admin rights for Dexbot. --Succu (talk) 21:15, 19 May 2014 (UTC)

+ all of actions are public and people can check them and be sure my bot is doing just approved tasksAmir (talk) 21:18, 19 May 2014 (UTC)
This is a bad bad choice many item are merged wrong, many item are empty because the article on wiki is moved to new name, all description and label gone lost i hope the community reconsider that choice --Rippitippi (talk) 03:16, 21 May 2014 (UTC)
Ok... please show us some specific examples of this happening then. Ajraddatz (talk) 03:18, 21 May 2014 (UTC)::::
The first item I have open no label (Q5769812) merged to Charles Green (Q5078401) but label not, label losted. Who work in deletion (checking the items..) see hundreds of item to fix with disambiguation page wrong merged thousends... --Rippitippi (talk) 03:27, 21 May 2014 (UTC)
Another example no label (Q3353454) is merged with 2012 China Open (Q1188125) but the merge is wrong because the fr voice is on only male tournament but the bot in the future delete the item because empty ecc.... if we want reach the 100000000000 item number this is the right way, if we want the correctness of the information this not the right way --Rippitippi (talk) 04:01, 21 May 2014 (UTC)
Both of the examples that you linked to were not issues. For the first, the enwiki page for Charles Green the balloonist refers to the same person as the page on svwiki (notice the birth dates). So that one was fine to merge. The second was also good, with the frwiki page referring to the China Open tournament, and the enwiki page on the item it was merged into being on the same topic. Look at who won; it's the same names. Please explain to me how the merges were incorrect in light of that. Ajraddatz (talk) 04:34, 21 May 2014 (UTC)
Hmm, I see the issue with the second (though perhaps frwiki should make their two articles into one to be like the rest of the projects :P) Ajraddatz (talk) 04:37, 21 May 2014 (UTC)
I corrected now the mistakes on the second example. Who want to do some more work? --Stryn (talk) 07:33, 21 May 2014 (UTC)
This is the point, if bot delete empty pages we can't see the problems, the empty pages are not a problem the errors are, so plesase consider to deflag bot and stream delete too--Rippitippi (talk) 15:31, 21 May 2014 (UTC)(Ajraddatz tennis tournaments is a minefields with apt, mens,womans both etc...)
The checks that my bot makes in order to prevent mistakes are:
  1. The item can't have any links
  2. The item can't have any statements
  3. The item can't have any backlinks (except from user and user talk namespaces)
  4. The bot goes through the history of item, In whole history if there is more than one site link, the bot skips
  5. The only linked sitelink has to be in at one of these situations: Added to another item (that means the item has been merged) or has been deleted.

The checks are so tight that makes bugs impossible to happen (frankly I think these checks are too tight), but if someone merges two items that needs to be remained separated, do you think admins checks them as carefully as you checked these examples? if you give this list to any admin they would delete them all (no admins checks contents of all of sitelinks) admins have deleted incorrect merges too, when someone realizes that a merge is incorrect the solution is simple, split them up into two items. Amir (talk) 16:58, 22 May 2014 (UTC)

Strange behaviour when trying to remove a statement

Q13717403. If I try to remove BbF identifier for good (one of those silly entries where a bot created a totally wrong VIAF entry, and then another silly bot took the corresponding BnF entry from VIAF), the statement won't be deleted, but I get a "malformed input" message. The input mentioned has nothing to do with deleting, though. In the meantime, I changed the value to "0" (better than the wrong entry). Here's a screenshot from when I try to remove the statement: http://www.bilder-hochladen.tv/pic/NQLmzOZr/ --AndreasPraefcke (talk) 11:17, 22 May 2014 (UTC)

I faced the same problem when I tried to remove the BnF entry. After this edit, where I changed the nl label (which was the only one to write "J.C." instead of "J. C.", with the error message also mentioning "J.C."), deleting the statement worked. --YMS (talk) 12:40, 22 May 2014 (UTC)
I think it's new bug. I had similar problems deleting a property. I was forced to remove the description in another language. Then I could delete the property. @Lydia Pintscher (WMDE):? --Succu (talk) 12:45, 22 May 2014 (UTC)
It is a bug indeed. It should be fixed in the next deployment on 27th. Sorry folks. --Lydia Pintscher (WMDE) (talk) 13:27, 22 May 2014 (UTC)
Thanks to all. --AndreasPraefcke (talk) 17:30, 22 May 2014 (UTC)

Authority control template for editors' user pages

Please could somebody import the {{Authority control}} template from English Wikipedia, (or Wikimedia Commons, or another variant) so that editors may add en:ORCID and other identifiers, to their user pages? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:25, 16 May 2014 (UTC)

@Pigsonthewing: ✓ Done The Anonymouse [talk] 15:13, 19 May 2014 (UTC)
Thank you, The Anonymouse! Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 19:33, 21 May 2014 (UTC)

Bot with admin rights

Just to inform you: Ladsgroup got unrestricted admin rights for Dexbot. --Succu (talk) 21:15, 19 May 2014 (UTC)

+ all of actions are public and people can check them and be sure my bot is doing just approved tasksAmir (talk) 21:18, 19 May 2014 (UTC)
This is a bad bad choice many item are merged wrong, many item are empty because the article on wiki is moved to new name, all description and label gone lost i hope the community reconsider that choice --Rippitippi (talk) 03:16, 21 May 2014 (UTC)
Ok... please show us some specific examples of this happening then. Ajraddatz (talk) 03:18, 21 May 2014 (UTC)::::
The first item I have open no label (Q5769812) merged to Charles Green (Q5078401) but label not, label losted. Who work in deletion (checking the items..) see hundreds of item to fix with disambiguation page wrong merged thousends... --Rippitippi (talk) 03:27, 21 May 2014 (UTC)
Another example no label (Q3353454) is merged with 2012 China Open (Q1188125) but the merge is wrong because the fr voice is on only male tournament but the bot in the future delete the item because empty ecc.... if we want reach the 100000000000 item number this is the right way, if we want the correctness of the information this not the right way --Rippitippi (talk) 04:01, 21 May 2014 (UTC)
Both of the examples that you linked to were not issues. For the first, the enwiki page for Charles Green the balloonist refers to the same person as the page on svwiki (notice the birth dates). So that one was fine to merge. The second was also good, with the frwiki page referring to the China Open tournament, and the enwiki page on the item it was merged into being on the same topic. Look at who won; it's the same names. Please explain to me how the merges were incorrect in light of that. Ajraddatz (talk) 04:34, 21 May 2014 (UTC)
Hmm, I see the issue with the second (though perhaps frwiki should make their two articles into one to be like the rest of the projects :P) Ajraddatz (talk) 04:37, 21 May 2014 (UTC)
I corrected now the mistakes on the second example. Who want to do some more work? --Stryn (talk) 07:33, 21 May 2014 (UTC)
This is the point, if bot delete empty pages we can't see the problems, the empty pages are not a problem the errors are, so plesase consider to deflag bot and stream delete too--Rippitippi (talk) 15:31, 21 May 2014 (UTC)(Ajraddatz tennis tournaments is a minefields with apt, mens,womans both etc...)
The checks that my bot makes in order to prevent mistakes are:
  1. The item can't have any links
  2. The item can't have any statements
  3. The item can't have any backlinks (except from user and user talk namespaces)
  4. The bot goes through the history of item, In whole history if there is more than one site link, the bot skips
  5. The only linked sitelink has to be in at one of these situations: Added to another item (that means the item has been merged) or has been deleted.

The checks are so tight that makes bugs impossible to happen (frankly I think these checks are too tight), but if someone merges two items that needs to be remained separated, do you think admins checks them as carefully as you checked these examples? if you give this list to any admin they would delete them all (no admins checks contents of all of sitelinks) admins have deleted incorrect merges too, when someone realizes that a merge is incorrect the solution is simple, split them up into two items. Amir (talk) 16:58, 22 May 2014 (UTC)

Strange behaviour when trying to remove a statement

Q13717403. If I try to remove BbF identifier for good (one of those silly entries where a bot created a totally wrong VIAF entry, and then another silly bot took the corresponding BnF entry from VIAF), the statement won't be deleted, but I get a "malformed input" message. The input mentioned has nothing to do with deleting, though. In the meantime, I changed the value to "0" (better than the wrong entry). Here's a screenshot from when I try to remove the statement: http://www.bilder-hochladen.tv/pic/NQLmzOZr/ --AndreasPraefcke (talk) 11:17, 22 May 2014 (UTC)

I faced the same problem when I tried to remove the BnF entry. After this edit, where I changed the nl label (which was the only one to write "J.C." instead of "J. C.", with the error message also mentioning "J.C."), deleting the statement worked. --YMS (talk) 12:40, 22 May 2014 (UTC)
I think it's new bug. I had similar problems deleting a property. I was forced to remove the description in another language. Then I could delete the property. @Lydia Pintscher (WMDE):? --Succu (talk) 12:45, 22 May 2014 (UTC)
It is a bug indeed. It should be fixed in the next deployment on 27th. Sorry folks. --Lydia Pintscher (WMDE) (talk) 13:27, 22 May 2014 (UTC)
Thanks to all. --AndreasPraefcke (talk) 17:30, 22 May 2014 (UTC)

Duplicates of species not being fixed

I see that user:Magnus Manske has not been fixing the damage he and his bot has caused. I reported to him last week that he had created ~2000 duplicates of taxons, and on the weekend user:GZWDer reported that his bot had also been creating duplicates (at least 500; probably many many more) with a 90% error rate.

To understand the impact of this, let me explain how I am using taxons in the real world. Wikimedia Indonesia is managing a WMF grant to an Indonesian photographer. They have been uploading all sorts of photographs, only some of which are 'part of' the grant. We need to monitor the grantees work, and report on it. To do this I wrote a script which finds the species in each photograph, and check whether it is an 'Indonesian' species. It uses Wikidata extensively to do this, and the WMID team (and the grantee) have been adding data to Wikidata as required so my tool can assess the species. The report was working flawlessly a month ago. Occasionally user:Dexbot or a human creates a duplicate, which my script reports and I would be happy to fix. However the script is now mostly errors. John Vandenberg (talk) 04:44, 22 May 2014 (UTC)

John, what do you call a duplicate and, is it in every respect a duplicate ? GerardM (talk) 05:17, 22 May 2014 (UTC)
Already before this last runs of the item creator were tons of duplicate taxons, thanks to the bot created articles on several Wikipedias which were not properly connected with each other and thus included separately here. Espcially most of Vietnamese Wikipedia articles need to be merged, not just taxons, also lots of small towns from all over the world. It seems that the current manual way to merge items does not scale to the big amount of merges still necessary, it gives editing conflict of the RfD page so often. Ahoerstemeier (talk) 07:15, 22 May 2014 (UTC)
A paper on Portuguese and English Wikipedia indicates that there is a vast number of missing inter language links. One aim of Wikidata is to be able to replace information in templates with information from Wikidata. This requires that each article has a link. To prevent problems with bot generated articles, MediaWiki should support articles generated on the fly based on information from Wikidata.
In the mean time we need tools that help identify either merges in Wikidata or articles that are on the same subject.
Sadly, Wikidata is NOT yet to be trusted for applications that are used in the real world. Assuming that you can, is accepting that it may prove you wrong. Thanks, GerardM (talk) 07:37, 22 May 2014 (UTC)
Well, I think it is unfortunate that so many taxon related items have been created on taxa having already an item on wikidata. To give you an example: I have been going through Carabidae taxa: most articles in the root category of en:Category:Carabidae are unlinked, while there is a corresponding article in nl:Categorie:Loopkever with wikidata item - we are talking about 1,755 in one category only. What is even more unfortunate, is that these newly created items do not contain a single property. Identifying the items by properties as taxa with a binomial/Latin name (P225) would make merge candidates more easily identifiable (and countable) by constraint violations. Taking more care when importing new items would be of considerable help. Kind regards, Lymantria (talk) 07:51, 22 May 2014 (UTC)
It is unfortunate indeed.. However, you can only add statements to items when they exist. Preventing new items is not an option. Thanks, GerardM (talk) 08:18, 22 May 2014 (UTC)
Preventing automated creation of duplicates is definitely an option that may need to be considered if the bots and widar-supercharged people that create duplicates do not actively minimise duplicate rate and fix their own errors. The admin corps here shouldnt be spending all of their time processing duplicates on WD:RFD because the bot operators are too lazy to do simple checks in their code or processes. With the current level of duplicates created by dumb bots and widar tools, the only effective way to reduce admin workload is to build automated tools with open source code to find duplicates and list them in batch mode so the admins can process them knowing that the tool has done all the necessary checks before merging them.
Preventing automated creation of items in some domains needs to be considered very soon, as there are some domains where Wikidata has almost every item in the domain, and we do not need a new duplicate every time another Wikipedia article is created. Otherwise Wikidata will never be able to be used. John Vandenberg (talk) 08:41, 22 May 2014 (UTC)
John, you indicated that you have a conflict of interest. In order to improve quality and in order to be able to serve information to info-boxes, your approach does not work / cannot work. Your unwillingness to talk makes for a great deal of imho unnecessary conflict. GerardM (talk) 09:34, 22 May 2014 (UTC)
We have identified nearly 1,8 millions scientific names (=taxon name (P225)). So how likely is it we haven't allready an item?! Simply search for the scientific name. In most cases you will find an item. Creating an empty item makes no sense. --Succu (talk) 10:18, 22 May 2014 (UTC)
Consider the approaches to the problem differently. What can we do to prevent the creation of new items. Consider approaches that identify the species while they are still articles. I do not dispute that many species may already have an item. Just having the same "taxon name" however does not imply that it is the same defintion of a species though. The issue at hand is that we should not prevent the creation of new items because they might be a duplicate. GerardM (talk) 10:47, 22 May 2014 (UTC)
Species with the same scientific name belong together. Interlanguage links are not capable to express a taxonomic opinion. --Succu (talk) 11:21, 22 May 2014 (UTC)
GerardM, what is this about a conflict of interest? I don't receive any benefit wrt this camera grant; actually worth noting that WMID staff have also done their part of administrating this grant as volunteers because WMF didnt provide funding for overheads. John Vandenberg (talk) 23:51, 22 May 2014 (UTC)
Your conflict of interest shows in that you hold that application higher than Wikidata quality. When invited to discuss Wikidata quality you decline. When you are in a conflict with others you refuse a conversation. When you are told that your behaviour is perceived as stalking, you argue that you are only doing this for the common good and that it is not personal.
I have shown you why I did not understand your actions and I blogged about it. I am QUITE happy to understand your issues and work towards solutions. HOWEVER, as you refuse to communicate, you put yourself on a collision course and the only reason that I can see is that you consider your external commitments of a higher value.
What I want is for us to talk and cooperate and appreciate each others point of view and work together on solutions that will help the project that is dear to both of us. Thanks, GerardM (talk) 13:15, 23 May 2014 (UTC)
Large data importations should be approved by a Wikiproject/task force which manage the data structure and do the evaluation of the data source. To many bots operators just import data from one WP without checking if a statement already exists and if the imported data is in conflict with the existing data. Snipre (talk) 11:36, 22 May 2014 (UTC)
What we need is a better approach. What you propose is having someone walk in front of a car with a red flag. What you do not address is that in order to support data in templates all articles need an item. We can harvest data from articles, we can serve that data and we can merge items later. The minimal requirement is an item. Thanks, GerardM (talk) 11:46, 22 May 2014 (UTC)
It is a bit annoying that you narrow down the discussion to preventing creation of articles or not. In a project with almost 15,000,000 items it would be unwise to just drop in new things and hope they get repaired more or less manually. You started a discussion about "quality is measurable", well IMHO the recent nobrainer imports measurably decrease the quality of wikidata, because of the many duplicates. I would not like to prevent creation, but I dislike that used as argument to not consider improvements and ask people who do mass imports to implement these. Kind regards, Lymantria (talk) 11:56, 22 May 2014 (UTC)
The process is simple:
  1. try to find a taxobox template
  2. read the scientific name of the lowest taxon rang
  3. ensure it's a binomen
  4. use WDQ to identify an existing item with the same scientific name
  5. a) add the interwikilink to this item or,
  6. b) create a new item, add taxon name (P225) and taxon rank (P105) = species (Q7432)
These steps ensure to minimize the creation of redundant items. --Succu (talk) 12:16, 22 May 2014 (UTC)
It would be anyway desirable if items could only be created if they included at least one statement. Especially at Special:NewItem, but also for Widar and Bots. -Tobias1984 (talk) 12:21, 22 May 2014 (UTC)
The process is not that simple given the numbers of involved. This should not be done one at a time. GerardM (talk) 12:23, 22 May 2014 (UTC)
Why not? ---Succu (talk) 12:25, 22 May 2014 (UTC)
The merging afterwards has to be done one at a time as well. Lymantria (talk) 12:36, 22 May 2014 (UTC)
It seems to me it is already too late to discuss how to better import articles, most of the articles have been imported here by now, so we have to identify and kill the duplicates here. I most cases this is already possible with the labels, but if a bot would set taxon name (P225) then it would be even easier to find the matches. But the biggest problem - with the current way to merge using RfD we won't be able to kill all the 100,000 duplicates, it took me hours to merge some of the Quercus (Q12004) species, and most of the time was wasted because of editing conflicts on RfD. Ahoerstemeier (talk) 13:03, 22 May 2014 (UTC)
I wish it were true that we had a link for each article in each project. We do not have that. What you describe in the merge process is unfortunate. It points very much in the direction of a need for better tools. I am very confident that there are at least a million duplicates in Wikidata. It could be as much as 4/5 million Thanks, GerardM (talk) 13:28, 22 May 2014 (UTC)
Yep. For the first time "Unique value" violations of taxon name (P225) droped today below 1000 items. --Succu (talk) 12:54, 22 May 2014 (UTC)
@succu When a Wikipedia made the unfortunate decision to include *.000 articles as bot articles, you want to look at those articles with a bot and decide that they are the same as existing items using an algorithm. This allows for the insertion of those articles in existing items. When this is not possible, it does not negate the need to allow for the provision of data to those articles from Wikidata. It follows very much that a "project" is not finished once an initial number of articles have been created and data is imported. Follow up activities include the management of such merge actions.
@Lymantria In the recently announced games, Magnus shows that we can merge items with external tools. When we spend effort in building tools that recognise articles as belonging to specific items, it follows that the *current* manual activities can be automated to a large extend as well.
@Tobias1984 I totally agree that items should include statements. I am on record that I am looking for 5 data items for each item. A data item is a link, label, statement or qualifier. It is however to be understood that when you create an item that it can be part of a workflow. For instance, we added items for en.wp people who died in any given year. The next step is to mark them as human and it is to be followed with the import of date of birth etc data. It takes time to do all that. Appreciation for such consideration would grow when people, particularly admins assume good faith and engage in conversation particularly when they are invited to do so. Thanks, GerardM (talk) 13:00, 22 May 2014 (UTC)
The already existing merge.js is a better tool, usually completes the merge (Widar does not always do that) and is easier to process for admins at the Requests for Deletion pages. Perhaps things can be automated, but it is less easy than adding them to the correct items in one time. Kind regards, Lymantria (talk) 13:39, 22 May 2014 (UTC)
I use the merge.js all the time. I do not understand why it is not available by default.. When you say that is easier to do them one at a time, you are right when you add them individually, it is not easier when you add a large number. Typically people lose interest. Thanks, GerardM (talk) 14:03, 22 May 2014 (UTC)
@GerardM. Data import needs preprocessing work and that what is missing now. Preprocessing work avoid hours of manual merging and other correction jobs. We have not enough contributors to expect a rapid correction and we have to code bots to correct bots. A pity. Snipre (talk) 13:44, 22 May 2014 (UTC)
The approach I proposed for importing when there is nothing and comparing when there is, is being tried out by Amir. It has the added bonus that those people keen on such things may help to find cotchas and suggest improvements. Magnus introduced a tool that helps find merge candidates.
We should stop bitching and look for things that make things easier for us. Articles about humans for instance that have names, dates of birth / death and a link make it easier to spot duplicates. Adding basic statements really helps. We need tools that make things easy but we cannot forget what Wikidata is intended for. Thanks, GerardM (talk) 14:13, 22 May 2014 (UTC)
I am with Snipre and Succu (and others) here. Bots should not be used to create a big dump of loose items here, with the expectance that the little people will clear up the mess. A bot should be used to reduce the amount of tedious work, not create more. If this is the best a bot owner can do, he should do nothing till he can do better. - Brya (talk) 16:58, 22 May 2014 (UTC)
BTW, @ John Vandenberg: do not use "taxons" unless you want to be taken for French, the plural of "taxon" is "taxa". - Brya (talk) 17:01, 22 May 2014 (UTC)

My bot has been brought up in here several times but because the discussion is so long I can't read it very carefully. My main question is the problem of duplicate that happened by my bot is mainly about taxa or the duplicates has very wide range of subjects? Amir (talk) 19:20, 22 May 2014 (UTC)

I just saw this giant thread. TLDR; can't do much in the next few days. Most of the duplicates would have been fixed by the merge game I wrote, but other professionally upset people like yourself made me deactivate it. If you know how many duplicates there are, you must know their Q numbers, which should make it reasonably quick to merge them. Otherwise, I may get around to run a specific query in a week or so to find them. I then could extend Toolscript to automatically merge them, but I'm sure someone would be upset about automated merging. Frankly, I'm getting sick of this. Nothing anyone does is without side effects. What I do brings large and significant improvements to Wikidata, and I'm quite happy to improve things where I have the time and ability. But this constant bitching about issues that are, in the worst case, minor annoyances starts to grind down even on my calm. --Magnus Manske (talk) 21:27, 22 May 2014 (UTC)

@Magnus Manske: Happy merging... --Succu (talk)
I think we should not blame Magnus for these duplicates. The tools of Magnus are well written, are easy to understand and they improve Wikidata. The problem are the users who are not aware of the fact that after the creation of a new item a second user has to add statements, a third user has to merge the items and a fourth user has to confirm the merge respective to delete the empty one. --Pasleim (talk) 22:57, 22 May 2014 (UTC)
The dups reported in this thread were created by Magnus and his bot. John Vandenberg (talk) 01:00, 23 May 2014 (UTC)
@Pasleim. It is not a question of blame it is a question to define if bots importations/works are a help for Wikidata. The proposition is to put more restrictions on bot importations/works because their side effects are becoming important. If 90% of a bot work have to be corrected I think we have a problem. Snipre (talk) 08:49, 23 May 2014 (UTC)
90%? I'm pretty sure these errors ain't more than 5% but the scale is so big that 5% seems so big to you (see edits of my bot as an example) Amir (talk) 18:35, 23 May 2014 (UTC)
More than three quarters of these items were unnecessary created. The existing items could have been easily detected by the process I described above. --Succu (talk) 18:58, 23 May 2014 (UTC)

Navigation tree?

One of the features I miss the most on item pages is the lack of a navigation template. I think the strongest point of Wikidata is the ability to create knowledge structures, but at the same time there is no way to get a sense of the whole picture, other than resorting to external tools like WD tree. And the right part of the screen is so empty... However, each item has a different contextual needs (some need sorting by taxon, other by "subclass of", other by "part of", other need a deeper query...), so no generic solution is possible. I also think that creating this kind of knowledge maps is something that could be reused in Wikipedias to structure better their articles and find missing gaps. Any thoughts about this?--Micru (talk) 07:46, 23 May 2014 (UTC)

Reasonator or {{Item documentation}} (and {{Class documentation}}) do not have these constraints. Wait for Simple Queries and we will be able to build these navigation templates without a lot of problems. Reasonator is also hackable and Magnuses tools already have this ability. I don't think that Wikidata will change its model agnostic paradigm because it's in the roots of the project. At best instance of (P31) and subclass of (P279) will get special treatment, this was in the original data model, but it's not in the roadmap if I remember well. I don't remember, was you opposed to pushing Reasonator ? ;) TomT0m (talk) 09:41, 23 May 2014 (UTC)
TomT0m: {{Item documentation}} comes close to what I envisioned, but they are so hidden in the talk pages that I didn't even knew about their existence... Lydia, any chance to have a free field for templates? Or would be too much to have something like this on the side panel?--Micru (talk) 10:52, 23 May 2014 (UTC)
It's not completely out of the picture. It should be started as a gadget. When developing it please keep in mind that the current plan is to move identifiers and sitelinks into a sidebar similar to how Reasonator does it. (Not 100% decided yet. Mockups still have to be finished and posted.) --Lydia Pintscher (WMDE) (talk) 11:36, 23 May 2014 (UTC)
General info, a "downstream" tree is already present on wmflabs, see for example engine subclasses. LaddΩ chat ;) 14:08, 23 May 2014 (UTC)
Lydia is there any chance that the UI can be changed to show statements which link to the current item? That would (in my opinion) provide enormously valuable additional data about an item - or make that data accessible. We could also seriously start looking at deleting some reverse properties which would be redundant, especially if these statements were available for use in sitelinked wikipedia pages. Is that even possible? Filceolaire (talk) 17:18, 23 May 2014 (UTC)
Filceolaire I also like that idea, but I would put it in a collapsed section at the bottom (as in Reasonator), because sometimes they can be too many uses. The navigation tree makes visitors inmediately aware of the context of the item in the structure, unlike now that there can be parts, subitems, etc, but it is impossible to notice. If it starts as gadget, it would need some experimentation, and also some configuration options because for some items of the tree that LaddΩ posted, only a branch is relevant, be it upwards, or downwards.--Micru (talk) 17:44, 23 May 2014 (UTC)

Preview gadget

  1. 1) The preview gadget is broken since jquery got upgraded to version 1.9. (.after is no longer supported, the way it was used in the gadget). I propose a fix for this MediaWiki talk:Gadget-Preview.js#Proposed fix for broken gadget.
  2. 2) I have been using a modified version of the gadget for a while, that by default shows a default preview (guessed, based on my user language, shows preview from the associated wikipedia). Would folks like that as part of the official gadget? MediaWiki_talk:Gadget-Preview.js#Default_display

Aude (talk) 20:08, 23 May 2014 (UTC)

Media Viewer


Greetings, my apologies for writing in English.

I wanted to let you know that Media Viewer will be released to this wiki in the coming weeks. Media Viewer allows readers of Wikimedia projects to have an enhanced view of files without having to visit the file page, but with more detail than a thumbnail. You can try Media Viewer out now by turning it on in your Beta Features. If you do not enjoy Media Viewer or if it interferes with your work after it is turned on you will be able to disable Media Viewer as well in your preferences. I invite you to share what you think about Media Viewer and how it can be made better in the future.

Thank you for your time. - Keegan (WMF) 21:29, 23 May 2014 (UTC)

--This message was sent using MassMessage. Was there an error? Report it!


Species in one wiki, genus in another

Apologies if there's a better place for asking these sorts of questions.

Playing the Wikidata game, I was presented with it:Delapparentia turolensis and en:Delapparentia. Delapparentia is a dinosaur genus with only one known species, Delapparentia turolensis. Both articles cover both the genus and the species (there is no genus article in :it and no species article in :en). The game asks if they are the same topic. In a sense, they are (since both articles cover both topics), but in another sense they are not (since one title is the genus and the other is the species).

It certainly seems like we should link the two articles together. But in the event a second species is discovered, we would have to break the link. That seems weird to me. Is there a standard for handling these situations?

-- LtPowers (talk) 14:17, 22 May 2014 (UTC)

Genus and species have to be separate items on Wikidata. The genus should be linked from the species by parent taxon (P171), in case of a monotypic genus there can be the backlink towards the species using taxonomic type (P427). Whether the Wikipedia articles should be added to the genus item or the species item depends on what is the focus of the article, usually its the species and the genus will be without sitelinks. Ahoerstemeier (talk) 14:25, 22 May 2014 (UTC)
Thanks for the info. It seems that :it is the only Wikipedia with its article titled as the species; :en, :cs, :es, :nl, :or, :pl, :ru, and :zh have articles titled as the genus. Would the solution, then, be to move the en,cs,es,nl,or,pl,ru,zh links from Q1706554 to Q3704931, and leave the former empty of Wikipedia links? LtPowers (talk) 14:48, 22 May 2014 (UTC)
Since the articles deal with the genus and all of it's (only one) species I would think it is more appropriate to link all the articles to the item for the genus. Filceolaire (talk) 05:55, 23 May 2014 (UTC)
Mostly it is a matter of taste. For algae, fungi and plants there is strong emphasis to regard the species as the basic unit, but the people who deal with dinosaurs do everything upside down. BTW, the number of species assigned to a genus is a taxonomic opinion, and may well vary from one taxonomist to another. - Brya (talk) 16:28, 23 May 2014 (UTC)
Okay, I took the easy way out and linked them all to the genus entry. LtPowers (talk) 16:19, 24 May 2014 (UTC)

Requested merge

Q3051580 (Elminster : La Jeunesse d'un mage) in French and Q3723580 (Elminster - La nascita di un mago) in Italian are the same - could someone please merge the Italian page into the French one? 129.33.19.254 19:40, 22 May 2014 (UTC)

✓ Done --ValterVB (talk) 19:45, 22 May 2014 (UTC)
ValterVB These are not the same. They are translations of the same work but they have different values for translator (P655) so we need 3 different items - an item for the original work in English and items for the French and Italian translations. Can you undo this merge please. Filceolaire (talk) 05:50, 23 May 2014 (UTC)
@Filceolaire: Yes I can do it, but I want ask to italian literature project how they want manage this situation, because the italian page of the book isn't about a italian version. --ValterVB (talk) 18:43, 23 May 2014 (UTC)
ValterVB, the only edition detailed on the it wp article is the Italian translation. See it:Elminster_-_La_nascita_di_un_mago#Edizioni so that edition is certainly discussed there, though it is true that the infobox describes the English original. However it:wp decide to organise their articles we still need 3 different wikidata items because each item has statements you can make about it that are not true about the other items. Filceolaire (talk) 09:18, 24 May 2014 (UTC)
✓ Done --ValterVB (talk) 09:47, 24 May 2014 (UTC)
This is stupid. No Wikipedia will have separate articles for different language editions, even for world-famous books. The items should be merged--151.15.147.208 11:38, 24 May 2014 (UTC)
This is not Wikipedia. GerardM (talk) 12:12, 24 May 2014 (UTC)
ValterVB, Filceolaire, GerardM: I have corrected the item(s), please take a look. Just a clarification: even if wikipedia articles are in different languages, they refer to the same work (Elminster – The Making of a Mage (Q3051580)), so they should put together in the same work item. Usually there are no Wikipedia articles about editions, but it is useful to create items about editions to store bibliographic information. See the modified item linking to its 3 editions/translations (English, French, Italian). For a more detailed information check Wikidata:Books task force.--Micru (talk) 12:50, 24 May 2014 (UTC)

Wikidata Query: How to get a list of people born in any city from a given Country?

Hi!

@Magnus Manske:: Is it possible to extend this query to get people born anywhere in Brazil (instead of in a specific city)? Helder.wiki 22:57, 23 May 2014 (UTC)

@Helder.wiki:: I tried one and added it to Wikidata:AutoLists:
Somehow I tried it the other way round though (using place of death):
Currently there are 10 people who died in Brazil since 1.1.2014. There are 40 Brazilians who died, but most don't have a place of death defined.
Maybe it works how it should. --- Jura 10:19, 24 May 2014 (UTC)
How about this: this query?? It shows 41 people who have the Brazilian nationality. Thanks, GerardM (talk) 12:26, 24 May 2014 (UTC)
@Jura1: I can't see what you are trying to do with country (P17) and contains administrative territorial entity (P150), but it works better without them (17 results)--Zolo (talk) 12:43, 24 May 2014 (UTC)
@Zolo: I took it from http://wdq.wmflabs.org/api_documentation.html --- Jura 12:49, 24 May 2014 (UTC)
Oh I see sorry. Now I get 17 results for your query as well (p17 and p150 are just security checks here, and apparently our data are so good that they do not affect the results :). --Zolo (talk) 12:58, 24 May 2014 (UTC)

Lochs

According to Wikipedia, Loch is just "the Irish and Scottish Gaelic word for a lake and a sea inlet". So the word seems just less precise than "lake". So can I remove "type of lake: loch" [12] and replace it with "instance of Lake" ? --Zolo (talk) 13:13, 24 May 2014 (UTC)

@Zolo: - Somebody should add this as an example to Wikidata:Wiktionary. It makes sense to use the most-used word for the statement, and give the synonyms a property that link to the item that is used in the statements. If Wiktionary is imported won't users have the possibility to choose a word for lake in 300 different languages plus synonyms? -Tobias1984 (talk) 14:45, 24 May 2014 (UTC)
Tobias1984 English speakers in Ireland and Scotland use 'loch' for certain lakes and sea inlets, when speaking English. It (and 'Lough') appears on maps in English. The translations into other languages appear on the labels for this item in other languages, not as aliases to the English label. Filceolaire (talk) 19:38, 24 May 2014 (UTC)
Zolo Yes. You could loch say is not a 'type of lake' but some lakes are called 'loch' or 'lough' and others aren't so there may be a place for it. Personally I think 'type of lake' should be used for the scientific classification for the lake type (as en:Types of lakes) and not for the name so I would oppose using 'type of lake:loch'. Filceolaire (talk) 19:38, 24 May 2014 (UTC)
@Tobias1984:. My understanding of Wikidata:Wiktionary is that word entries will be stored in a different namespace, which should avoid confusions. --Zolo (talk) 04:54, 25 May 2014 (UTC)
@Filceolaire:. I would think that the most nature place for telling that some places are called 'loch' or 'lough' is the label (even though for some topics, a single, unqualified label is not enough to be precise, so I may support more "name" properties). -Zolo (talk) 04:54, 25 May 2014 (UTC)
@Zolo, Filceolaire: Anyway imho we should not have a type of lake property at all, but a type of lake class of classes item. Wrongly assuming loch is a type of lake we could have
< Lochness lake > instance of (P31) View with SQID < Loch >
,
< Loch > instance of (P31) View with SQID < Type of lake >
and
< Loch > subclass of (P279) View with SQID < Lake >
. TomT0m (talk) 10:53, 25 May 2014 (UTC)
Except that 'Loch' is used both for lakes and for sea inlets. It is not a subclass of Lake nor an instance of a 'type of lake'. It is a subclass of 'body of water' and an instance of a 'term imported from Gaelic'. Filceolaire (talk) 11:27, 25 May 2014 (UTC)
@Filceolaire: That's why I wrote wrongly assuming :) TomT0m (talk) 13:13, 25 May 2014 (UTC)
@TomT0m: oops, so you did. my mistake. Sorry. Filceolaire (talk) 16:25, 25 May 2014 (UTC)

Problem with the Merge gadget

Hi. I recently got an error message using the Merge it with... gadget that said that there was an editconflict reporting the merged item to RfD. Is there a way to change the gadget so as to avoid there ever being editconflict problems? (E.g. if it adds a new section instead of editing the page or something?) It Is Me Here t / c 15:29, 25 May 2014 (UTC)

I believe the developers are hard at work on redirects. Once these are working the merge gadget will need to be rewritten to redirect the merged item rather than deleting it so edit conflicts on the rfd page should no longer be a problem. Filceolaire (talk) 16:23, 25 May 2014 (UTC)

0-statement items

Since February the items with 0-statements have stopped decreasing: (http://tools.wmflabs.org/wikidata-todo/stats.php). Is there any way we can get this number to drop again? I have suggested a few item-groups here: Wikidata:Bot requests#Stagnating fall of 0-statement items, but there are probably much better bot-tasks hidden in those 5 million items. The problem might also be connected to the many items that only have one label. A good idea would be to put a message (only visible for logged in users) on the Wikipedia pages of one-label-items, that asks them to translate the item into as many languages as they know. We could also do the same asking the user to add one statement if the item has none. -Tobias1984 (talk) 15:57, 21 May 2014 (UTC)

Hi Tobias1984, I think we created quite a few missing items over the last couple of weeks. These start as items without any statements. On the other side my bots have been working on identifying categories. That seems to have cleared about 100.000-150.000 items and I still have some left to work on. Looks like I can tag about 30.000 more. After that I could do the same trick for the templates. Probably another 50.000. A good way to approach is to make a list of items without claims and try to find subsets. I made the list for nlwiki and enwiki. Plenty of stuff to work on. Multichill (talk) 18:39, 21 May 2014 (UTC)
I believe that we should invent some new properties and class items for many of such "0-staters". Like "Outline of sports" or "Russia in the Eurovision Song Contest 2014". --Infovarius (talk) 13:28, 26 May 2014 (UTC)
@Infovarius: - Your example are very difficult cases. And we can't ignore those items because Wikidata is supposed to be the database for Wikipedia. The first example could say "instance of = outline" &qualifier "topic = sports". The second example could say "instance of = subjects participation of an event" &qualifier "subject = Russia" &qualifier "event = Song Contest 2014". But there might be easier ways of classification. -Tobias1984 (talk) 17:30, 26 May 2014 (UTC)

How to map fracking chemicals?

I'm trying to figure out how to import this list of fracking chemicals. I think we have most of them if not all, but what is missing is the classification. I was thinking of using "subclass of:hydraulic fracturing proppant (Q2826784)" in all of them, plus "subclass of" the column labeled as "product function" (acid, biocide, breaker, etc). Would that be correct or should I use use (P366) instead?--Micru (talk) 11:22, 24 May 2014 (UTC)

@Micru: I wakly favor using the property use (P366) and use subclass of (P279) to climb up the lather of chemical classification. We have molecular function (P680) which is used for "what does the chemical do in an organism". Maybe we should have another property for "what does the chemical do in a watery solution" and use it for the "Chemical Purpose"-column in the report you linked. -Tobias1984 (talk) 14:41, 24 May 2014 (UTC)

I think we could need other properties. In particular, a property
< product > chemical properties search < acid >
could be useful. To express how they are used in industry, we could use something like
< product > used for property search < acid >
by industry search < industry >
, or the other way aroud
< mining industry > uses (P2283) View with SQID < product >
for its property search < property >
on substance search < gold >
results in search < pure solid gold >
.
More generally, I think we need a model for processes :
  • An algorimth as typically input datatypes, output datatypes, a purpose, a complexity ... ;
  • an industrial process is used by some industry to transform something into some other thing using some kinds of machines and some other products ;
  • ...
TomT0m (talk) 14:45, 24 May 2014 (UTC)
Tobias1984, ok, then "subclass of" and I'll propose a new "chemical purpose" (actually use (P366) could be renamed as "purpose" which is more generic and then we wouldn't need a new property).
TomT0m: Regarding your proposal
< product > chemical properties search < acid >
, wouldn't be that the same as
< product > intrinsec property search < acid >
? I suggested it with other uses in mind, but since it is a general property, it could be used for chemicals as well. As you, I also noticed a lack of general properties for systems and machines. The other day I was just thinking how to express that a
< machine > system input search < energy/raw material >
and
< machine > system output search < product >
. We have almost no vocabulary for basic concepts like this, and we could do so much better. Do we need a "task force for systems" and give it more thought about possible use cases or do you want to propose only those properties for now?--Micru (talk) 16:26, 24 May 2014 (UTC)
Heu do you have an idea of the number of applications of some chemicals ? Snipre (talk) 11:34, 26 May 2014 (UTC)
@Snipre: can you be more specific ? TomT0m (talk) 15:57, 26 May 2014 (UTC)

Error adding page link to item

I just tried to add en:Saeculum obscurum as a Wikipedia article link to Q867808, and got the following message:

An error occurred while trying to perform save and because of this, your changes could not be completed.

Using the latest Chrome on Windows 7. Any ideas? Please ping me if you reply, I don't check in here often. Thanks. — Scott talk 22:29, 25 May 2014 (UTC)

@Scott: Looks like it's already linked to Q1359213. --AmaryllisGardener talk 22:43, 25 May 2014 (UTC)
I see, thanks. That error message could do with some improvement!
I've virtually never used this project; what do you do when there are two items for the same thing? The topic that some of the Wikipedias call the "Saeculum obscurum", others call the "Pornocracia"/"Pornokratie"/"Pornocratie"/"Pornocrazia" (etc.). It seems to be a historical matter of some contention. On the English Wikipedia, the topic was also previously at the title "Pornocracy". — Scott talk 23:11, 25 May 2014 (UTC)
@Scott: You merge them. I like the merge.js tool (which can be enabled in your preferences). It merges the items and adds the duplicate item to WD:RfD. --AmaryllisGardener talk 23:19, 25 May 2014 (UTC)
Great. Looking further into this particular case, I can see there are a minority of Wikipedias that have articles under both titles, some of which even disagree with each other (on a Wikipedia? Who'd have thought?) so I can't resolve that, and will leave it as is. Thanks for the quick help, see you around. — Scott talk 23:34, 25 May 2014 (UTC)
Wikidata:Interwiki conflicts/Unresolved/2014#Pornokratie--Oursana (talk) 10:08, 26 May 2014 (UTC)

What do do with Wikipedia:Draft articles

English Wikipedia recently created a Draft namespace. This is for articles which are under development. Articles there are not subject to notability rules until they are ready to be moved to the main namespace. They are also excluded from Search engines.

  1. Should we allow sitelinks to these draft articles from existing Wikidata items?
  2. Should we allow the creation of new Wikidata items associated with these draft articles?
In favour
  • We want articles to have infoboxes linked to wikidata. Allowing new items and sitelinks to existing items associated with draft articles will mean editors of draft articles can work at including these infoboxes in their draft articles before they publish them to the main wikipedia.
Against
  • Some of these articles may never become mainspace articles, perhaps because they are just not notable. Allowing sitelinks and items associated with them is just a way of side stepping our notability rules.

I'm not sure either way.What do you think? Filceolaire (talk) 05:42, 23 May 2014 (UTC)

Is there a way we could have Draft items on wikidata - say regular Qitems with 'Draft:' at the start of their names in every language? Filceolaire (talk) 05:42, 23 May 2014 (UTC)
IMO drafts and articles less than 2 weeks only shouldnt be included in Wikidata, let alone imported into Wikidata. We should give the local project time to review the article. We have enough 'data' to work on, without us doing double handling of the firehoses of 500+ projects. Infoboxes are often wrong until the relevant WikiProject has processed the new article. This often happens because editors copy the infobox from another article, and dont replace/empty all of the values. John Vandenberg (talk) 05:53, 23 May 2014 (UTC)
+1 to wait before creating the item.--Micru (talk) 07:35, 23 May 2014 (UTC)
We could say that draft articles are allowed to be connected to items here, but that a link to only a draft article is not enough to be notable. This way draft articles can use the data on Wikidata and we don't end up with dozens of empty items. Multichill (talk) 16:50, 23 May 2014 (UTC)
There are definitely benefits in letting a draft be linked to an existing item which has a lot of properties. If we allow this for 'Draft' on English Wikipedia, we also need to allow it for userspace drafts. The draft functionality is not available on all Wikipedias (only available in English Wikipedia at present, I think).
However IMO we need to prevent people creating items specifically for a draft, or worse sucking up data from an infobox on a draft into Wikidata. Your wording comes close to achieving this, but the focus on notability will result in people creating items for a draft and then leaving the Wikidata to determine notability, which would be a lot of unnecessary notability discussions duplicating the same discussions that will happen on English Wikipedia once the article is published into mainspace.
How about:
Draft articles may be connected to an item which already has sitelinks to another language of the same project family.
i.e. if the topic is already considered notable in French Wikipedia, the English Wikipedia draft can be added as a sitelink. If the topic has only sitelinks to English Wikiquote or English Wikisource or Wikimedia Commons, the draft is not allowed to be added as a sitelink, as those projects have much lower notability levels. This isnt ideal, but it assists article translators and does prevent abuse.
If this is to proceed, we need to decide what happens to interwiki links on the client projects. I believe the Wikidata Client should not display the sitelinks to drafts on other Wikipedias (i.e. French Wikipedia shouldn't have an interwiki link to the English Wikipedia draft) by default or without the local community support for this. Showing drafts in interwikis might be desirable once they are marked as a draft using a 'badge'? John Vandenberg (talk) 00:00, 24 May 2014 (UTC)
+1 to Multichill. --Ricordisamoa 23:51, 23 May 2014 (UTC)
Pictogram voting question.svg Question Couldn't there be several drafts for the same article? So linking one instead over another might be confusing? --- Jura 04:28, 24 May 2014 (UTC)
Pictogram voting question.svg Question To make a good draft, they might need information from Wikidata available. In that case, shouldn't they create an item as soon as possible add add information here? --- Jura 04:28, 24 May 2014 (UTC)
Pictogram voting comment.svg Comment Not to mention the Wikidata notability rules are more relaxec than Wikidata. The presence of one Wikidata item is sufficient to create an item, but an item does not need a Wikipedia article. Think of a book author who is not notable by himself in Wikipedias but is notable as the author of a reference book. I think we should stick to this : we create the item if he his notable in Wikidata, and if there is an item, we make the connection to the Draft (no reason not to, it's an information related to the item.) This is simple and applicable. An item on Wikidata just identify a subject, who is useful if e can give datas about it and if it is an identifiable subject. The Notability on Wikipedias is another subject. TomT0m (talk) 09:50, 24 May 2014 (UTC)

Symbol oppose vote.svg Oppose Draft articles are excluded from search engines for good reason. Consider wikidata as a search engine as well, and you will bring up additional unwanted links to all search engines that use wikidata as a source. It is early enough to link it to wikidata as soon as it is in main namespace, you also don´t need to monitor if an article has moved to a different namespace.--Giftzwerg 88 (talk) 16:40, 24 May 2014 (UTC)

Symbol oppose vote.svg Oppose As a schematic batch of information we already have Wikidata itself. We don't need linking proto-articles. I don't see the need for hurrying, if the draft doesn't go wrong it'll be linked to Wikidata once it's finished. I see no problem having a Q item here, while the article is being developed in en.wikipedia or whatsoever, but clearly (to me, at least) it's not neccesary having them related.Totemkin (talk) 17:38, 25 May 2014 (UTC)

Symbol oppose vote.svg Oppose Draft namespace has been created for articles who may or may not adhere to notabilty rules on en.wiki - which are IMHO already far too permissive, considering the it.wp (or de.wp) ones. I don't think we need that. --Sannita - not just another it.wiki sysop 10:03, 26 May 2014 (UTC)

Symbol oppose vote.svg Oppose per above. Unfinished articles have no place here. Mushroom (talk) 11:00, 26 May 2014 (UTC)

Symbol oppose vote.svg Oppose draft articles are not part of the encyclopedia (yet) --CutOffTies (talk) 12:10, 26 May 2014 (UTC)

Symbol oppose vote.svg Oppose per above. --AmaryllisGardener talk 12:24, 26 May 2014 (UTC)

Symbol oppose vote.svg Oppose - Drafts are articles in development and are little different to user subpages. Once they are mainspace articles they can be added. Green Giant (talk) 12:53, 27 May 2014 (UTC)

Wikidate Person Game: fictional people?

I have been playing this game and several fictional people have come up. I have played it safe by picking "not sure", but what is the desired answer? —Phil Boswell (talk) 09:31, 23 May 2014 (UTC)

Hey :) Clicking "person" will add "instance of: human" to the item as I understand it. Last I heard fictional people should not be tagged in that way (but it is a community decision). See for example Q4653. It is tagged as fictional human. --Lydia Pintscher (WMDE) (talk) 09:36, 23 May 2014 (UTC)
I thought as much, so just as well I played it safe. Myself, I would prefer to be able to mark something as "instance of :<species>" and "fictional", the latter possibly coupled with the author and/or the work within which it appears, but I guess that will have to await developments ;-). Thanks anyway. —Phil Boswell (talk) 09:41, 23 May 2014 (UTC)
@Phil Boswell: : there will be no such development coming (from the dev team) as it's out of their plan. Here we regroup fictional in fictional classes, and we use the fictional analog of (P1074) property to link the classes to their real world equivalent, for example
< Fictional Cat > fictional analog of (P1074) View with SQID < Cat >
and
< Felix the cat > instance of (P31) View with SQID < fictional cat >
. TomT0m (talk) 10:05, 23 May 2014 (UTC) But I understand now you meant Wikidata the game devs /o\
Thanks for those answers. I'm also hanging fire on disambiguation pages for the time being: maybe at some point these could get some special attention…—Phil Boswell (talk) 12:02, 23 May 2014 (UTC)
My take was that fictional humans should be marked as 'instance of:human' and also as 'instance of:fictional character' or 'biblical character' (sidestepping the question as to whether these are fictional) or 'legendary character' (sidestepping the question as to how historical these are) or 'mythical character'. Felix the Cat would be 'instance of:funny animal (Q2369882)' (a specialised term for cartoon animals who act human). At least that is my take. Filceolaire (talk) 17:27, 23 May 2014 (UTC)
They are not human, a human is an instance of the "Homo sapiens". GerardM (talk) 12:14, 24 May 2014 (UTC)
Various people in the 'Lord of the Rings' are described as being 'men' or 'women' and are clearly meant (by the author) to be humans and fictional characters. Sherlock Holmes is fictional but is also a human. Robin Hood may or may not have existed or may be based on someone who existed. I would say he is an instance of a 'legendary character' and a 'human'. Filceolaire (talk) 19:18, 24 May 2014 (UTC)
@Filceolaire: they are instances of fictional human (Q15632617). The reason why it was decided to use fictional X item rather than an "is fictional" standalone property was because fictional people should not appear by default in queries about people, etc. Tagging p31ing them as "human + fictional person" would defeat this. --Zolo (talk) 04:48, 25 May 2014 (UTC)
The problem with this policy is that there are people where it is not clear if they are fictional or not. Notably no label (Q14943515) - much of the old testament is written as history and is as likely to be true as some of the characters in the Babylonian king lists - and no label (Q16934977) - many of which could be historical or at least based on historical characters. By using these items as values for instance of (P31) instead of just fictional character (Q95074) we can sidestep those disputes and separate out the question as to whether the character is a human or a chimera or an elf as a separate statement. At least that is how I see it. Filceolaire (talk) 16:11, 25 May 2014 (UTC)
I agree and I was not really in favor of this solution either. But now that we have settled on it, it seems best to stick to it. It also has benefits. Actually, we still need to find out a good way to express hesitations and alternatives. It may be that the issue is just the same for "instance of: (human or fictional human)" as with any other form of alternative like "author: (x or y)". --Zolo (talk) 17:58, 25 May 2014 (UTC)

──────────────────────────────────────────────────────────────────────────────────────────────────── This subject was discussed in great detail in Wikidata:Requests_for_comment/Migrating_away_from_GND_main_type#Fictional_entities. Simply put, Wikidata is doomed if users query "people born in United Kingdom in 1980" and get back "Harry Potter" among the results. Stating that Harry Potter is an instance of human and an instance of fictional character would entail that dooming scenario. The solution here is to use disjoint class hierarchies for (real) entities and fictional entities. Fictional entities should be excluded from queries by default even if that makes modeling fictional entities unwieldy. Wikidata should be unabashedly biased toward the natural world.

Filceolaire, the question of how to model people of dubious historicity is interesting, but I don't think we can sidestep it by classifying a subject as, say, "instance of: Babylonian king". The term "Babylonian king" -- without further qualification -- should be interpreted to mean non-fictional, actual kings of Babylon. All instances of Babylonian king were also instance of (P31) human (Q5); that is, they were all actual people.

Zolo, regarding a "way to express hesitations and alternatives", I think we have two decent mechanisms. First and foremost, the built-in rank feature for statements seems suitable for this: if the statement Abraham (Q9181) instance of (P31) human (Q5) is held to be incorrect by a preponderance of reliable sources (as seems to be the case per en:Abraham#Historicity_and_origins), then we could add the statement Abraham (Q9181) instance of (P31) fictional human (Q15632617) and assign the latter the 'preferred' rank and the former the 'deprecated' rank. Second, we can capture information about disputed statements via the property statement disputed by (P1310).

Below is how this approach would model the Biblical patriarch Abraham:

instance of (P31) fictional human (Q15632617) (preferred rank). Source: The Historicity of the Patriarchal Narratives (Q16947623), Abraham in History and Tradition (Q4669283)
instance of (P31) human (Q5) (deprecated rank). Source: Kenneth A. Kitchen (Q1669960), William F. Albright (Q451799), Albrecht Alt (Q84895)
depicted by (P1299) Old Testament (Q19786)

Perhaps statement disputed by (P1310) would be redundant here, perhaps not, but I think this example demonstrates how we can reasonably model subjects of questionable historicity through ranking. Emw (talk) 17:46, 26 May 2014 (UTC)

@Emw:. Yes, actually it seems to work here, because fictional and non fictional is mutually excludable, so it seems that there is insuperable technical issue here. But other cases about alternatives seem harder to express (like A says X was written by both Y and Z while B says it is either Y or Z). --Zolo (talk) 18:06, 26 May 2014 (UTC)
@Zolo: I don't think we got a problem. Clearly, if we take Jesus for example, we can have several items. There is probably historians who work on religion and can source or argue that it's an historical character. Then there is Jesus in fictional works, who are clearly different from the historical or even the biblical one. We could even add some properties like inspired from or other like parody of, classes like biblical character. It's nothing but a POV conflict if some persons thinks something existed and some other else think otherwise. We can deal with this. TomT0m (talk) 18:13, 26 May 2014 (UTC)
@TomT0m: I think the main issue here is when we do not know if the depiction in a given text is fictional or not. Some people think the Bible's story about Abraham is all true, some that is all false, and some that it is a romanced version of the story of a real person, but it is all based on a single text in the Bible. Can we use inspired by (P941) here ? That seems a bit imprecise. said to be the same as (P460) does not seem quite right either. So a new property ? --Zolo (talk) 05:56, 27 May 2014 (UTC)
I think we're fine if we stick to objective things. A religious story is a religious story, it's up to individuals to believe it or not. Abraham might or not have been a real human, this does not change the fact it's a biblical character. I think there is for example no problem to say that the Abraham character in The Ten Commandments (Q746733) (view with reasonator) is inspired by the biblical one, as the story is a free interpretation of the bible. I'm also not opposed to use the statement disputed by (P1310) View with SQID property here. For example we could use a statement , sourced by whatever religious authority. We also might have a problem with some entities like saint (Q43115). Anyway, I'm not aware of a real controversy about the existence of Adam for example, it's too old for a real historical controversy, hence it's a pure matter of belief. We should not push too hard this conversation without a real use case with documents about the controversy, I think. TomT0m (talk) 08:29, 27 May 2014 (UTC)

Adding statement Commons category does not work

Yes. After entering and klicking on "save", there is just the "Saving..." message and you can wait until the cows come home for it has really been saved. Please fix. --A.Savin (talk) 14:29, 24 May 2014 (UTC)

I have the same in one Firefox installation since Thursday afternoon with every statement (well, only tried the statements linking to items yet), whereas the Firefox on my laptop continues to work as usual. I already tried the standard procedure - clearing the cache, rebooting. Chrome on the same system continues to work fine. Ahoerstemeier (talk) 22:03, 24 May 2014 (UTC)
Thanks, but unfortunately purging does not help, and changing the browser from FF to anything else isn't a solution for me. --A.Savin (talk) 08:14, 25 May 2014 (UTC)
Maybe you misunderstood me - purging didn't help for me either. So I don't have a solution yet, and don't like to change browser just because of this annoyance. Ahoerstemeier (talk) 10:39, 25 May 2014 (UTC)
I had today the same problem but only if the UI-language is set to English. With other languages adding statements works fine for me. --Pasleim (talk) 11:45, 25 May 2014 (UTC)
Indeed. Thanks. --A.Savin (talk) 18:08, 25 May 2014 (UTC)
Yes, that "fixed" it for me as well, just strange that the problem isn't present on my other Firefox installation, which also uses the English UI. Ahoerstemeier (talk) 07:38, 26 May 2014 (UTC)
And now it works again in English on the system which was broken before. Strange... Ahoerstemeier (talk) 07:42, 27 May 2014 (UTC)

Subject of a book

Do we have a property to say that one object (e.g. The King of Rome (Q594)) is the subject of another object (e.g. the book, The King of Rome (Q15098285)); or vice versa? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 14:37, 27 May 2014 (UTC)

Andy Mabbett: That would be main subject (P921). --Micru (talk) 15:24, 27 May 2014 (UTC)
Thank you. Does that include specific subjects like the above example, rather than generic subjects (like "pigeon racing")?
We also have depicts (P180) but that seems to be just for images. Filceolaire (talk) 19:11, 27 May 2014 (UTC)
Andy Mabbett: you can be as specific as you want, as long as you use a reasonable number of items to describe it. Even better if those items that you use are properly classified with "subclass of".--Micru (talk) 21:07, 27 May 2014 (UTC)

Rollback Malfunction?

For a mysterious reason, my rollback right does not work properly from today. I can still see the rollback links, but when I click them, what I only receive back is the message "Rollback failed — edit conflicts". Is this a technical problem? Bluemersen (+) 13:33, 28 May 2014 (UTC)

This is a known issue and is a result of some js changes made to the Wikibase extension included in wmf6. The bug for this is bugzilla:65831. A backport patch has been made so it will likely be fixed at some point today during a deployment :) John F. Lewis (talk) 13:40, 28 May 2014 (UTC)
Thanks for the quick reply :) Bluemersen (+) 13:42, 28 May 2014 (UTC)
Thanks for asking the same question question everyone's been thinking :) John F. Lewis (talk) 13:43, 28 May 2014 (UTC)
Was wondering why it wasn't working for me. Good to know I haven't broken it. —Tom Morris (talk) 15:53, 28 May 2014 (UTC)

instance of (P31): sibling duo (Q14073567) or human (Q5)

(Regarding Protus and Hyacinth (Q3394205))If an item is about a (human) sibling duo, should that item be tagged with the instance of (P31): human (Q5) too, or is sibling duo (Q14073567) enough? (tJosve05a (c) 21:21, 28 May 2014 (UTC)

And if the item is about multiple people like Sullivan brothers (Q1850286). (tJosve05a (c) 21:35, 28 May 2014 (UTC)
Josve05a, the statement instance of (P31) human (Q5) should only be used on an individual human. I'd go with some other value for P31 for your example, like sibling duo (Q14073567). Emw (talk) 01:32, 29 May 2014 (UTC)
try "group of people".. siblings can be deduced. GerardM (talk) 15:56, 29 May 2014 (UTC)
Thanks! (tJosve05a (c) 17:34, 29 May 2014 (UTC)

STOP „The Game“ (Merge)

Please STOP using Magnus Merge-Game! It's malfunctioning. --Succu (talk) 21:52, 21 May 2014 (UTC)

Just the merge game or the whole game? --AmaryllisGardener talk 22:06, 21 May 2014 (UTC)
[13] Vogone (talk) 06:32, 22 May 2014 (UTC)
@Succu: What exactly is it doing wrong? --Jakob (talk) 22:13, 21 May 2014 (UTC)
  1. There are requests for deletion without having really merged the items.
  2. The tools does not check if the items have backlinks.
  3. There are undoubtedly more false merges than on other days. --Pasleim (talk) 22:20, 21 May 2014 (UTC)
Pasleim: Thanks. --Succu (talk) 05:25, 22 May 2014 (UTC)

I think it's a great improvement over the other user interfaces for the same tasks. --- Jura 05:08, 22 May 2014 (UTC)

Someone edited the topic of this thread and his previous comment. No good. --- Jura 06:52, 22 May 2014 (UTC)

So the trouble is going on. I have the strong feeling I'm wasting my time... Echinocactus mihanovichii (Q14943168) has a backlink to Gymnocalycium mihanovichii (Q137994). It was proposed by Magnus new merge game as a merge candidate and merged. Geohintonia (Q8178626) has a backlink to Geohintonia mexicana (Q137774) This monotypic genus was also proposed by Magnus new merge game as a merge candidate and merged. And so on. I'm really mad. --Succu (talk) 15:50, 22 May 2014 (UTC)

@Succu: Want me to message those that appear to be playing the merge game? --AmaryllisGardener talk 17:01, 22 May 2014 (UTC)
@AmaryllisGardener: If this is possible, yes. But it would be better Magnus could fix the problems, but he seems not to be available at the moment. --Succu (talk) 17:29, 22 May 2014 (UTC)
In all the lists based by bot, there are always false positive item... Ex : User:Ivan A. Krestinin/To merge (In this case, there are a lot false positive item). The users have just to use properly the tool... Idem for backlinks. And wikidata have to show its "rules" and "problematique" to the newbee or the ancient who didn't know those rules, but come because a tool is nice. --Nouill (talk) 18:55, 22 May 2014 (UTC)
@Nouill: You missed the point. All in all this game is really usefull and a great improvement. But this tool has some serious flaws that needs to be fixed. I doubt that a lot of users playing this game noticed my reverts. --Succu (talk) 19:11, 22 May 2014 (UTC)
For me, we just have to stop the game temporally. The community and the admin have just to much work to check the item merged, and to explain the contributor explicitly when they do wrong merge. --Nouill (talk) 19:15, 22 May 2014 (UTC)
@Nouill: Thats exactly what I asked Magnus yesterday (before I opened this thread). I hope you noticed my revert? --Succu (talk) 19:23, 22 May 2014 (UTC)
Sorry. I put a message on Magnus's page. --Nouill (talk) 19:34, 22 May 2014 (UTC)

My internet connectivity is limited for the next week or so. I'll try to deactivate the merge sub-game. FYI, the game is working well. It's the Wikidata API that's buggy. --Magnus Manske (talk) 21:05, 22 May 2014 (UTC)

Update: Merge game deactivated for now. --Magnus Manske (talk) 21:15, 22 May 2014 (UTC)
Thanks, Magnus. I hope we can restart the game in a few days. But with the buggy Wikidata API and players who just click for fun on 'same topic' without knowing the consequences, it was a real pain for all admins. --Pasleim (talk) 22:43, 22 May 2014 (UTC)

@Magnus Manske: „the game is working well“ FYI, why did you post request on Rfd if the merge had returned with error? I don't think is working well. --JulesWinnfield-hu (talk) 08:27, 23 May 2014 (UTC)

If you had actually read the complete sentence from which you copied, you might have learned that it is not my game that is broken, but the part of Wikidata that does the merging. But, I guess, reading half a sentence is the best that one can expect these days...--Magnus Manske (talk) 10:59, 23 May 2014 (UTC)
If the API doesn't succeed, it gives you an error code. You posted Rfd requests despite the error. See bugzilla:65215. --JulesWinnfield-hu (talk) 11:17, 23 May 2014 (UTC)

Many invalid merges appeared last days. Very interesting merge sample. Human and geographical object were merged. Is the tool provide to user enough information for decision? Or information is available only by additional clicks? — Ivan A. Krestinin (talk) 11:19, 24 May 2014 (UTC)

@Magnus Manske: The „Merge-Game” started again, but nothing changed. Backlinks are not checked (example see basionym). --Succu (talk) 13:51, 29 May 2014 (UTC)

Ahhhhh! --AmaryllisGardener talk 13:55, 29 May 2014 (UTC)
"The „Merge-Game” started again, but nothing changed." Wrong. API merge function was fixed, game now resolves edit conflicts. Also, big fat "be careful with da merging!" message on top ;-)
"Backlinks are not checked" Correct. Is that a requirement somewhere? Automatically fixing all backlinks would be quite difficult (sources? qualifiers?), and should be implemented centrally and not in every merging tool, IMHO, if so desired.
I have written a helper script to better cope with the flood. Also, note that the merge game is "self-limiting"; we're seeing a flood now, but once it's through (at 7% now), rates will drop massively again. The underlying cause are the many item duplications in Wikidata. --Magnus Manske (talk) 14:23, 29 May 2014 (UTC)
I know about the API fix, but I'm the one who has to how fix such wrong merges. --Succu (talk) 19:48, 29 May 2014 (UTC)
Anything for admins playing the merge game in the works? ;) --AmaryllisGardener talk 14:36, 29 May 2014 (UTC)
As an admin, you should now have an option for direct deletion here. I'm not an admin here and can't test, so I have no idea if this actually works. Please tell me either way :-) Note this will not resolve links to the deleted item! --Magnus Manske (talk) 18:15, 29 May 2014 (UTC)
Great. But now it says I'm not logged in to WiDaR, and when I tried to log in to it, I get an error saying "No webservice". Nevermind, working now. --AmaryllisGardener talk 18:22, 29 May 2014 (UTC)
Update: Upon trying the delete feature, it says "DELETION FAILED:Unrecognized value for parameter 'action': =delete Not trying again." --AmaryllisGardener talk 18:24, 29 May 2014 (UTC)
Oops, please try again! --Magnus Manske (talk) 19:01, 29 May 2014 (UTC)
Now it says "DELETION FAILED: Permission denied Not trying again". --AmaryllisGardener talk 20:53, 29 May 2014 (UTC)
It appears that you have the right to delete pages, but my OAuth consumer doesn't. Waiting for approval of new version. Sorry for the "remote debugging", but I can't test admin functions here myself... --Magnus Manske (talk) 21:46, 29 May 2014 (UTC)
It's working now. Thank you. --AmaryllisGardener talk 22:01, 29 May 2014 (UTC)
@Magnus Manske: It would be nice if the deletion summary was something like: "merged into Qxxxx" instead of the meaningless default one. Regards, Vogone (talk) 22:21, 29 May 2014 (UTC)
Should work now, please confirm. --Magnus Manske (talk) 22:51, 29 May 2014 (UTC)
Looks good. --AmaryllisGardener talk 23:53, 29 May 2014 (UTC)

Profession

Which using will we prefer for a human (for example, Jean-Daniel Kieffer (Q8340034))? a) occupation (P106) linguist (Q14467526) + orientalist (Q1731155) or b) field of work (P101) linguistics (Q8162) + Oriental studies (Q476294) ? --Infovarius (talk) 05:00, 27 May 2014 (UTC)

I would say Occupation:Civil Servant, Author, Translator. plus b) above. Filceolaire (talk) 12:26, 27 May 2014 (UTC)
Didn't get your point. Are you talking about specifically Q8340034? My question concerns general principle. --Infovarius (talk) 20:20, 29 May 2014 (UTC)
Yes I am talking about Jean-Daniel Kieffer (Q8340034). He worked for the french foreign ministry as a translator so I would say that his occupation (P106) is Civil Servant + translator + author (maybe Diplomat too) and his field of work (P101) is linguistics (Q8162) + Oriental studies (Q476294). I wouldn't put his Occupation as Orientalist. I think this illustrates the general principal of what kinds of things I would use for each of these properties.
I hope this helps. Filceolaire (talk) 21:22, 29 May 2014 (UTC)

Bug

I can't see the 2014 archive here, but exists here. --Kizar (talk) 19:33, 29 May 2014 (UTC)

I can see a link to the 2014 archive on the former page as well as links to the 2012 and 2013 archives. What exactly is the problem? Vogone (talk) 19:49, 29 May 2014 (UTC)

Create Account is broken

I try to create a new Wikidata account by clicking 'create account' at the top right of any wikidata page. I then fill in the form and click the green 'create your account' button. If I entered the correct fields and got the CAPTCHA right then I got the following page:

Error Not Found

The requested URL /w/index.php was not found on this server.

Additionally, a 404 Not Found error was encountered while trying to use an ErrorDocument to handle the request.

Further investigation reveals that my account was not, in fact, created. I've tried a few times now, but always with Chrome Stable (35.0.1916.114 m) on Windows.  – The preceding unsigned comment was added by 192.0.183.56 (talk • contribs).

This page is about how to improve the project chat page. Please ask your question there. --AmaryllisGardener talk 21:34, 29 May 2014 (UTC)
This issue was known at the time and has been resolved. Please try now :) John F. Lewis (talk) 21:40, 29 May 2014 (UTC)

People from Brazil

In a recent chat the question was asked how to query for the number of people from a given place of birth or how to query for the number of people who died from a given country. I presented a query for Brazil for the people who died in 2014. However, I had a good look and found that more than half the people from Brazil are not known to be Brazilian. I blogged about this.

What is needed to add the missing Brazilians is add 1,128 items and 12,127 items need to become both human and Brazilian. Thanks, GerardM (talk) 09:47, 25 May 2014 (UTC)

@GerardM: The new "nationality" mode of Wikidata Game might help with this :-). Helder.wiki 15:45, 30 May 2014 (UTC)
Regarding this "game", I see a few problems. The game seems to suggest countries based merely on the birth town's current national status. This may cause many users to think e. g. someone who lived in the 19th century was a citizen of the Russian Federation though obviously this state didn't exist back then. The game should also consider these historical aspects and suggest countries accordingly depending on the date of birth resp. date of death. Vogone (talk) 16:19, 30 May 2014 (UTC)

Links and ID edit gadget and template

Edit links to social networks
Edit Authority Control IDs
Result representation

Hi. Sorry for bad English. Believe this may be interesting. There is a new gadget in ru-wiki that helps to edit external links to social networks, linked sites and authority control identifiers from local site without visiting wikidata. Still there is one problem with it: API actions of wikidata w/api.php does not preserve "summary" field. Because of that it is not possible to distinguish manual edits and edits via gadget in wikidata history. Any help with gadgets (incl. bugreports) will be appreciated.

Also there is a new template that shows all links edited by this gadget: w:ru:Шаблон:Внешние ссылки (as a future replacement of Template:Authority control (Q3907614)). You can see how it works here. -- Vlsergey (talk) 23:38, 26 May 2014 (UTC)

  • Hi Vlsergey, looks like a promising project. Two questions: 1) You've deleted in the ruWP articles the old auhority temple. Did you check if the deleted IDs are identical with the IDs in Wikidata? 2) Your template (like the old template) has problems with VIAF. There are often two or more VIAF IDs for one person. The template is not showing the one that is marked as "preferred rank". --Kolja21 (talk) 01:25, 28 May 2014 (UTC)
  • Kolja21, they were not always identical. Sometimes ID (VIAF, GND) from ru-wiki was deleted from viaf.com -- i've just dropped those. Sometimes it was still present, so I added second values to wikidata and let wikidata team (who has contacts with VIAF/GND) to handle it. Yep, current template doesn't handle multiple values nicely, but new template will do in one week (after merge). It will output all values. -- Vlsergey (talk) 09:28, 28 May 2014 (UTC)
  • Update. New template supports multiple values now. -- Vlsergey (talk) 16:44, 30 May 2014 (UTC)
  • Nice project. It's very interesting to edit Wikidata from Wikipedia. Pyb (talk) 04:23, 28 May 2014 (UTC)
  • I find it a super idea and I hope that it can be internationalized and used in other projects too.--Micru (talk) 06:58, 28 May 2014 (UTC)
Вообще-то API Викиданных позволяет добавлять описание правки. --Infovarius (talk) 20:15, 29 May 2014 (UTC)
Оказывается, он игнорируется, если передавать в POST, и очень хочет видеть себя в URL. Хотя может я просто позабыл формат передачи. -- Vlsergey (talk) 08:00, 30 May 2014 (UTC)

Proposal to change English description of Q476028 from "football team" to "football club"

I propose to change the English description of Q476028 from "football team" to "football club" to reflect reality on how Q476028 is currently used in Wikidata. As this change would affect quite a few other items in Wikidata, I created a proposal on the talk page of Q476028 so that we can talk about this proposal there.--Bthfan (talk) 13:55, 29 May 2014 (UTC)

Make sure the old label is kept as an alias (posting this comment on Project Chat as it applies to other cases where items are renamed). Filceolaire (talk) 21:14, 29 May 2014 (UTC)
Ok, will do. --Bthfan (talk) 07:33, 30 May 2014 (UTC)

Q16999835

Hi

I've just created item Dolores Puthod (Q16999835). Since I’m pretty new at this, can someone have a look and see if everything is ok/nothing missing? That would be too kind. Thanks in advance. Jihaim (talk) 18:43, 29 May 2014 (UTC)

ːI’m not sure about “DNB editions” ːP Jihaim (talk) 18:58, 29 May 2014 (UTC)

The item looks good. --Pasleim (talk) 11:19, 30 May 2014 (UTC)

Property value

I'm looking for a way to have a list of item and value for a specific property. Ideas? --Sbisolo (talk) 15:25, 30 May 2014 (UTC)

Types of name

Playing "The Wikidata Game", I ran into Crafton (Q16275246) View with Reasonator View with SQID which is ... both a family name and a given name

I search the subclass tree :

<family name> on wikidata tree visualisation (external tool)(depth=1) This does not seem right as personal name (Q1071027) View with Reasonator View with SQID in french is, in a full person name, everything but the family name (it is the first given name with maybe the second, third, ...). I think we should have more something like:

But what to do with item which denotes names that are both family and given name ?

  1. < Crafton (Q16275246) View with Reasonator View with SQID > instance of (P31) View with SQID < name >
     ?
  2. < mixed name > subclass of (P279) View with SQID < name >
    and
    < mixed name > subclass of (P279) View with SQID < given name >
     ;
    < family name (Q101352) View with Reasonator View with SQID > instance of (P31) View with SQID < mixed name >
     ?

I like the second option as we naturally capture the intersection of both sets of names.

Main question, @Magnus Manske: how to integrate that in the game (I think it's worth it as it's a common cases) ? what to do with personal name ? TomT0m (talk) 16:40, 30 May 2014 (UTC)

Where to find phase 2 roadmap and progress?

Wikidata:Introduction says phase 2 started to roll out more than a year ago, but I'm unable to find any info about the its roadmap and progress. Where to find it? When it would be possible to use Wikidata infoboxes in wikipedias? -- Alexander Vasenin (talk) 18:25, 29 May 2014 (UTC)

@Alex.vasenin: - The original phase plan has probably underestimated the amount of work the database requires. I would say we are now in a mix of phase 1, 2 and 3 - depending on the subject your working on. You could join or create a WikiProject, if you are interested in a particular subset of data. General direction of Wikidata is summed up here Wikidata:Development plan (and a similar question: Wikidata talk:Development plan#Phase 3). -Tobias1984 (talk) 18:37, 29 May 2014 (UTC)
@Tobias1984: Any estimations about taxobox infobox availability date?  – The preceding unsigned comment was added by Alex.vasenin (talk • contribs).
@Alex.vasenin: - That depends very much on what data and which Wikipedia. But for basic information the taxobox could already be deployed to any Wikipedia. It is just up to that community to decide if they want to use Wikidata-data or not. You could also discuss that at: Wikidata:WikiProject Taxonomy - Tobias1984 (talk) 19:32, 29 May 2014 (UTC)
@Tobias1984: I'm interested in Wikidata taxoboxes in English Wikipedia. What's holding their introduction? -- Alexander Vasenin (talk) 19:58, 29 May 2014 (UTC)
@Alex.vasenin: - I think its mostly that it would replace a established and well-working system with a rather experimental system. Wikidata still has bugs, missing information and disagreements on how to structure the data. See for example the list of depreciated properties for taxonomy (Wikidata:WikiProject Taxonomy#Deprecated properties). When Wikidata was started it was thought that each infobox line should get its own property. This would be a simple data-import, and infoboxes could be generated using simple code (just replace each information on Wikipedia with a link to the property on Wikidata).
But this approach would mean that Wikidata would be a relative "dumb"-database, just saving and providing information like a simple spread sheet. So instead of going the path "first infoboxes and lists and gradually improve the database, through inheritance and weeding-out of duplicate information", the community decided to first improve Wikidata, to the point where information is intelligently stored and cross-linked, and worry about infoboxes later. That is a superior approach, but also means that much more work has to be done than simply importing every template line on Wikipedia.
Now the community has to worry about what properties should connect items and what information is inherited from these links. For example should we use a property "sibling" to connect siblings instead of "brother" and "sister"? The first inherits the sex from the other items, while the latter doesn't. The same questions also appear for taxonomic-data.
There is an additional problem that is new to Wikipedia. While it was perfectly fine not to source each line of an infobox on Wikipedia, Wikidata does require such a rigorous approach. That is because only if each statement of a long chain of statements is sourced, the information generated through such a query can be trusted. If you query for example "red birds in South America", then it doesn't help you if "feathers: red" and "subclass of: bird" are sourced, but "living in: South America" is unsourced. You can not be confident that the generated list is valid information. The information might even be wrong and you get too many birds listed.
So after we deal with sourcing, data-structure, and the problems that appear in such a young project, we can think about infobox deployment (probably starting with smaller Wikipedias and not the English one). -Tobias1984 (talk) 09:25, 30 May 2014 (UTC)
@Tobias1984: Too big expectations. Though I think Taxonomic data model is still not perfect because of the use of parent taxon which can be substituted by subclass of, I think taxonomic datas are ready to be used in an infobox. Imho waiting a source for each line of an infobox before deployment is just insane, infobox deployment is one of the key to get contributors. And speed up the source import. Otherwise it will take decades. TomT0m (talk) 11:01, 30 May 2014 (UTC)
@TomT0m: - I am any way in favour of gradual deployment. For data-areas that are ready we replace one infobox at a time (meaning around 100 a day). Then see what problems appear (data-accuracy and increase on Wikidata-server). The question of contributors is really a question of usability. Many people would probably contribute, if it didn't feel like proof-reading a calculus-text-book. I think Magnus is really building some important brdiges with accessible interfaces and making contributing more fun (Wikidata-game). I think a massive deployment would just flood all our pages with complaints and the current interface will give us a massive amount of good-faith vandalism (spanning 270 languages). And yes 100% sourcing is idealistic, but it is at least the direction we should go for. -Tobias1984 (talk) 19:23, 30 May 2014 (UTC)
@Tobias1984:@TomT0m: It looks like we face a chicken and egg dilemma: we need accurate data to deploy our infoboxes => we need inordinate amount of time to perfect the data => the only way to attract editors is to deploy infoboxes (back to square one). Personally I believe we should deploy infoboxes as fast as possible, but give editors the choice: use Wikidata taxobox with something like {{wd:taxobox}} or use traditional WP taxobox. Eventually the editors will gradually replace most old-style taxoboxes with Wikidata taxobox, and probably even make it a policy. What do you think? -- Alexander Vasenin (talk) 21:14, 30 May 2014 (UTC)

What statement (instance of) to use for anthroponymy articles?

Hi, the Wikidata "Disambiguation items" game presents to users (among other pages) so-called anthroponymy pages. These pages list persons with the same surname or given name (e.g. https://en.wikipedia.org/wiki/Archdale_%28surname%29 (Q16479440)) and the user should then decide if the item/page is a disambiguation page, a disambiguation page about a surname or a disambiguation page about a given name. When the user now clicks on surname, the game tags the item with "instance of Wikimedia disambiguation page (Q4167410)" and "instance of family name (Q101352)". Now some people think this is wrong, also see the bitbucket issue on this (https://bitbucket.org/magnusmanske/wikidata-game/issue/47/disambiguation-pages-are-not-anthroponymy), given that https://en.wikipedia.org/wiki/Wikipedia:Manual_of_Style/Disambiguation_pages#Given_names_or_surnames states:
"Articles only listing persons with a certain given name or surname, known as anthroponymy articles, are not disambiguation pages"

Magnus sees the issue a bit differently, see the bitbucket issue for details. Now what do you think about this, how should those anthroponymy pages (not) be tagged? "instance of [name]" only? Or maybe a new item should be created for this anthroponymy page type? Or is the current tagging by the game fine? --Bthfan (talk) 08:41, 30 May 2014 (UTC)

My opinion: Diasambiguation page are all the pages that Wikimedia software recognize like disambiguation with magic word __DISAMBIG__ see also Extension:Disambiguator. I don't think that to add "instance of family name" is correct, because we can't are sure that all the wiki add only disambiguation about family. --ValterVB (talk) 18:41, 30 May 2014 (UTC)
In my opinion also we should treat this pages as disambigs. As far as I check ~50 of them they don`t describe name - its only listings of people who have the same surname. So in my opinion its disambig. PMG (talk) 22:08, 30 May 2014 (UTC)

Why I can not add DDC properties?

Wikidata:Project chat#DDC
Hi! It happened yesterday and today:

✓ Done
The values do not store. I have purged the pages several times without result. It happened yesterday as well. FYI: 1177+ items with DDC values. Regards gangLeri לערי ריינהארט (talk) 07:14, 24 May 2014 (UTC)

לערי ריינהארט, I was able to add Dewey Decimal Classification (P1036) = 491.41 on Q33997. No idea why you may have had problems. Do you want to try again with another of those items to see if the problem has voodoo disappeared? John Vandenberg (talk) 08:04, 24 May 2014 (UTC)

@John_Vandenberg I managed to add b) and c) now. I am using a nightly build of FF "Mozilla/5.0 (Windows NT 6.3; WOW64; rv:32.0) Gecko/20100101 Firefox/32.0" with Windows 8.1. This happened yesterday as well and two times today (with an intermittent period of error free behavior). At some undefined time (later) addition of statements may "recover". Normally I have 80+ FF tabs open. This is why I use "purge" before adding values. Thanks for your help! לערי ריינהארט (talk) 08:17, 24 May 2014 (UTC)

It is back again: image:Doria Shafik.jpg needs to be added to Doria Shafik (Q4120224) . לערי ריינהארט (talk) 20:54, 24 May 2014 (UTC)
image:Alaedin.jpg needs to be added to Muhammad Aladdin (Q954686) . I can add labels and descriptions but no statements. לערי ריינהארט (talk) 21:07, 24 May 2014 (UTC)
image:Portrait of Ahmed Zaki Abushady, (1892-1955) as a young man, ca 1909, taken in Cairo, Egypt.jpg to Ahmed Zaki Abu Shadi (Q552273)
occupation: archaeologist should be added to Hermann Thiersch (Q1612904) .
@John_Vandenberg I am using more then 2,5 GByte of internal memory. This might be a combination of a FF memory managment problem and faulty Javascript execution. לערי ריינהארט (talk) 21:34, 24 May 2014 (UTC)
It does not relate to memory management. I restarted FF: Max Karl Ernst Ludwig Planck should be added as "birth name" to Max Planck (Q9021) .
It is an interface problem. I can use a clone of Magnus 's AC Javascript. see changes to Laozi Laozi (Q9333) View with Reasonator View with SQID . Regards לערי ריינהארט (talk) 22:11, 24 May 2014 (UTC)
@John_Vandenberg It was a private faulty Javascript whith a faulty button action.
✓ Done לערי ריינהארט (talk) 23:56, 24 May 2014 (UTC)

@John_Vandenberg There is a fault in the event queue handling: normally the user performs an action and waits on the response from the system. This process is iterated. Imagine that the user performs another action - for example a second click on the action button area - before the system responds. This event should be deleted from the event queue (but it is not). I experienced that the save button does not react properly. I experienced that the area beside the statement field just says "saving" but nothing happens. The only way I am aware to overcome this this "deadlock" is to make a (dummy) edit in the special:MyPage/common.js page. A simple CTR_R (force reload) does not help. Can you please confirm? Thanks in advance! לערי ריינהארט (talk) 00:59, 25 May 2014 (UTC)

Just now, I can't add anything either in Firefox. "Saving" appears and that was it. It doesn't store, nothing helps (reloading, purging, deleting browser cache). It works in Google Chrome, though. --AndreasPraefcke (talk) 08:32, 31 May 2014 (UTC)
Thanks @AndreasPraefcke for confirming the behavior. In order to identify the source for this buggy behavior I used Chrome with an alternate account i18n and noticed that Chrome is much faster. There I use different tools at my /common.js and /vector.js subpages and the selected gadgets differs too.
FYI: At the bottom of the i18n user page you may find (some of) the recent changes related to Javascript changes relevant to Wikidata. gangLeri לערי ריינהארט (talk) 10:25, 31 May 2014 (UTC)

Instance of human(s)

I've found (and have flagged in the Wikidata game) a number of entities which represent duos or groups of people, and which are tagged as "instance of human". Is this acceptable? Should they instead be marked as something like "instance of duo"? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 17:39, 30 May 2014 (UTC)

You should use "instance of:group of humans (Q16334295).--Micru (talk) 22:43, 30 May 2014 (UTC)
Or one of its subclasses such as duo (Q10648343) View with Reasonator View with SQID , see
Classification of the class group of humans (Q16334295) View with Reasonator View with SQID
For help about classification, see Wikidata:Classification.
parent classes (classes of items which contain this one item) 
subclasses (classes which contain special kinds of items of this class)
<group of humans> on wikidata tree visualisation (external tool)(depth=1)
Miga external tool (does not work in Firefox) 
listing of subclasses, number of super and subclasses, properties of the instances: <group of humans> on Miga
Browse the classes starting from this one 
Browse classes from < group of humans > with Taxonomy Browser

now in the game we had citizenship but this is a very delicate work for example François L'Huillier de Hoff (Q3085042) is suggested as France (Q142) but in those days we had First French Empire (Q71084) or Second French Empire (Q71092) --Rippitippi (talk) 02:19, 31 May 2014 (UTC)

Rippitippi, that is wonderful, I didn't know that so many subclasses existed! I hope we can get some sub/upper class tree discovery method soon other than going to an external site every time...--Micru (talk) 07:28, 31 May 2014 (UTC)

Starting Help:Classification

After creating {{Classification}}, I thought that we're ready to start an help page of the same name.

I propose a 2 main parts plan :

  1. first the class/token relationship
  2. second the class of classes level with several examples :

I think this deserves to be community validated, that's why I talk here first and do not start the page right away. Any thoughts and comments ? TomT0m (talk) 15:58, 31 May 2014 (UTC)

Two items to merge

N'Tenetou and Tenetou are the same place. Djembayz (talk) 17:40, 31 May 2014 (UTC)

It looks like they're different. N'Tenetou is in Bougouni Cercle (Q895013) and Tenetou is in Sikasso Cercle (Q1537714). The coordinates are also different. --AmaryllisGardener talk 17:50, 31 May 2014 (UTC)

manual de-capitalization of labels

Sorry if this has been asked before. Is there a tool/gadget for quick de-capitalization of labels in Wikidata interface? --Papuass (talk) 13:58, 29 May 2014 (UTC)

Maybe can be useful a new Wikidata game :) @Magnus Manske: --ValterVB (talk) 19:12, 29 May 2014 (UTC)
Can do, if there is a "likely subset" of items, a heuristic that would return a small (say, <1M) number of items, each with a >30% chance of needing "de-capitalization". Otherwise, JavaScript is your friend ;-) --Magnus Manske (talk) 21:32, 29 May 2014 (UTC)
@Papuass, @Magnus_Manske Hi! First you should get familiar by making manual changes. I identified more classes of items:
a) disambiguation pages should preserve the first character as a capital
b1) encyclopedia (Q5292) View with Reasonator View with SQID where only de German and related languages / dialect including frr Nordfriisk and stq Seeltersk are starting with a capital character, hy needs de-capitalization while ka does not
b2) a list to test: user:Gangleri/sandbox/index/extended periodic table uselang=el versus today's uselang=be example: fermium (Q1896) View with Reasonator View with SQID
c) Pitcairn Islands (Q35672) View with Reasonator View with SQID a proper noun starting with a capital in all languages using capitalization (ka, ar, he etc. do not)
d) Inuit languages (Q27796) View with Reasonator View with SQID where en and variants including some creole languages as pih Norfuk / Pitkern are using a capital for the words derived from proper nouns
note: there is an ontological mismatch between languages in item labels and in the content of the WMF articles: some are talking about one language some about a language family (Q25295) View with Reasonator View with SQID / language group / variety (Q3329375) View with Reasonator View with SQID
e1) mixtures Pitcairn Anthem (Q1209767) View with Reasonator View with SQID probably better Pitcairn anthem
e2) mixtures https://www.wikidata.org/?diff=prev&oldid=134486909#top reverted by 콩가루
e3) Indo-European languages (Q19860) View with Reasonator View with SQID in German and relates dialects / variants: indogermanische Sprachen
e4) https://www.wikidata.org/?diff=135376993&oldid=126602509#top example for changes behind the first character at Indo-European languages (Q19860) View with Reasonator View with SQID
note: descriptions and aliases may need a de-capitalization too Regards gangLeri לערי ריינהארט (talk) 00:31, 31 May 2014 (UTC)
special cases:
i) https://www.wikidata.org/?diff=128374437&oldid=128374370#top in nl Dutch IJ is the two byte / two character coding of a "starting character"
ii) de-capitalization is an UTF-8 related issue / task / job
iii) in addition el Greek is using accents; their encoding needs to verified; see lang: el όλμιο (Q1846) (view with reasonator)) for holmium (Q1846)
gangLeri לערי ריינהארט (talk) 02:20, 31 May 2014 (UTC)
FYI: please see user:Jitrixis/nameGuzzler.js made by user:Jitrixis which handles a related topic
see also MediaWiki:Gadget-MainLangFirst.js created by Reza1615
P.S. Is there a place about multilingualism where this paragraph van be moved? If it is hidden in the archive it was a waste of time ...
P.P.S. |stq-0|war-0|wo-0|sma-0|sjd-0|sjt-0|smn-0|smj-0| are valid babel values; they are used at the "incubator" see en:Sami languages#External_links
regards gangLeri לערי ריינהארט (talk) 08:59, 31 May 2014 (UTC)

Sorry for wrong revert. But I re-reverted my wrong revert: https://www.wikidata.org/w/index.php?title=Q5866990&diff=next&oldid=134486909 --Konggaru (talk) 10:43, 31 May 2014 (UTC)

pattern related de-capitalization: Please see OmegaWiki:DefinedMeaning:5684 - Dutch and Dutch (Q7411) View with Reasonator View with SQID . There might be more patterns different then DefinedMeaning:1164 - енциклопедия and encyclopedia (Q5292) View with Reasonator View with SQID or DefinedMeaning:1144260 - brothel and brothel (Q131295) View with Reasonator View with SQID . I thing we should identify the patterns suitable for automated de-capitalization first. Gangleri לערי ריינהארט (talk) 15:43, 31 May 2014 (UTC)
proper names including brands as "Contergan" and brand type names as Interlingua (Q35934) View with Reasonator View with SQID are written with a starting capital in all languages using capital characters. לערי ריינהארט (talk) 20:41, 31 May 2014 (UTC)