Wikidata:Project chat

From Wikidata
(Redirected from Wikidata:PC)
Jump to: navigation, search
Shortcut: WD:PC
Wikidata project chat
Place used to discuss any and all aspects of Wikidata: the project itself, policy and proposals, individual data items, technical issues, etc.
Please take a look at the frequently asked questions to see if your question has already been answered.
Requests for deletions and mergers can be made here.
IRC channel: #wikidata connect
Start a new discussion
On this page, old discussions are archived. An overview of all archives can be found at this page's archive index. The current archive is located at July.




for permissions

for deletions

for deletion

for comment


Javascript devs : an entity selector for Wikidata discussion[edit]

@Tpt, Magnus Manske, Lydia Pintscher (WMDE), YMS: (@...) Hi, I begun (sketched actually) the dev of a javascript gadget on test Wikidata to insert an entity in wikitext but I can't find the time to develop it right now. It's upsetting me because I think it's important, so I through a bottle here, if someone think it's important we can make this effort collaborative :)

I thought I would replicate the search or entity selector in item pages in the wikitext edit gadget. Except validation would insert something like {{subst:Q}} in the Wikitext. One other solution is adding something to the search widget to insert in wikitext instead of going to the entity page (Lydia could this be a good query to add a ticket or a feature of the future UI ?)  – The preceding unsigned comment was added by TomT0m (talk • contribs).

Thematic mapping of geographic statics, US and World[edit]

My interest lies in statistical phenomenology that can be shade mapped by states in the US and countries around the world. Such phenomena as demographics, crime, opinion, health, etc. -- any phenomena that has statistics by area for which the statistic can be listed, statistically summarized, then I want to convert the statistical phenomena into shade maps of the geographic areas they represent and possibly apply bi-variate statistics on the geographic distribution such as center of gravity, etc.

I am new to Wikidata and want to get an understanding of the statistical data you are assembling and if data mapping is already produced or would be amenable to adding such map graphic products to your data series.

I am having trouble getting connected into Wikidata. I contribute financially to Wikipedia.

Please help me get connected and directed appropriately.

Charles Barb

Forth and back conversions of items between class and instance[edit]

I changed instance to class [1] then had a look into the page history to obtain a link for my edits and ..., just on 2014-07-01 User:Humatiel did the opposite [2].

The editor did create a Wikidata:Database reports/Constraint violations/P279, since the item was target of "subclass of" since 2013-08-06 [3], created by User:TomT0m.

There are still 6000 unrooted subclasses listed at Wikidata:Database reports/Item classification. Can a bot inform editors that unrooted an item, i.e. that created a P279 or P31 constraint violation? Tamawashi (talk) 13:13, 15 July 2014 (UTC)

The difference between "subclass of"/"instance of" is confusing many users. After many discussions the conclusion has been that "instance of" represents something material of which there can be only one ("single exemplar of"), whereas "subclass" represents something abstract of which there can be many ("subcategory of").
I don't know how to make that more clear, but the current labels don't convey their meaning, otherwise there wouldn't be so many wrong uses. Any suggestion about how to convey better the difference?--Micru (talk) 13:47, 15 July 2014 (UTC)
@Tamawashi, Micru: Develop and make Help:Classification more visible ? It's a really important page imho. Even people like @Emw: seem to disagree with me on some topic like the type of the isotope class, after many discussions. I, (I speak for myself) am pretty confident I am at a point where I sorted this out and the instance/class/class of class scheme works pretty well for me and is really useful to model things in Wikidata, see many examples on Wikidata talk:Item classification of why I think this works and is useful. But we need to be really clear on that and not make that discussion over and over again. I'd like to make Help:Classification approved by community when it is ready for beginners and reasonably complete.
A constraint cannot do the job of explaining that to a user. @Micru: The abstract/concrete view is probably confusing. There is often several level of abstration : we divide concrete things like atoms in chemistry in classes of concrete things like "oxygen", who is a class of atom, and "oxygen-18", who is a class of atoms to, a class of oxygen more precisely. But we also use more abstract levels : isotopes for example regroups classes, not instances : we can say that oxygen-18 and oxygen-16 are isotope, but oxygen-18 is not a subclass of isotope. The definition of subclass is not "things that links abstract concept", it's more precise. TomT0m (talk) 06:33, 16 July 2014 (UTC)
@TomT0m: It doesn't seem that you have sorted out so well the distinction, otherwise you wouldn't be saying that "oxygen-18 is not a subclass of isotope". My reasoning is as follows:
  1. There are things that happen in nature by themselves
  2. By being in sensory contact with differentiable things (instances) we create representations of them
  3. By observing repeated occurrences, we create models of these things (classes)
Now I apply the same reasoning to "oxygen-18"
  1. There are oxygen-18 atoms that happen in nature by themselves
  2. By being in sensory contact, in this case through devices, with distinct oxygen-18 atoms (instances) we create representations of them (instance:oxygen-18 [one atom])
  3. By observing repeated occurrences, we create models of oxygen-18 (class:oxygen-18 [represents all atoms of oxygen-18 that may exist] )
Usually I see very often the inverse reasoning, specially from IT people, who put the class first and that is totally meaningless. Either we agree that if we want to model reality, reality has to come first, or we we'll be circling around this till the end of the times.--Micru (talk) 08:27, 16 July 2014 (UTC)
@Micru: : I have a discussion going on with Emw on this very page about this. See in particular about the proposition of query by Emw built with the <isotope> class and my answer on why it does not make sense : if <isotope> is a superclass of <isotope-18>, then the <atom> class and the <isotope> classes are indistinguishable for a reasoner, he can substitute one with the other. And I think I took part in pretty much every discussion about class/instance relationship and how they are implemented in the semantic web. I argumented and proposed to use the metaclass concept : <isotope> is a class of concepts whose instances are classes such as <isotope-18>, who exists as an entity in the human conceptualisation. What a human call an isotope is a class of atom. When we say oxygen-18 is an isotope, in common language, we don't say every oxygen-18 instance is an isotope. Otherwis the whole isotope concept is useless compared to the atom concept. The <isotope> concept by itself is useful because all atoms in an isotope class have the same numbers on nucleons. We have an isotope concept because this property is shared by a big numbers of important classes of atoms, we can regroup in a set : the set of all isotope classes, which is NOT the set of all atoms at all. I backed up by citing the french Wikipedia's definition of the <isotope> concepts which says exactly that : an isotope is a class of atoms who share the same numbers of nucleons. In short, in Wikidata we do not model reality only, we also model how human models reality. Which is a real thing too somehow. TomT0m (talk) 09:12, 16 July 2014 (UTC)
@TomT0m: if the "the <atom> class and the <isotope> classes are indistinguishable for a reasoner" it is not because the classing, it is because they do not contain enough information to make them distinct enough. We need to translate the definition in human language (an isotope is a class of atoms who share the same numbers of nucleons) into ontological language (for instance, like this, but it could be done better). Anyhow let's keep the atomic discussion in one place.--Micru (talk) 09:41, 16 July 2014 (UTC)
@Micru: No it cannot. The best stuff we have for this is to class not atoms, but substances, especially a pure substance is a substance made of only one type of atom. An isotopically pure substance is a substance made of only one isotope of one chemical element (as a class of atoms). But this concepts still do not map the isotope concept as all the kinds of isotopes types of all the atom types. But feel free to join the discussion on the query: here is my answer to Emw. TomT0m (talk) 10:18, 16 July 2014 (UTC)
@TomT0m: Substances (class) have atoms (class) as parts, and an isotopically pure substances (class) have atoms of the same isotope as part (class), but ok, I'll read the (now long) discussion.--Micru (talk) 10:40, 16 July 2014 (UTC)
I followed exactly the logic that Micru presented. In programming languages, it is usual that an instance being an object and a subclass being a subcategory. For example, an apple is an object/instance of fruit, and fruit is a subcategory/subclass of the kingdom Plantae (plants). That's why I changed it, since wikidata is to be human and machine readable. Regards, Humatiel (talk) 22:49, 15 July 2014 (UTC)
Humatiel, an instance it is not only an object, it is *the only* object that exists of something in the universe at a molecular level. One indeterminate apple is a "subclass of fruit", this apple that I have on my desk is an "instance of the class apple". Everything that is abstract (awards, movies, albums, concepts, etc) cannot be instances because there can be many of the same.
In Wikidata there is this confusion all over the place, even myself I had it until very recently, that is why I was suggesting to change the property labels to something that is more self-explanatory... but I don't know what.--Micru (talk) 23:14, 15 July 2014 (UTC)
@Micru: I french we have nature for instance of (P31) as a default label. I guess the subclass of in english would be better with something like special kinds of. How sounds "the nature of Micru is Wikipedian" in english, and "wikipedians are special kinds of humans" ? The plural for subclass of (P279) seem to me an idea worth investigating. Though instances of seems not really good, how reads Wikipedian instances of human (just random thought letting it going). Or <Wikipedians> are all <humans> with <Micru> example of <Wikimedian> ? (I'm not very aware of how natural sounds the instance word in everyday english. <Micru> yet another <Wikimedian> (no offense :)) ... <Apple> yet another <Fruit> works too. Hard to get rid of the is a easily substituted to are all in common language. <Apple> is a <fruit> works as well as <apples> are all fruit. <Apple> kind of <Fruit> or <Apples> are all <fruits> works, but can we add plural marks to all classes ??? (I like the idea, not sure it is a good one). TomT0m (talk) 07:40, 16 July 2014 (UTC)
I would go with "@TomT0m: <one exemplar of> human" and "apple <all exemplars are> fruit". More opinions on this?--Micru (talk) 08:27, 16 July 2014 (UTC)
@Humatiel: a fruit is part of a plant and not a subclass of the kingdom Plantae. BTW: there are some substantial differences between an object orientated programing language and OWL. --Succu (talk) 06:46, 16 July 2014 (UTC)

@Humatiel, TomT0m, Micru, Zolo, Emw: - To my understanding, in the tree "abstract object" there are no instances. And probably "physical object" is a tree that contains all instances. That means, two little queries would reveal already some errors in classification. There are still ~5900 unrooted classes listed by autolist2. If you could help here :-). Numbers are going down, but very slowly: Wikidata:Database reports/Item classification. And sometimes I found it hard to find an upward class, that is not "entity". Tamawashi (talk) 11:28, 16 July 2014 (UTC)

@Tamawashi: That is more or less correct, and it matches the formal definitions of 'continuant' and 'process' that Emw brought with this paper. I can help rooting, but in some cases it will be useless (as you have seen first hand), since some concepts are best defined in other terms.--Micru (talk) 12:03, 16 July 2014 (UTC)
There are individual things which are 'instance of' a class of things.
In some cases however a class is an instance of a 'type of class' as well as being a subclass of a larger class.
Oxygen-18 a class of millions of individual atoms each with 10 neutrons and 8 protons. This is a subclass of:oxygen atoms.
The class of Oxygen-18 atoms is also an 'instance of:isotope' - a special type of class.
The class of Oxygen atoms is on the other hand a subclass of:atom but an instance of:element - another type of class.
Note that 'isotope' is not a 'subclass of:element'. They are both 'subclass of:ways of classifying atoms'.
Other 'types of class' that classes can be an instance of:
  • Moby Dick subclass of:book, instance of:novel
  • Ford Model T subclass of:car, instance of:model.
At least that is how I see it. Filceolaire (talk) 21:53, 17 July 2014 (UTC)

"The class of Oxygen-18 atoms is also an 'instance of:isotope' - a special type of class." I do not agree with this. An instance has to be either a material thing (continuant) or a unique process in order to be considered "instance", in other words it has to be something that exists "here" and "now". All the rest are class relationships, either as "subclass of" or "superclass of", that we don't use. The relationship between "the class of all Oxygen-18 atoms" and "the class of oxygen isotope (classification of atoms)" is a subclass relationship.
I also don't agree with the book/novel, car/model examples. It was a mistake that we did on the Books task force that is spreading all over the place. It should be:
  • Moby Dick subclass of:novel
  • Ford Model T subclass of:car model
Of course the terms "subclass of"/"instance of" are confusing, that is why it would be better to change them. Possible synonyms for "subclass of": "is a", "nature", "all exemplars are", "all instances are". Possible synonyms for "instance of": "unique exemplar of", "real-life specimen of", "mass-energy occurrence of".--Micru (talk) 14:55, 18 July 2014 (UTC)
@Micru: You can't stay with an argumentation as poor as it was a mistake witout begin solid on this. You disagree with everybody one way or another with 3 people at least here : with Emw (talkcontribslogs), who could be your best friend here because he do not like metaclasses, @Filceolaire:, me ... We discussed this a lot, you will have to do better than just saying everybody else is wrong. TomT0m (talk) 13:50, 19 July 2014 (UTC)
@TomT0m: After the discussions with you, with Emw, and with others I have now a more coherent view that I had in the past, for that I am very grateful that you have helped me with great patience in our long discussions. I say that it was a mistake because the method used back then doesn't hold to deep scrutiny and makes the properties p31 and p279 indistinguishable from one another, as I have been argumenting here and on the other conversation below. Funny that you use argumentum ad populum (Q251695)... history is full of examples where that was a bad idea. For instance, that many people agreed that the Earth was the center of the universe doesn't make it more true, just means that there was an agreement and a legal body to keep that acceptance. In this context I do not believe in "right" or "wrong", just in approximations. If there is a general consensus then of course I will let it flow as an "accepted approximation", I am just pointing out that there are better ones. Please, do not take my disagreement personally since I consider you, Emw, and all others, allies in the mission of bringing the sum of all knowledge to all sentient beings. If you consider that this conversation is straining, let's stop it now. Sometimes it is just better to wait.--Micru (talk) 14:52, 19 July 2014 (UTC)
If you consider that this conversation is straining, let's stop it now. Sometimes it is just better to wait. That's why I went meta on the discussion and stopped argumentation at that point :) TomT0m (talk) 07:06, 20 July 2014 (UTC)

I agree and disagree with things Micru says above. Foremost, I think changing the labels of P31 and P279 from instance of and subclass of to something else would be a bad idea. Those labels have been established for over a year, have been used in academic publications and media coverage about Wikidata, and give an immediate, glanceable, conventional visual hook that illustrates a fundamental distinction used throughout the Semantic Web and philosophy at large to describe the nature of things. Changing their aliases is OK, changing their descriptions with care is OK, but we should not change their labels.

I agree with Micru that "Moby Dick subclass of novel" is preferable to "Moby Dick instance of novel". However, I disagree with the statement "Ford Model T subclass of car model". Cars models are like biological taxa in that they are not physically instantiated. Stating otherwise is a subtle mistake with significant consequences. For example, if we say "Chevrolet Malibu (Q287723) subclass of car model", then that entails the incorrect statement "The Peekskill Meteorite Car (Q7756463) instance of car model". The Peetskill Meteorite Car is not a car model, it is a car (specifically, an instance of a Chevrolet Malibu). Although Wikidata covers far fewer instances of car than instances of human, we should keep the application of instance of and subclass of consistent.

And that is why the statement "oxygen-18 instance of isotope" is problematic. Oxygen-18, cars, and humans are all types of material entity. Oxygen-18 is a type of isotope, car is a type of vehicle, and human is a type of Homo. Stating "oxygen-18 instance of isotope" creates an inconsistency in how we apply instance of and subclass of in cars and humans and all other physical things. The fact that Wikidata will likely never have an item about an instance of oxygen-18 does not mean we shold state that oxygen-18 is an instance. instance of (P31) is not simply what you use at the bottom of Wikidata's concept hierarchy. For at least physical things, outside rare and straightforward usages of explicit metamodeling, instance of should only be used on items that have a unique location in space and time. Emw (talk) 15:34, 19 July 2014 (UTC)

@Emw: I am glad that we are approaching postures. You might be right that it might no longer possible to change the labels of p31 and p279 without causing major disruption that we might want to avoid, but it worries me that some languages have departed from the canonical names and now they have come to represent different things (in German p31 is now "is a"), perhaps it will be enough making sure there is no confusion by explaining it better, or creating a tutorial.
Just a minor correction regarding your car example. You say: if we say "Chevrolet Malibu (Q287723) subclass of car model", then that entails the incorrect statement "The Peekskill Meteorite Car (Q7756463) instance of car model". I think we should examine deeper what a "car model" is. If we assume that a car model is a blue print for manufacturing an automobile with shared characteristics, then the statement "The Peekskill Meteorite Car (Q7756463) instance of car model" is not incorrect, because we have a blueprint and from there we generate something that is an automobile. Of course the statement is less accurate than saying "The Peekskill Meteorite Car (Q7756463) instance of Chevrolet Malibu (Q287723)", but still right. Same case as the example you gave me "Pluto (Q339)<part of>universe (Q1)", which is not wrong, just "less right".--Micru (talk) 16:27, 19 July 2014 (UTC)
@Micru: : this depends on the definition we use for car model. Of course we can take a car, scan it and make a copy with a 3D printer, and our copy has a model : the original car. But it's not the common definition car manufacturers uses for car model. We can define the <car model> item such that its correct wikidata use is a car model in the car manufacturing industry sense, and we get rid of this correct in a loose sense notion that is ... Uncommon. Of course if there is a use for this definition we can create an item with this definition, but let's be precise and say that car that are 3D printed with some scanned car and car model in the car industry sense are different item. The vast majority of car models in Wikidata are on the second sense. That's indeed a compelling case for class classification. TomT0m (talk) 07:52, 20 July 2014 (UTC)
@TomT0m: What is a car model in the car industry sense? Put the definition on the operation table so we can dissect it ;) --Micru (talk) 09:57, 20 July 2014 (UTC)
@emw:. My understanding is that O-18 is general would be considered an isotope of oxygen but an individual molecule of O18 would not. That sounds remarkably similar to the Ford T case.
As stated in #Issue with "instance of" for text, I would argue for a similar analysis with novels. Novel is a subclass of text, and a text is, roughly speaking, an (immaterial) string of words and Moby Dick is best viewed as an instance of novel starting with "Call me Ishmael." It can be materialized in a book, an audio recording whatever, but those are instances of book, audio recording etc, not instances of novel. Of course defining a text as a string is a very crude approximation, but a definition of text would have to do with the us of words, the existence of a plot, etc. These are not features that a material object can have It implies that the class "text" is disjoint from the class "material object".--Zolo (talk) 17:35, 19 July 2014 (UTC)--Micru (talk) 14:30, 20 July 2014 (UTC)
Agreed with Zolo on the similarity with the <car model> and <isotope> discussions, that's why the instance/class/metaclass modeing 3 layer scheme I push other and other again make sense. "oxygen-18 instance of (P31) isotope" creates no inconsistency if we make clear that isotope IS NOT a class of physical objects itself, which can be made explicit in the ontology by a disjoint with statement beetween the <physical entities> class and the <class> class, so it is absolutely not a logical inconsistency. It's not an inconsistency with the Wikidata usage if we stop pushing the Token/class philosophical principle to newbies and start pushing a Token/class/metaclass principle which actually helps a lot into understanding how to model the world. Using subclass of (P279) as the only relationship beetween classes of tokens (I don't say instances intentionally) is a mistake and logical inconsistency source and should as a result not be pushed to users. So, Emw, I think you should revise your way of thinking and starts thinking not car models as weird exceptions, but as the general case. When a scientific theory starts to have one exception scientists can be fine saying OK, that's a weird exception. When there is a second, a third and so on they'll start thinking they need a Paradigm shift (Q689971) (View with Reasonator) (paradigm shift) in their theories. I think token/class principle is such a paradigm who put our mind in a bottle, but everything is simpler if we break that bottle and shift to the token/class/metaclass paradigm. I provided several usecases for which this paradigm is useful, for me it's a compelling argument it's a solid paradigm, and more consistent than the "token/class with exception" you are proposing, who is unsatisfactory : if we got a 3 concepts principle which fits almost all wikidata, wrt. a 2 concept principle with 30 exception, the first has 3 concepts, the second 32 concepts. Ockham rasors says "the first one is better". TomT0m (talk) 07:52, 20 July 2014 (UTC)

Zolo, an actual atom of oxygen-18 is an instance of an isotope. Instances of isotopes is what is measured in e.g. isotope-ratio mass spectrometry. Thus the statement "oxygen-18 instance of isotope" is indeed inconsistent with how we use that property to describe things like humans, cars, and other physical things. It is like saying "human instance of Homo", not "human instance of taxon".
Some domains talk of types of types of things. Biological taxonomy is one, but chemistry does so less, certainly in the way TomT0m envisions:
Which is easier for users to understand, the "atom type class" model proposed above or the one without a metamodeling layer, as used in widely-cited, professionally curated chemistry ontologies like ChEBI? I do not think pervasive explicit metamodeling will ameliorate users' unfamiliarity with instances and classes. I think it will bewilder them.
Instance of and subclass of can be used for explicit metamodeling in a way that accommodates type-token distinction. The usage seen in biological taxonomy is an example. However, it is easy to do metamodeling in a way that breaks that useful distinction, as illustrated in statements like "hydrogen instance of chemistry element" and "oxygen-18 instance of isotope."
Doing so compels us to make absurd statements like "isotope is not a class of physical object" as TomT0m does. As any chemist or high school student who has completed a course in chemistry will tell you, isotopes are physical objects. Stating "oxygen-18 subclass of isotope" rather than "oxygen-18 instance of isotope" consistently applies classification properties in a way that does not compel us to paraphrase away basic scientific facts. Emw (talk) 14:13, 20 July 2014 (UTC)
@Emw: I share your opinions to some extent. I think we should not use "instance of" for cross-domain relationships, but we can attempt to cross it with a different property, even if that departs from ChEBI. The way I envision it is like this:
I don't think using metaclasses it makes it more complicated if it is explained properly (i.e. not using technical jargon).--Micru (talk) 16:27, 20 July 2014 (UTC)
Micru, I think it is essential for Wikidata to use vocabulary from the wider Semantic Web. The connection between Wikidata, the Semantic Web and ontology is that "class" is synonymous with "type" or "concept", as outlined in e.g. The Role of Foundational Relations in the Alignment of Biomedical Ontologies. The conventional term for "classes of classes" like "taxon" or "car model" is metaclass. Implying that classes are not types or concepts as your graphic above does would misalign Wikidata with most literature relating Semantic Web vocabulary to that from ontology. Styling instances as "phenomena" and classes as "substance" is also highly idiosyncratic and would lead to even obscurer philosophical wanderings than those we currently see on Wikidata.
Metamodeling in the domain of chemical elements seems generally extraneous and unhelpful. What makes "atom" a class and "isotope" and "element" metaclasses? How do "chemical substance" and "physical object" fit in; are they class or metaclasses? As explained above in my reply to Zolo, isotope is clearly a class of physical object and individual atoms are also isotopes. Why should I have to include nature of as outlined in your graphic (note: as presented it should be has nature) rather than use subclass of? What is the precise definition of has nature -- in terms used by the Semantic Web, i.e. OWL? Precisely how would it relate to P31 (i.e. instance of, rdf:type) and P279 (i.e. subclass of, rdfs:subClassOf)?
I think it would be far simpler to say something like "oxygen-18 subclass of isotope of oxygen" for such classes of isotope. We could then say "isotope of oxygen subclass of oxygen, isotope". Alternatively, we could say "oxygen-18 subclass of oxygen, isotope", and consider "isotope of oxygen" to effectively be a query result. Both of these approaches adequately model isotopes in a way that is consistent with major existing ontologies and does not pose compatibility issues with Semantic Web standards. Emw (talk) 18:09, 20 July 2014 (UTC)
Brilliant paper, Emw! It brings many insights and I dearly recommend Zolo and TomT0m to read it. I will do it again, and again. On a first read I found some gems:
  • inst: defined as bound to a specific time and space. An instance can be related to its class (a instanceOf b)
  • class: defined by its instances, which means that they can change in time. A class can be related to its upper class (b subclassOf c)
  • set: a timeless entity, idealized class, not bound to time and space. A set or a class can be related to a set (b is_a d)
The definition of what I called "nature" is what they call "is_a". Please, do notice the differences with "subclassOf" as cleverly explained in the section "Classes vs. Sets". I can go further and say that "is_a" is equivalent to "type" (well, they also say it in the paper), and "taxon rank" is a subproperty of "is a".
Coming back to my drawing, yes, you are right, it is wrong, I'll try again:
That is my interpretation of the provided paper, with the nomenclature stated in the paper. True that for atoms it looks quite redundant (although here we wouldn't use classes), however it models accurately the relationship between classes-metaclasses and between metaclasses themselves. To use subclassOf for both purposes is missleading and brings to many confussions and long discussions as we are seeing. I hope we can all reach the same interpretation and consider it both acceptable and practical.--Micru (talk) 23:18, 20 July 2014 (UTC)

Micru, that paper formally defines is a, a foundational property in major ontologies like the Gene Ontology (GO), ChEBI and others. It is referenced as the basis for is a and part of as discussed in Relations in Biomedical Ontologies, a very widely-cited, classic paper in modern ontology. That paper establishes the link between "is a" and the language of the Semantic Web: "A has_subclass B = [definition] B is_a A." -- in other words, "A rdfs:subClassOf B" equals "A is a B". This equivalence between BFO/OBO's is a and RDF/OWL's rdfs:subClassOf is noted in the draft BFO 2.0 release documentation and clear from RDF/OWL exports of GO, ChEBI and other ontologies among the Open Biomedical Ontologies.
The relation between classes and sets is a recurring topic in these discussions on Wikidata. Classes are treated essentially as sets in the Semantic Web. The foundational subclass of property is even symbolically represented with the symbol as "subset of" -- i.e. ⊆ -- in the Interpretation of Axioms and Facts in the OWL Direct Model-Theoretic Semantics specification. The relevant OWL specifications address this by noting that a class is distinct from the "extension" of that class.
It is interesting how your subsumption hierarchy (the "class tree") of atom in your interpretation of the FMA paper is virtually identical to that used in ChEBI. See e.g. the "graph view" visualization in the ChEBI model of oxygen-18 at Of course there is no mirrored subset tree, because is a is equivalent to subclass of.
The ChEBI ontology does not explicitly account for some things we may want to, e.g. isotopes. One suspicion is that this is because all instances of atom are also instances of isotope, and vice versa. However, we might be able to model the notion that "oxygen" itself is not an isotope -- while preserving consistency with ChEBI and those other widely used ontologies -- with an approach I outline above. Emw (talk) 01:35, 21 July 2014 (UTC)
@Micru: Thanks for the link. Some notes:
  • It states that nothing can be both a class and an instance, and that conflicts with the latest OWL version. The current theoretical view appears to be that we should distinguished between the statements about the item seen as an instance of X and those of the item seen as a subclass (seen as a instance of species, gorilla beringei is an endangered species, seen as an subclass of animal, it is hairy).
  • Nothing in the theoretical part of the article says that an instance must be localizable in time and space. They indeed allude that they should in the conclusion, and for highly empirical topics like biology, it is certainly true that most instances have a location, but that would be an empirical fact, or a methodoligical advice, not a formal imperative (yes
  • You are misreading the paper, it does not say that classes are defined by its instances, quite the opposite: contrary to a set, a class "survives the turnover in its instances".
  • I like their idea that contrary to a set, a class does not have 2n subclasses (actually it think we can say that "subset" is a subclass of "class" with some special features, except that it would sound a bit odd to include the empty set and singletons in the list of subclasses). --Zolo (talk) 07:56, 21 July 2014 (UTC)
@Emw:. Ok, if individual molecules are isotopes, then clearly it is fine to call 018 and subclass of isotope. --07:56, 21 July 2014 (UTC)~
@Emw: first of all some term equivalences:
* what OWL calls "extension of a class" = what FMA calls "class"
* what OWL calls "class" = what FMA calls "set"
For the particular case of chemical elements I agree with you that there is an "almost" 1:1 relationship between class-set (real atom vs conceptual atom), but not making this distinction is problematic for several reasons. The main one is the superposition of "natures" that an element has, which collapse into one as soon as we try to access experimentally (adding instances to) one of them (cf. w:wave function collapse). Only by avoiding to use the "isotope" set, they are able to keep an equivalence between "class" and "set". It is not a lie, they just don't consider all the range of possible truths.
In our particular case, it would be perfectly fine to model chemical elements considering an equivalence between subclassOf = is_a because we don't instantiate them. For the classes that we do instantiate then metamodelling (making a distinction between class and set) is useful (taxonomy, cities, literary works, etc). But that of course forces us to juggle with properties, when by using two distinct properties (FMA-subclassOf, FMA-is_a), would remove all obscurity from our model.
The other reason to make the distinction is to avoid mixing up "extension of a class" and "class". If we use OWL-subclassOf for both cases we might arrive to the weird conclusion that a person might have the permanent nature of being an actor (!). That means that we accept instantiating permanent concepts as impermanent instances, with all the inconsistencies and headaches that it brings.
For these reasons my recommendation is to abide to the FMA definitions (instanceOf=phenomena, subclassOf=OWL-"extension of a class", is_a=OWL-"class"). This will guarantee compatibility with all existing and yet to exist ontologies, because we'll be able to cast both subclassOf and is_a to the same external ontology property (if they have so defined), or to distinct properties (if they have so defined).
@Zolo: They refer to "SNAP and SPAN: Towards Dynamic Spatial Ontology". Basically: SNAP=continuants (permanent) and SPAN=occurrants (impermanent). According to this model, yes, you can instantiate a permanent class (set), but not as an impermanent entity!--Micru (talk) 08:26, 21 July 2014 (UTC)
@Micru: That is true, but that seems to work well as long as we do not need to assume that everything is a SPAN and SNAP.
To keep things in plain language:
  • entity:
    • unlocalizable entity (concept)
    • localizable entity
      • event
      • material object.--Zolo (talk) 10:03, 21 July 2014 (UTC)

──────────────────────────────────────────────────────────────────────────────────────────────────── @Zolo: Your tree puts again together sets and classes (the "entity" you put on top cannot be both "abstract" and "real"), and as showed before that is problematic. Here there is another one:

  • unlocalizable entity (concept, set)
    • qualities: physically identifiable characteristics (shape, color, etc)
    • attributes: abstractedly identifiable characteristics (the "is_a"-ness)
    • entity: the concept that something might exist either as unlocalizable or as localizable entity
  • localizable entity (phenomenon, instance)
    • event
    • material object

@Emw: I'm sorry that I'm not following OWL conventions :( I realize that the definition of entity is self-referential... this is not my personal preference, it is just how nature appears to me... very weird... :-/ I also realize that there are more "unlocalizable entities" that only apply to living beings, but it is better to focus on this first.--Micru (talk) 12:57, 21 July 2014 (UTC)

Micru, is a is equivalent to subclass of, i.e. rdfs:subClassOf. The lead author on that paper you like is also the lead author on Relations in Biomedical Ontologies, and also the head of the team that produces the BFO, which uses that equivalence. Please take another at the first paragraph in my previous comment where I describe how is a equals subclass of in that major school of formal ontology. Please review the BFO class hierarchy, which accounts for the relation of 'entity' to 'material object' and 'qualities'. Let's use conventions from the Semantic Web and not add yet another upper ontology to it. Emw (talk) 12:55, 22 July 2014 (UTC)
@Emw: Which property shall we use to express "extension of a class"? And for metaclasses?--Micru (talk) 15:28, 22 July 2014 (UTC)
Micru, subclass of (P279), i.e. rdfs:subClassOf. Here's the definition in OWL:
The rdfs:subClassOf construct is defined as part of RDF Schema [RDF Vocabulary]. Its meaning in OWL is exactly the same: if the class description C1 is defined as a subclass of class description C2, then the set of individuals in the class extension of C1 should be a subset of the set of individuals in the class extension of C2. A class is by definition a subclass of itself (as the subset may be the complete set).
And reviewing the definition of classes in OWL:
In general classes are used to group individuals that have something in common in order to refer to them. Hence, classes essentially represent sets of individuals. In modeling, classes are often used to denote the set of objects comprised by a concept of human thinking, like the concept person or the concept woman.
The equivalence between BFO's is a relation and OWL's subclass of is used, for example, in the chemistry ontology ChEBI:
All database entries are now 'is_a' classified within the ontology, meaning that all of the chemicals are available to semantic reasoning tools that harness the classification hierarchy... To comply with our goal of increasing interoperability with other ontologies in the biomedical domain, ChEBI has provided a mapping to the upper level ontology Basic Formal Ontology (BFO) (12), version 2.0.
Feel free to verify this by downloading any of the OWL files in and seeing the substitution of all usages of is a used through the ChEBI OBO files with rdfs:subClassOf.
Reviewing further, Role of Foundational Relations formally defines is a, and Relations in Biomedical Ontologies harmonizes that definition with Semantic Web standards from the W3C by clearly defining the equivalence of is a and subclass of. The equivalence is made explicit in the BFO 2.0 release notes. The three aforementioned works are led by the same individual and the latter two are produced by the same group. The equivalence is used throughout the world's most widely used biomedical ontologies.
Nowhere in any of those works is the discrepancy you assert between is a and subclass of alluded to. The ontologies in question that use the equivalence of is a and subclass of in practice, Gene Ontology and ChEBI, are maintained by content experts with advanced degrees and deep experience in chemistry, computer science, philosophy and ontology. If your concern with the discrepancy between sets and classes because of wave function collapse were warranted, don't you think they they would have noticed that, and raised that issue somewhere in a publication? Do you really think that's why they don't explicitly have an "isotope" concept in ChEBI? I suspect the cause was much more mundane. A discrepancy between is a and subclass of would directly contradict what that network of researchers have clearly stated in multiple major publications and spent over a decade building. Emw (talk) 04:29, 23 July 2014 (UTC)
@Emw: the last message by TomT0m about intensional vs. extensional definitions clarifies quite well the class-set duality paradox that was besetting me. Seen under that light, yes, I agree on the equivalency between subclassOf and is_a, not because any researcher or produced bible says so, but because it is possible to reconcile that duality the same way it is done in quantum physics. Any entity has thus an "existential state:indeterminate", that can collapse into "determinate existence" -when seen from the perspective of its instances (extensional view), it can represent "extension of that class"-, or into "determinate non-existence" -when seen from the perspective of its abstract meaning (intensional view) it can represent an ideal set, non-localizable in time or space. It is not necessary to declare this explicitly, because the presence or absence of instances already serves that purpose.
Do you still hold the opinion that aligning our vocabulary with CHEBI, BFO and other major ontologies by changing the label of p:p279 to "is a" would be a bad idea? As TomT0m noted, it feels more natural to to write "<isotope> is a <nuclide>" than with the current label.
it is indeed a terrible idea. Did not you really understood the risk for users to mix instance of (P31) with the is a label ? Micru is a talkative Wikidatian is not
< Micru > subclass of (P279) miga < Talkative Wikidatian >
. TomT0m (talk)
Ha, ha :D Considering that the German label of p31 is "is_a", I take back by words, and I will not keep insisting on this. --Micru (talk) 09:46, 24 July 2014 (UTC)
About CHEBI and their lack of isotopes, why don't we stop our elucubrations and just ask them about it? :) --Micru (talk) 07:47, 23 July 2014 (UTC)
Good idea, let's not forget to include a word about nuclides :) With this concept, we got something very similar to the biological taxonomy usecase for class/instance. TomT0m (talk)
I already sent an email to the BFO-discuss list about the proposed "embodiment of" as a way to bridge abstraction and reality, so I leave contacting CHEBI to you or to Emw. Contact page.--Micru (talk) 09:46, 24 July 2014 (UTC)
I might be able to contact them today or early tomorrow. Emw (talk) 12:47, 24 July 2014 (UTC)

@Zolo, TomT0m: Continuing expanding the reasoning that I was exposing below, I realize that it is very similar to the 3 layer model that TomT0m is talking about:

  1. there is an idealized class of something, that at the same time is an instance of itself (see my example with the natural number "5" below)
  2. there is a class of all real occurrences of that class
  3. there are instances of the class of the real occurrences

So applying it to the case of "a novel"

  1. there is the idealized class of a novel, and the class is abstractedly equal to their instances (all thoughts about a "novel" are both a class and an instance, even if they happen with different material substrate)
  2. there is a class of all real occurrences of the class abstract novel
  3. there are instances of the class of real occurrences of the former

Now the problem is to find what is the relationship between 1-2 (I hope we can all agree that the relationship between 2-3 is clearly class-instance). For 1-2 is not that clear, because it seems that this relationship is neither of instantiation, nor of classing, but of abstraction. 2 is an materialization of 1, 1 is an abstraction of 2. It would be interesting to see if there are precedents of these kind of relationship in the literature (pun intended :P).--Micru (talk) 11:34, 20 July 2014 (UTC)zz

I don't follow you, and I think numbers are a bad example: they are already exists as a datatype in Wikidata, so I think we are fine using properties like length search or cardinality to use them. Otherwise clearly we are fine with the set membership/subset relationships for classes here : Natural numbers are a subset of numbers in the most general sense, "5" is a member of that set, so I'd be happy with
< number 5 > instance of (P31) miga < natural number >
for Wikidata, and
< Natural numbers > subclass of (P279) miga < numbers >
. I don't think it's interesting to go deeper here and enter philosophical or math foundation like number set construction problems.
Otherwise I think it's interesting to keep with instance of (P31) for linking classes to metaclasses because it is user friendly imho : in french, instance of (P31) is labelled nature (by the way Emw the english labels are not really supposed to be a reference here so I'm fine with changing the english labels, there is aliases and definition for precisions) and it's pretty intuitive to say that the nature of Porshe 911 is car model imho. It's a cool entry point, not boring the user with obscure philosophical and maybe even confusing considerations.
It would be interesting to see if there are precedents of these kind of relationship in the literature (pun intended :P) I don't see the pun, we're building something here, it has to make sense, that's all, not following absurdely some sourcable rule that may fit less well our needs. We are taking great care of standards in our discussions and are pretty standards friendly nonetheless. TomT0m (talk) 13:28, 20 July 2014 (UTC)
@TomT0m: The pun was because I was giving an example with a "novel", which is also literature... but yes, I agree with your statement, we have to be consistent mainly internally, because if we try to follow every source, the most we can aspire to achieve is inconsistency.
Ok, so we have metaclasses that are abstract concepts, classes and instances, and your proposal is to use p31 to link both metaclasses-classes and classes-instances... wouldn't that put metaclasses and classes in the same tree? It seems a bit like using "part of" or "subclass of" for everything... valid, but doesn't allow us to do consistency checks as the ones Tamawashi is doing. On the other hand, I like a lot the label "nature" for the relationship between classes and metaclasses, but I would prefer another for the relationship between classes-instances, perhaps "occurrent of", "phenomenon of" or even keep "instance of"?
If numbers are such a bad example, why do you bring it from another subsection? :)--Micru (talk) 14:30, 20 July 2014 (UTC)
Micru, "occurrent" is a direct subclass of entity in BFO, the upper ontology used in the world's most cited ontologies. It translates roughly to "event"; it is disjoint with "continuant", which includes things like material and immaterial entities. A complete rendering of BFO's class hierarchy is shown here. Thus changing instance of to occurrent of wouldn't be ideal. Changing it to phenomenon of would make Wikidata very idiosyncratic in the Semantic Web. The label instance of makes it clear that that property is based on Semantic Web conventions and the description logic that undergirds it. These disciplines use one property for instantiation -- instance-class classification -- and one property for subsumption -- class-class classification. It also makes it clear that the subject is an instance, whether that instance is an occurrent, a phenomenon, a human or a car. Emw (talk) 15:00, 20 July 2014 (UTC)

@Emw, Micru:

I lost some part of the discussion, but I'm not sure we are fine with some useful definitions of set theory : the so called set intentional definition concept : a set is defined by a property of their instances. This is an equivalent to class expressions in OWL : a class is then defined not by its instances, but by properties of its instances, see Intensional definition (Q1026899) (View with Reasonator) .

This notion is useful to understand my proposition (which is not an invention of mine btw, it's a used definition) because it fits : clearly <Oxygen-18> and <Oxygen> are both classes of atoms. But what makes one an element-class and the other an isotope-class is that we use different criteria in the intensional definition : in the element classes like <Oxygen> or <Hydrogen>, we use the property atomic number of atoms in the intensianal definition formula (class expression in OWL), and in the other one we use the number of nucleons of the atom. Now imagine we want to class, as in automatic classification in computing with learning machines or other mechanisms, ie. define a set of classes and put every class expression into some of those classes, clearly the chemical element class expression and isotope class expression definitions would become very handy and have a lot of sense. Imagine, in the sense of OWL punning, the classes item (<Oxygen-18> or <Oxygen>) becomes some kind of reification ( the senses on computer science for example, or in knowledge representation, I'm also aware of uses in logic), in some sense, is the act to treat something abstract like the other object in a model to be manipulated as the other basic objects of the model).

Then we got something very solid here : metaclasses are classes of reified (OWL2 allows this) classed using properties of their intensional definitions (as classes). We're fine with the class/token principle if we don't forget that reifications are not real world objects. It's natural to a lot of people (ie. this is exactly how the <isotope> concept is defined by french Wikipedia : an isotope is a nuclide (Q108149) (View with Reasonator) , nuclide is a type of atom. (seems the article have been rewritten since I read it the last time, it did not have the nuclide concept in it ! By the way @Emw:, were you aware of this concept ? How does this fits in your views ? It seems to me that it is exactly the atom type class concept I proposed ! Once more something I did not know but prove my reasoning solid, there is a word for this concept. So Wikidata has to be able to deal with this. Just merged the items.) I don't have time to rewrite all, but feels like a victory to me. TomT0m (talk) 17:00, 22 July 2014 (UTC)

WEF gadgets update[edit]

This is an gadget update notification.

External links gadget update

(«WEF: Links» in "Tools" menu at the left)

  1. all changes are stored using single edit query (second can be used for removing claims, if any were deleted), not query-per-edit as before
  2. support qualifiers (usually -- language (P407)). language (P407) qualifier used by (no label) (Q17116619) to select and display links (for example, if we have Russian and English links with normal rank, and French one with preferred rank, template will output Russian and French only)
  3. add encyclopedia links support (via own properties or described by source (P1343))
  4. add VIAF lookup support

Gadget is available in ruwiki directly (from prefferences). In all other projects (incl. Wikidata) it can be added to common.js with the following line:

mediaWiki.loader.load( '//' );
New gadget to edit person data

(item «WEF: Person» in "Tools" menu at the left)

Gadget is available in ruwiki directly (from prefferences). In all other projects (incl. Wikidata) it can be added to common.js with the following line:

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

Source code is available here:

Gadgets are localized in Russian and English. One of the following languages is used:

  • language from user preferences
  • content language of the project
  • English (fallback)
  • Russian (fallback)

Property and entity names are obtained as localized labels from Wikidata and stored in local storage cache.

Also, creation of new gadget is a matter of hours. If any project need a similar gadget (movies, cities, countries, languages) -- feel free to ask here or at my talk page. -- Vlsergey (talk) 05:47, 16 July 2014 (UTC)

Interesting and useful gadget I think! But somehow loading localized labels does not work in all cases fine though. I just tried with German language for the person gadget, when I click on "Update labels" in the "General" tab, sometimes it loads the German labels for the bottom three properties and sometimes it loads the German labels for the other properties in that tab. And sometimes it does not display any labels at all, just the property number.
Actually, I just logged in into Wikidata (I was only logged in into Wikipedia at the time I tried it) and now it works fine. So, problem resolved I guess? Not sure what logging into Wikidata changes here. Or it was just some random glitch. --Bthfan (talk) 07:32, 17 July 2014 (UTC)
@Bthfan:, labels are not loaded immediately, it takes some time to send requests to Wikidata server and load all labels and descriptions. But i will try to reproduce the case and check what can occur. -- Vlsergey (talk) 07:59, 17 July 2014 (UTC)
@Bthfan: I was able to reproduce the problem and fixed it. -- Vlsergey (talk) 15:38, 17 July 2014 (UTC)
@Vlsergey: Many thanks for doing this, it is so useful! The only thing I missed is some link to the property here in wikidata. Sometimes they are not translated, so I have to come to WD and look for the page here. I also missed these properties for people: movement (P135), notable works (P800) and this in general: Gran Enciclopèdia Catalana ID (P1296). Another gadget for taxon info would be useful too.--Micru (talk) 08:13, 17 July 2014 (UTC)
@Micru: i've added movement (P135) and notable works (P800) to person form; Gran Enciclopèdia Catalana ID (P1296) and ID (P1422) to external links editor (on "Encyclopedias" tab). Let me check, what property types are required for taxon editor. -- Vlsergey (talk) 17:25, 17 July 2014 (UTC)
@Vlsergey: Wonderful! I have some problems when using the search function of Gran Enciclopèdia Catalana ID (P1296), for some reason it uses the English name in Wikidata instead of using the text I input on the box. It would be nice if CANTIC (P1273) had the search function enabled too, it looks something like:$1&index=1, where $1 is the search string with white spaces converted into "+" signs.--Micru (talk) 09:30, 18 July 2014 (UTC)
@Micru: All search on external websites uses the titles of wikipedia pages, not the value fields. So for Gran Enciclopèdia Catalana ID (P1296) it now uses the title of the page from cawiki first (if present), and after it -- title of the page from 'enwiki'. Also i've added search button for CANTIC (P1273). -- Vlsergey (talk) 17:23, 18 July 2014 (UTC)
@Vlsergey: I've translated the gadget in French. Could you add fallback for it please?
Thanks. — Ayack (talk) 10:01, 17 July 2014 (UTC)
@Ayack: thank you for the translation! Please, also, have a look at the commons file used by both gadgets, it may need some translation as well: [4]/[5]. -- Vlsergey (talk) 17:25, 17 July 2014 (UTC)

Great tool! One question, should Property:P1037 really be on the person editor? --Ainali (talk) 11:36, 24 July 2014 (UTC)

Taxon editor[edit]

@Micru: please check the new Taxon Editor:

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

I'm not a specialist in biology, so feel free to propose any changes to grouping/splitting/etc on the form (at github, for example: [6]). -- Vlsergey (talk) 08:59, 18 July 2014 (UTC)

Josve05a (talk)

FelixReimann (talk)
Infovarius (talk)
Daniel Mietchen (talk)
Soulkeeper (talk)
Brya (talk)
Klortho (talk)
Tobias1984 (talk)
Delusion23 (talk)
Alexander Vasenin (talk)
Notified participants to Wikiproject Taxonomy.

Some remarks: botanist author abbreviation (P428) and author citation (zoology) (P835) belong to the "person editor", not to the "taxon editor". I would also add ecoregion (WWF) (P1425).--Micru (talk) 09:23, 18 July 2014 (UTC)
@Micru: botanist author abbreviation (P428) and author citation (zoology) (P835) moved to "person editor", ecoregion (WWF) (P1425) added to "taxon editor". Thank you! -- Vlsergey (talk) 09:36, 18 July 2014 (UTC)


Small update: just wrote up some instructions for framework. -- Vlsergey (talk) 20:41, 23 July 2014 (UTC)

Sometimes DOB and DOD get set to today's date, I fixed a few. Sample Q6025241 (still needs repair).--- Jura 21:20, 27 July 2014 (UTC)

Issue with "instance of" for texts[edit]

Currently, Help:Sources say that items like. Even though there has already been long discussions about text-items, it seems that we have failed on this particular point. It is recommended that items like Hamlet or the Bible should be marked as instances of book (Q571). This is not correct. "Subclass of Q571" would not completely correct the issue as Q571 refers to "set of written, printed, illustrated, or blank sheets, made of ink, paper, parchment, or other materials, usually fastened together to hinge at one side.". What the Bible and Hamlet are is not a book but a text. I am not sure whether is should be instance or subclass of text (I would think "instance of" would be more convenient, but possibly not completely correct). Also, "text" seems a bit vague, so we could use some of its subclasses. What Hamlet is is a tragedy, what Ode to a Nightingale (Q3349126) is is a poem. So, it seems that the p31 (or p279) value should be the same as the p136. So I would argue that p136 is useless (at least for works) and that its content should be merged with p31, and that "instance of book" should be only used for iotems about real material books.---Zolo (talk) 08:10, 16 July 2014 (UTC)

@Zolo: I agree with you, it was a mistake from the early times that we should correct. We are also discussing here if there is a better label for "instance of/subclass of" which would avoid misunderstandings as these. It could also be considered to have a specific property for creative works. Notification Books task force participants--Micru (talk) 08:40, 16 July 2014 (UTC)

According to WIDAR autolist there are 38,000 or so items with instance of = book. Would many be solved by linking to Q8261 (novel), Q179461 (religious text), Q5185279 (poem) etc. instead? The novel is a more abstract concept e.g. JK Rowling has written seven Harry Potter novels which have been distributed via the medium of books. Therefore Harry potter and the Chamber of Secrets = Q8261. Non-fiction texts would be more difficult to specify without a new item. 08:56, 16 July 2014 (UTC)

We should probably have a generic "non-fiction text" item but many coulds have more specific values (biography, essay, handbook, etc.)--Zolo (talk) 09:25, 16 July 2014 (UTC)
In general, I am in favor of adding instance of (P31), even if the item used is overly generic or slightly "off". Items without instance of (P31) are hard to find, if you look for something specific; but making instance of (P31) more specific is much easier if you have an "anchor" to work with. Maybe not the best example, but instance of:book in en:category:Poems gives us ~270 candidates. Autolist2 can add and remove statements, so selecting the poems in that list (or de-selecting the non-poems) and changing instance of (P31) is straight-forward. In the future, we could even have a tool list us "usual suspects", such as "instance of:book" automatically, and suggest "popular" replacements. --Magnus Manske (talk) 12:21, 16 July 2014 (UTC)
@Magnus Manske: The thing is "instance of book" is supposed to mean a material book like Lorsch codex (Q538377). These call for properties items about text do not (for their physical location and their material description). I am in favour of adding a generic p31 in the absence of a generic one, but I think it should be text (Q234460) rather than book (Q571), so that we do not mix differnt classes of objects. --Zolo (talk) 15:11, 16 July 2014 (UTC)
I agree. There are a few physical books which have their own articles and these particular tomes should be 'instance of:book'. Items about a text should be 'instance of:novel, poem, etc.' and 'subclass of:book, ebook, etc.' Filceolaire (talk) 21:34, 17 July 2014 (UTC)
@Filceolaire: Text is an abstract object if it is not a physical text, so "instance of" is wrong as it is for most books. Tamawashi (talk) 14:16, 18 July 2014 (UTC)
Tamawashi An abstract object can be an instance of a class of abstract objects. A book is a physical object so a book and a text must be in different classes. A whole taxonomy for books has been developed by librarians but I've lost the link. Filceolaire (talk) 19:39, 18 July 2014 (UTC)
@Filceolaire:. I don't think texts should be subclasses of books,~ebook etc. either, as they are conceptually independent from their physical materialization (they would remain the same text if they were written on a wall or recited aloud). However, in the terminology of Help:Sources, instances of {Q|3331189}} could be subclasses of books or ebooks. --Zolo (talk) 20:43, 18 July 2014 (UTC)
@Filceolaire: If an abstract object can be an instance of another abstract object, then how can editors decide whether to use "instance of" or "subclass of"? Tamawashi (talk) 02:18, 19 July 2014 (UTC)

@Filceolaire, Zolo, Tamawashi: There cannot be instances of abstract concepts in Wikidata, although instances of abstract concepts do exist in real life. For example, let's take a look at The Raven (Q22726):

  • if I memorize it (the process of transfering it to my pasive memory), I encode it in my neuronal circuits (as material substrate)
  • if I recite it (the process of activating passive information), other sentient beings or sensing devices (recorders) can become affected by the organized patterns of sounds that is my performance and perhaps create new instances of it
  • if I write it down, then again I am modifying a material entity (paper) to encode passive information that can be eventually decoded by someone else.

All those are instances, but if I want "something" to represent all of them, then I need a class, therefore <The Raven (Q22726)>subclass of<literary work (Q7725634)>. Same applies to music, movies, etc. which might be confusing to users (even for me it was confusing until not long ago), that is why I suggested above changing the label of p:p279/p:p31 to something easier to distinguish.--Micru (talk) 08:32, 19 July 2014 (UTC)

@Tamawashi, Micru: I really do not understand why an instance should be something with a location in space and time. Beside that would not be identical with being a material object. A football match can be located in space and time but it is not a physical object.
It seems much more useful to define instances and subclasses in terms of logical implications (like in Help:Basic membership properties). This is the cleanest way to use these properties for making inferences. Instance = the item has this feature. Subclass = Instances of this item has these features. A text is a defined sequence of word. Everything that is a defined sequence of word is an instance of text. Everything whose instances have these features is a subclass of text.--Zolo (talk) 08:55, 19 July 2014 (UTC)
@Zolo: There are not only material (or boundary-defined) objects, there are also processes. A football match is a process that is voluntarily engaged by agents (players) according to agreed rules.
There are two views: either we have models and we "create" things from those models (instances of a model), or instances exist on their own in real life and we create models from them. In practice it doesn't matter much, because you can express reality one way or the other and still reach the same reality representation. However, for the common mortal (non-IT people), the notion that the reality comes from the representation of it, it is, to say it in plain terms, a bit weird.--Micru (talk) 09:50, 19 July 2014 (UTC)
@Micru: as you say, the way we see the causal structure of the World does not really need to show in the use of the properties. I am just saying that the simplest and most useful definition of a an instance is "something that has the features defined by a class" An irrational number is a number that cannot be expressed as a ratio of integers. "Irrational number > 3" is not a number, so not an instance of irrational number. Other the other hand pi is an instance of irrational number. Adding requirements about physical location does not appear to have any practical benefit. Moreover, OWL now allows something to be both a subclass of X and an instance of Y. It is hard to see how an object with a physical location can be a class, so it must be that something without one can be an instance.
If we define a text as a unique sequence of words, or something like that, an instance of text must be an abstract object, not a physical one. This would make "Ode to a Nightingale" an instance of poem, and that would tell us that it is a unique poem, not a type of poem the way ode (Q178985) is. It would also permit additional constraints like "values of the edition (P747) property must be instances of text". -Zolo (talk) 11:26, 19 July 2014 (UTC)
@Zolo: If you define an instance as "something that has the features defined by a class", then what is a class and a superclass?
"Irrational number > 3" is a "subclass of:mathematical concept", when you write it down or, when you read about it, think about it, or otherwise create it in your mind, then you have an "instance of:Irrational number > 3". Pi, as any other irrational number, is also a "subclass of:irrational number concept". When I write down "Pi", that is an "instance of:irrational number concept", when I calculate it and write down the value that becomes an "instance of:approximation to irrational number".
The instances of the neurological processess of John Keats combined the linguistic patterns that he had acquired through his life to give rise to the first instance of "Ode to a Nightingale" in his mind. He penned it down and created another instance of "Ode to a Nightingale", then it was printed and more instances of "Ode to a Nightingale" were created, each one of them had a different material support, but all of them have a strong simmilarity and for that we can speak of a "the class of all instances that exist of 'Ode to a Nightingale'".
If you think about it, the only reason we have the sepparation between "instances" and "classes" is because for some things we have the factual certainty that they exist (instance, token, empirical evidence, phenomenon, the actual thing, reality) while with the others we imagine or suppose that they exist (class, type, kind, rational evidence, nuomenon, fabrication). Even the concept of gravity is a class that we have inferred after seeing instances of its effects repeated times. That is the basis of the scientific method, create classes from the observation of repeated instances. If you don't make that distinction, then it is equivalent to say that there is no difference at all, and in that case we can just use one property for everything and rely on bottom concepts.--Micru (talk) 12:08, 19 July 2014 (UTC)
@Micru: An instance is "something that has the features defined by a class and a subclass is something whose instances have the features of a class. Pi is a unique number and it is irrational, so it is an instance of irrational number. "Irrational number > 3" is not a itself a number so it cannot be an instance of number, but everything that meets the conditions "is a number", "is irrational" and "is > 3 is an instance of irrational number. So "irrational number > 3" is a subclass of irrational number.
Is the mental representation of an armchair -or the neural configuration producing this mental representation- an instance of armchair ? No because a mental representation cannot have a back and two arms. But just the same, how can a mental representation be an irrational number or be > 3 ?
That said there are certainly issues with allowing something to be both a class of something and an instance of something else. How do we know that the the statement "is > 0" means "this item is > 0" for pi (Q167) and "all instances of this item are > 0" for natural number (Q21199). That may not be trivial, but it appears to be a practical issue as opposed to the logical dead-end that seems to be created by requiring instances to be phyiscally localizable objects. --Zolo (talk) 14:02, 19 July 2014 (UTC)

@Zolo: Saying that "an instance is something that has the features defined by a class" is the same as saying that you have an object because you have a general representation of it... It doesn't seem rational to say "I know that the Mona Lisa is a painting because I always knew what a painting is". When I was little and I had no knowledge of language I wouldn't have been able to recognize the Monalisa as a painting, so most probably someone showed me a painting saying "this is a painting, and this, and this, and this", so that is why now I have a model (class) of what a painting is based on my exposure to the instances of that class and why now I can say "Monalisa instance of painting".

Pi is a pattern that happens in nature and we have idealized it (class). "Irrational number > 3" is a fabrication that you made up in your mind and I can replicate it almost identically in my mind (I say "almost" because I cannot have knowledge of your subjective experience while you are evoking that representation, so I cannot compare it to mine), for you that is enough to call both instances "the same", even if they are not represented with the same mind-substrate.

A mental representation of an armchair is the closest our mind can be to that existing armchair, that is why we call it "instance" and also because we can agree with others that we are referring to exactly the same object in time and space. Even if we now that it is a lie (our "mind object" or "object representation" can never be the object itself), it is "less lie" that pretending that things can exist without material substrate. In the same way, a mental representation is the closest we can be to an irrational number. If there wouldn't be matter where to store that representation, that representation wouldn't exist.

You cannot know the answer of "pi > 0" because pi is an abstraction and as such it has no value, but you can ask if "approximation of pi > 0", because you have executed some algorithm that gave you Approximations of π (Q1130396) and you can use that as a value for the comparison. The comparison itself "z = (x > y)" is a subclass of all comparison algorithms (passive process) that give a binary value of z as a result of the execution of the comparison process. The algorithm has an agent that executes the comparison process. x and y have to fulfill certain characteristics for the agent to be able to perform the comparison, but most probably the conditions are to have enough memory to store the representation of both x and y, and enough processing capacity to manipulate both representations. "z = (x > y)" as such may represent 3 superposed states: indeterminate, true, false. The agent selects which one is active at a given time. But all in all you need an instance of that algorithm (implementation, be it in your brain or in a computer) to be able to use it, which requires matter, energy, and takes space.--Micru (talk) 16:01, 19 July 2014 (UTC)

@Pasleim, DavidMar86hdf, Zolo, Micru, Filceolaire: And now classification as an instance [7]. Tamawashi (talk) 16:53, 19 July 2014 (UTC)
@Micru: I have a feeling that we do not speak the same language here. This is not about the psychological origins of knowledge. If we want to know if Mona Lisa is a painting, we just need to check whether it matches the citieria set in painting (Q3305213) or in some extrernal source.
"irrational number > 3" does not mean "someone's idea of a irrational number larger than 3", it means the set of real numbers that are larger than 3 and cannot be written as a quotient of real numbers. The neuronal circuit creating your idea of an irrational number may well be completely different from the one creating mine, but it is really irrelevant."Irrational number" is a mathematical object defined in mathematical terms. How the minds encodes it does not change a thing to its definition or its constituents. --Zolo (talk) 17:40, 19 July 2014 (UTC)
@Zolo: No, no, it is not about the "psychological origins of knowledge", it is plainly about the "origins of knowledge", or just about "knowledge". You are saying that it is possible to have information without matter to represent it - I am saying that is impossible. You are saying that if you write the number 3 in one paper, and the number 3 in another paper, you don't have two instances of the number three - I am saying that is impossible. You speak from the perspective of mathematics, where things do not need to exist in order to be imagined - I speak from the perspective of ontology, where even the concept of non-existence needs a material substrate to be represented. Our postures are very different, but it is not necessary to keep talking about this. Let's keep working on productive tasks and may some insight come to any of us that allows us some day to come closer :) --Micru (talk) 18:04, 19 July 2014 (UTC)
@Micru:, yes maybe it is better to focus on other things, but I am absolutely not thinking that we can have information without matter, just that:
a) The simplest and most productive definition of instance is "something that has the features defined by a class"
b) The simplest and most productive definition a subclass is "something whose instances have the features of a class".
c) A mathematical concept should be defined using the definition given by mathematicians.
d) Mathematicians define pi as "the unique real number whose value is equal to the ratio of a circle's conference to its radius", and prime number as "any natural number greater than 1 that has no positive divisors other than 1 and itself."
It directly follows from the points above that pi is an instance of real number and prime number is a subclass of natural number. These are plain logical implications. It does not purport to say anything about the nature of knowledge.
Obviously, we could use choose other definition for "subclass" and "instance". We could say that an instance is something that has a location in space and time. But the latter definition has serious drawbacks, like not making the distinction between a number and a set of numbers, or between an individual novel and a literary genre. Given that definition b solves this issue, I really do not see any reason why we should not use it. --Zolo (talk) 08:02, 20 July 2014 (UTC)

@Zolo: After reading your comment, I think we are closer that we imagine we are. I agree with your inderdependent definition of class-instance. For a class to exist (in a representation system) it needs an instance, and for an instance to exist (in a representation system) we need a class to represent it with. Our difference is in where this inderdependece relation originates. You say that it can exist "in the void", where I say that they need to be "grounded", that is, that instances must exist first in space and time. The definition of Pi adapted to ontological language is "there is an (idealized) class that represents a unique (idealized) result equal to the ratio of an (idealized class of) circle's conference to its (idealized) radius". It is important to understand deeply that the definition refers to classes, not instances, even if the result seems to be a unique number, because it is not. I think what is confusing is that you can have an instance of Pi (the idealized class) and also an instance of Pi (the class of all aproximations). For instance if I write: π, π, π, π, π, π, π, those are 7 instances of the idealized class of Pi (the ones that we can only think about abstractedly), but if I write 3.14159, 3.14159, 3.14, 3.14, those are 4 instances of the class of all Pi aproximations (the ones that we can process). If there was not a Pi (the idealized class), I wouldn't be able to write instances of that idealized class of Pi. Notice also that all instances exist in time and space, where the class is a supposition that more instances might exist.

As of "the latter definition [considering instances in time and space] has serious drawbacks, like not making the distinction between a number and a set of numbers, or between an individual novel and a literary genre." I don't see those drawbacks. Consider the following:

  • there is an idealized class of a set
  • there are instances of that idealized class of a set
  • there is a class of all real occurrences of a set
  • there are instances of a real occurrence of a set

So now a practical example with a natural number "5"

  • there is an idealized class of 5: the mathematical five
  • there are instances of that idealized class of 5: representations of the matematical five (5, 5, 5, 5, 5....)
  • there is a class of all real occurrences of 5: all fives that can happen in real life
  • there are instances of that class of real occurrence of 5: five apples here on my table

Of course it wouldn't be practical to type in all this structure in Wikidata, because we just need the mathematical idealized class, but that one is clearly a class, not an instance, both a class and and instance otherwise we wouldn't be able to instanciate it. Remember your own definition: for a class to exist there must be an instance, and the other way round.--Micru (talk) 09:48, 20 July 2014 (UTC)

@Micru: sorry, it is growing way too long for the Project chat, but we do not agree at all:
  • Concepts and real objets are too disjoint classes. A material object cannot be an instance of concept and vice versa. Concepts do not have a material existence, they only have a defintion.
  • Numbers are concepts, so no material object can be an instance of number.
  • Five apple are five instances of apple or an instance of group of apples (with the property: quantity of apples: 5), but it is absolutely not an instance of 5.
  • Writing "pi" on a paper does not create an instance of pi, just an instance of a depiction of the pi symbol. --Zolo (talk) 14:57, 20 July 2014 (UTC)
@Zolo: Yes, it is getting too long for the project chat. You say that concepts do not have material substance and I agree, but they can happen simultaneously in two different places. True that "concepts and real objets are too disjoint classes". Concepts are also both a class and an instance of themselves (recurrence), I corrected my words above to reflect this. Classes (group of apples) and their instances (five apples) do not show that property. What bothers me is the wish to use the same property both to relate classes with instances and concepts with classes. If they are two disjoint groups, why to use the same property?--Micru (talk) 15:26, 20 July 2014 (UTC)
@Micru: If something meets the formal definition of instance, I do not see why it should not be called an instance. That concepts and material objects are disjoint classes is actually rather fortunate in that respect. This way the superclass tree of the p31 value of an item can tell us with certainty whether the item is a material object or not. --Zolo (talk) 15:50, 20 July 2014 (UTC)
@Zolo: I have made a drawing above. If the red lines were instances, then we would have instance of instance, which doesn't seem right.--Micru (talk) 16:40, 20 July 2014 (UTC)
@Micru: Yes, as you had already noted, concept is an instance of concept and number is also an instance of concept so instance is an instance of an instance of concept. But that's ok, ask user:tomT0m about metamodelling :). But individual texts and individual numbers are really not classes so this is not an issue here (no a number cannot exist at several places at the same time, because a number, according to the standard definition, really does not exist in a spatiotemporal sense). --Zolo (talk) 17:20, 20 July 2014 (UTC)
@Zolo: A book may contain text and a text maybe published as a book. (But of cause you can publish a collection of pictures without text as a book and not every text is published as a book.) If you want a perfect definition for properties do not use Wikipedia. Why: Because WP combines articles in different languages. Abstract concepts are not existing like a single person and therefore they will always set apart. (German WP: book = printed book, German language: book = printed books and e-books, Unesco: only publications with more than 49 pages are "books" etc.) In our (WD) definition a "book" - book (Q571) - is a kind of work, the published book is called "edition" - edition (Q3331189). Changing basic definitions is like shooting a running horse and than wonder why this horse didn't win the rase. --Kolja21 (talk) 11:08, 20 July 2014 (UTC)
@Kolja21: we may not always be able to get perfectly equal definitions among projects, but here pretty much every Wikipedia and "authority control" link define books as a sequence of sheets etc. Even if we add ebooks to the definition, we have something very different from book seen as a "kind of work, the published book is called "edition", the published book is called "edition". Beside we need an item about real material books like Lorsch codex (Q538377)) and it only makes sense that it is the one with the links to en:Book,, etc.
@Kolja21:It is true that we should try not to change the structure of Wikidata too often. If you do not want to change the statements made in other items, and the documentation pages, we can also move the sitelinks, statements, and labels to a new item, and move the content of text (Q234460) to Q571. That will affect less pages, and, with me at least, it would be perfectly ok. --Zolo (talk) 14:57, 20 July 2014 (UTC)
@Zolo: text (Q234460) has not been used for works/books yet. So there is nothing to move. For the book In Mexico (2010) Wikidata tells us that the author is a person called "text" but that is just one of the usual bot errors.[8] --Kolja21 (talk) 15:21, 21 July 2014 (UTC)
@Kolja21: Items like On the Origin of Species (Q20124) are marked as "instances of Q571". The Wikipedia and authority control links in Q571 say that a book is a material object made of leaves. On the Origin of Species (Q20124) is not a material object made of leaves. So we must either change the value indicated in On the Origin of Species (Q20124) or the Wikipedia/authority control links in Q571. --Zolo (talk) 17:04, 21 July 2014 (UTC)
@Zolo: As I said before you will never have a 100 percent match. Your definition of book is one possible defintion, but not the one of WD. Take for example the English description: book = "medium for a collection of words and/or pictures to represent knowledge, often manifested in bound paper and ink, or in e-books". (An e-book as you know is not made out of leaves.) This is the official English description - today. Tomorrow an editor might change it. And of cause the German description is different. So forget about the English word "book" and the rest of the 200 or 300 possible labels, descriptions and Wikipedia links. It's item Q571 and we are happy with it. --Kolja21 (talk) 23:20, 21 July 2014 (UTC)
@Kolja21: I am not sure that you have correectly read what worte. I said that (almost ?) all sitelinks to Wikipedia said a book is made of leaves, so it would make sense adopt this definition in the corresponding Wikidata item. And even when Wikipedia links do not match 100%, descriptions in Wikidata should. If the German and English descriptions in Wikidata do not match, then at least one of them should be changed. Beside, this is not an issue of not being 100% right. As I, and Filceolaire, and Micru, and other before have noted it is just 100% wrong to say that "The Origin of Species" is an instance of book - whether we include e-books or not. If we changed the definition of Q571 to make it mean "text" instead of "book", it may be 95% right. --Zolo (talk) 07:33, 22 July 2014 (UTC)
A text? Are you serious? If we have problems with a difinition of a simple thing like a book, don't even think about talking about a text. This term is so complicated you can write a doctor theses about it and we still wouldn't agree what a text is. The concept of WD is to connect properties with items. "Freiheit (German) = Freedom (English)" is not the same as "2 + 2 = 4." Move sitelinks as often as you like, split them up, rename them and later merge them again, but stopp changing the definitions of properties. We lost hundreds of working hours because of chaos resulting in these kind of changes. We said The Origin of Species is an instance of "book". That is a definition. It's good and it works. If we would have said it's a "work", it might be a the title of an opera. If we would have called it "literary work" it would exclude many types of books. If we would have said it's a "text" it could be anything, from DNA to an email. --Kolja21 (talk) 15:14, 22 July 2014 (UTC)
As it stands now, should be considered plain wrong for any of you. The German Language knows the word Schriftwerk implying a work predominantly textual in nature but not necessarily an intellectual or artistic work (think of a phone book which essentially is a database, i.e. the effort relies on the act of collecting). However this would also include individual articles (published in journals or newspapers) like essays, a generalization not everybody might find appropriate. The main obstacle seems to be to translate Schriftwerk into enough languages to pin down the meaning of the corresponding item here. -- Gymel (talk) 15:33, 22 July 2014 (UTC)
@Gymel:. Would "written work" sound ok in English ? Also, note that both the translation and the generality issue will not be too prominent if, as proposed, we recommend using more specific subclass like "poem" or "phone directory" (best avoid the word book :) ) for the value of p31.--Zolo (talk) 07:22, 23 July 2014 (UTC)
@Zolo: I'm by no means a native speaker. Work as it is used here is something extremely abstract and "writing" even in a metaphorical sense might not match it. (Also I have the impression that "written work" in english is used for homework at college level). -- Gymel (talk) 19:51, 23 July 2014 (UTC)
@Gymel, Zolo: I think we need a property to say "<book> embodiment of <literary work>". If you don't like that label it could be said as: "gives material form to", or "materialization of".--Micru (talk) 07:55, 23 July 2014 (UTC)
@Micru: My objection above was that literary work is too narrow a superclass for "book" as subclass of a suitable generalization of creative work (Q386724). An embodiment property as proposed would apply to the different meaning of a book as a physical object. By the simplified model of Wikidata:WikiProject Books it (i.e. the book on my desk) would be an instance of a certain edition (Q3331189) which in turn is related to the work. -- Gymel (talk) 19:51, 23 July 2014 (UTC)
@Gymel:. I was about to create the "Schriftwerk" item, but I just realised that we should solve hierarchical questions first: should it be a subclass of text ? How can a text not be a Schriftwerk ? Can a Schriftwerk have illustrations ? And if so is it still a text ? --Zolo (talk) 07:35, 28 July 2014 (UTC)
I can only give a subjective answer: For me DNA (or the specific DNA of my dog) is an instance of text, but not an instance of work even in the broadest sense I can imagine. Thus Schriftwerk being a subclass of text could be worth a try (but verifying that text is represented by text (Q234460) would not be trivial). As for the illustrations: For a certain work a combination of text, music and visual content may form an unit. And for another work illustrations are "only" added. Both may be reduced to their textual content by disregarding their other parts and whether this reduction is still a representation (or expression) of the original work will be open to debate. Or think of a poem where the words are arranged in a way to yield the graphic image of an apple: There is no purity but the thing would be still considered as mostly of textual nature? Of course there are printed books (no label) (Q860594) without any text content not to mention Artist's book (Q1062404)... One word of warning against Schriftwerk: It seems that at least linguists reason along the language content of text as being the fundamental characteristics, not its written-ness. That would be "Sprachwerk" in german which even more than "Schriftwerk" is mostly used in the context of copyright law and therefore stresses the work-ness more than will be helpful for our purposes (which still would be to catch the non-physical aspects of a general "book", which should include written-ness (or only the possibility of written-ness?) but work-ness only in the very broad sense of "result of a controlled activity"). As a consequence of the two points you raised I don't think any more that we can define book in a meaningful way: Something abstract is a book if and only if this something has been or could be instantiated / made manifest in a book as a physical object - defined by its sheets or pages of what material ever bound together. Unfortunately audio books or e-books would be books only very indirectly and most unfortunately the DNA of my dog would directly qualify as a book since now any text of finite length could be printed out and bound together effectively turning text into a subclass of book. Urgh. -- Gymel (talk) 07:24, 29 July 2014 (UTC)

What to do when an item refers to two things in one Wikipedia article?[edit]

Q4010540 The Wikipedia article on this topic is about both the fictional character named Vext as well as a comic book series entitled Vext. In part, this is due to parameters in the infobox that collapse together fictional characters as well as series when they have the same name (note also that sometimes, there will be more than one comic book series with the same title—they are usually disambiguated by volume). What do we do on Wikidata about this? Is this data item about the character, the series (there is only one volume), both, or neither? —Justin (koavf)TCM 17:25, 17 July 2014 (UTC)

In my opinion, the item, in general, should be about the larger concept. An article about a city and the same name surrounding county would link to an item about the county. In this case the item linked to the article would be about the comic series with a separate item for any characters notable enough to have their own items (do we know enough about them to make three statements?)
On the other hand, in this case the article starts by talking about the character, not the series so maybe...
On the third hand the article goes on to describe 17 other characters as well so by the end of the article it does seem to be about the whole series.
All the above influences whether the enWP article is linked to the wikidata item for the series or the wikidata item for the character. Either way we still need separate items for these as statements about one are not true for the other.
Hope this helps a little. Filceolaire (talk) 21:12, 17 July 2014 (UTC)
On the specific question you asked: This is what is known as the "Bonnie and Clyde problem". A wikidata item should not refer to 'two things' though it may refer to a pair of things or a bunch of things where we can make statements about that pair or that bunch. If a wikipedia article refers to two things then the item linked to that article will in turn have 'consists of' statements linking separate items for those things. Where an article discusses one class/group/big thing and also a bunch of other things that can be considered to be 'part of' or 'instance of' or 'subclass of' the group/class/big thing then the Wikipedia article should link to the wikidata item for the group/class/big thing. Filceolaire (talk) 21:22, 17 July 2014 (UTC)
@Filceolaire: But this isn't like w:Coen Brothers/Q56095 where there are two things (Joel and Ethan) who have some separate properties (e.g. their birth years) and some common properties (e.g. awards that they have won together or occupations that they have shared separately): it's not like a comic book series is an aggregate of comic book characters. One is a publication, the other is a fictional entity and they both just have the same name. I would guess that the series is the "main" topic but that's more or less arbitrary, really. —Justin (koavf)TCM 07:40, 18 July 2014 (UTC)
Note: I actually looked up that Wikidata item and there are parts segregated out to Q13595531 and Q13595311 even though neither has an independent Wikipedia article, Commons category/gallery, Wikiquote listing, Wikivoyage guide, or Wikisource bibliography... —Justin (koavf)TCM 07:42, 18 July 2014 (UTC)
Yes Justin except I didn't talk about the 'main' topic. I said the article should link to the group/class/big thing and that is the series. Each of the characters is featured in the series (and listed on the page). Each individual comic is a part of the series (and referred to in the 'publication history' section of the article). so the series is the group/class/big thing and the wikipedia article should link to the item for that.
I can imagine a language wikipedia with an article about Captain America which includes discussion of his appearance in his own comic, the Avengers, movies, TV series etc. where the article should link to the item for Captain America and not to one of the comic book series. Either way Wikidata still needs a separate item for each series, each character, each edition. Wikipedia can combine these topics in articles in many different ways and which of these items each article links to is then a separate question. Filceolaire (talk) 09:47, 18 July 2014 (UTC)
@Filceolaire: Ah. Thanks. That makes sense. So I can/should split this single item into two? —Justin (koavf)TCM 03:09, 21 July 2014 (UTC)
Statements in this wikidata item should (in my opinion) be about the series. If you have statements you want to make about the character or the individual comics (publication dates) then you will need to create separate wikidata items for these topics. Can you make enough statements about the character to justify creating a separate item? Filceolaire (talk) 07:28, 27 July 2014 (UTC)

Increase the minimum support votes for WD:RFA[edit]

I think we are now such a large community that it'd be safe to raise the number to 10 or 12. --Ricordisamoa 13:03, 19 July 2014 (UTC)

I agree. I've also thought about if we could make some rule to not accept users without any edits on Wikidata to vote. --Stryn (talk) 17:32, 19 July 2014 (UTC)
  • This needs an RfC, it can't be raised by a project chat discussion alone. But I do agree - I would even suggest 15 (12 is not that much greater than 10) and 20 for bureaucrats (do not change CU and OS).--Jasper Deng (talk) 18:04, 19 July 2014 (UTC)
not broken and does not need the change. If community is big enough, there will be enough votes pro and contra. If we have big community and not enough vote -- it means nobody cares. So, no big deal. I don't like the idea to force people to care. But, i agree to Stryn, votes should be required to have some qualification. -- Vlsergey (talk) 18:26, 19 July 2014 (UTC)
I don't really think we need a change in support levels but I won't be against implementing them. re. voting requirements - yes. John F. Lewis (talk) 18:35, 19 July 2014 (UTC)
+1. --AmaryllisGardener talk 20:15, 19 July 2014 (UTC)
I can not recollect recent closing problems because of the low participation, and I would not object rising the number of minimum votes, on the other hand I do not see any particular need. It should go through an RFC anyway after this preliminary discussion.--Ymblanter (talk) 20:58, 19 July 2014 (UTC)
I would think 12-13 is enough. Jianhui67 talkcontribs 01:38, 20 July 2014 (UTC)
Per Ymblanter. --Rschen7754 06:18, 23 July 2014 (UTC)
Can't see much of a need, but wouldn't be vehemently opposed to it. TCN7JM 06:21, 23 July 2014 (UTC)
Don't see a need either. More important is that there aren't too many "oppose" votes. Lymantria (talk) 07:15, 23 July 2014 (UTC)
10 supports should be enough for now, but I agree with what Stryn said. --Jakob (talk) 13:25, 26 July 2014 (UTC)

Blocked by two admins, where was the "pattern of local abuse"?[edit]

You are currently unable to edit Wikidata.

You are still able to view pages and data entries, but you are now not able to edit, move, or create them.

Editing from Tamawashi has been blocked (disabled) by John F. Lewis for the following reason(s):

Inappropriate use of user talk page while blocked: Refusal to listen/acknowledge comments by users, see also this AN discussion

This block (ID #6387) has been set to expire: 23:52, 20 July 2014.

Even if blocked, you will usually still be able to edit your user talk page and email other editors. To discuss the block, you may contact John F. Lewis or another administrator.

When following the link labeled "blocked", one comes to the page "Wikidata:Blocking policy", which says:

Administrators may block user accounts or IP addresses:

  • To prevent local abuse where a pattern of local abuse has been established. Local abuse includes, but is not limited to:
  • When consensus to block a user or IP has been developed on a discussion page, with at least a week of discussion and clear consensus on administrators' noticeboard.

QUESTION: Where was the pattern of local abuse? Where is "local abuse" defined? Tamawashi (talk) 00:29, 21 July 2014 (UTC)

@Tamawashi: First thing which people never understand; we don't have a blocking policy. Now, refusal to work in a co-operative and community environment by repeatedly removing questions and warnings is just a major no. The AN section in question pretty much shows me a consensus among the administrators that there is an issue here which we can't resolve because you refuse to communicate with us. On a side note; unless you have something real to discuss - I ask you to stop carrying on the behaviour which resulted in your talk page access being revoked. Thank you, John F. Lewis (talk) 00:33, 21 July 2014 (UTC)
My reply:
  1. OK, no blocking policy, admins can do what they like to. Thank you for at least admitting this.
  2. I ask you to stop carrying on the behaviour which resulted in your talk page access being revoked. Can you define such behavior? Oh, but wait, there is no blocking policy, i.e. talk page access can be randomly removed, it wouldn't follow any defined behavior.
  3. "consensus among the administrators that there is an issue here" - ah yes? A consensus among those with superior rights to keep users down that don't follow their random commands?
I made 300000 contributions in one month, I am at 200000 in the second month. I am discussing EVERY content issue, 1) if there is disagreement 2) if I am aware of the existence of such 3) if the discussion is lead in English. It is pure defamation to talk of "refusal to work in a co-operative and community environment". You are now at least the second admin to post defamatory statements. 01:06, 21 July 2014 (UTC)
The comments by these two admins are not defamation under any standards. Please refrain from using that word lacking a proper context. Now, we do have a blocking policy and is called common sense. We are here to take care of a wiki and make sure everyone can work and discuss matters on a timely and friendly manner. This is not the behaviour you have shown, and as such you received warnings and, ultimately, a block. — ΛΧΣ21 01:11, 21 July 2014 (UTC)

@Tamawashi: I'll be completely blunt about this. In addition to John's and Hahc21's comments above, you appear to not assume good faith nor assume that we do with you. "Defamation" and "trolling" accusations (such as here, against @TomT0m:) fundamentally fail to do either, without substantial evidence. As for blocking you, I do not know of any administrator who opposed the block of you, nor did I see any other users comment to that effect. It is true that our blocking policy needs to be actually fleshed out, but en:WP:NOTTHEM behavior is invariably frowned upon by the community and not an appropriate use of your talk page while blocked. My honest advice for you is to just drop the stick, because your behavior, if you choose to continue it, will lead to further blocks.--Jasper Deng (talk) 01:49, 21 July 2014 (UTC)

User:Jasper Deng - you again fail to properly parse text. It was TomT0m that made the troll allegation against me. So, now the way you treated me, go to TomT0m and treat him like that? Let's see, if you are fair, and apply your standards to everyone. "en:WP:NOTTHEM" - First, this is Wikidata not English Wikipedia. Second, why do you refuse to talk about your behavior? It was a separate issue. My behavior is mine, yours is yours. Putting the NOTTHEM sticker all the time makes you look a little bit like a hypocrite, since when a discussion about your behavior is started then you use NOTTHEM to evade the discussion. Regarding "drop the stick" - well, you chose to attack me in the first place, by stating "I will also say that Tamawashi's been not so careful with his editing of items" at the AN [9], without providing evidence that I was not careful. Tamawashi (talk) 02:04, 21 July 2014 (UTC)
Can we please desist from this matter now? No more mentions of it, no more accusations etc. that goes for both of you (Tamawashi and Jasper Deng). I shouldn't really have to say this but if you make another accusation whether provoked or not Tamawashi, I'm afraid we may have to block you again for carrying on behaviour immediately after your block expired. Take this as a final warning please. John F. Lewis (talk) 02:10, 21 July 2014 (UTC)
May I ask, why threatening with a block is only applied to Tamawashi, and not to other people? Can they make accusations against Tamawashi, but if he accuses them of unfair accusations, then he gets blocked, irrespective of whether he was right or not? I would like to drop the whole matter, but I FEEL TREATED UNFAIRLY. Tamawashi (talk) 02:17, 21 July 2014 (UTC)
This is nothing about unfair treatment. The mention of a block with regards to you is only because you came off a 24 hour block and immediately started this thread and began to accuse another user of trolling, therefore you came off a block and started to be disruptive again. If Jasper Deng just came off a block and did this; he would get the same response as you. John F. Lewis (talk) 02:21, 21 July 2014 (UTC)

Section break[edit]

This still is not answered: "Where was the pattern of local abuse?" - Any idea why? Any diff? Tamawashi (talk) 02:13, 21 July 2014 (UTC)

There does not need to be a pattern of local abuse. The block comes up 'consensus to block at WD:AN' and general discretion I exercised by WD:UCS. Now as I said, please desist from this matter now. John F. Lewis (talk) 02:15, 21 July 2014 (UTC)
Thanks for the clarification that "pattern of local abuse" was not the reason. Tamawashi (talk) 02:22, 21 July 2014 (UTC)
Great, can we drop the matter now and get back to focusing on editing? John F. Lewis (talk) 02:23, 21 July 2014 (UTC)

This place is starting to look even worse than the English Wikipedia. Great.—Al12si (talk) 10:04, 27 July 2014 (UTC)


Oluwole 'Segun Michael A native of Iloro-Ekiti in Ijero LGA area of Ekiti State, South West Nigeria.

General question regarding subjects of important works[edit]

Hi I have been updating the items for people portrayed by Frans Hals, but I notice that it's sometimes a bit blurry whether an article is about the person portrayed or about the creative work (in this case, a portrait). To solve these, I have been creating two items, one for the person and one for the creative work. See Aletta Hannemans: Q16859703 and Q17275957. With the portrait painting, I want to link to the pendant (for which I use the property "part of"). With the person, I want to link to her family (and soon also, the house where her brewery was). My question is, does she really need two items in this case? I want to be sure before I do more of these. Thanks in advance. Jane023 (talk) 07:11, 22 July 2014 (UTC)

Yes, you should create two separate items like you did here. Multichill (talk) 17:28, 22 July 2014 (UTC)
OK Thanks! Jane023 (talk) 20:20, 22 July 2014 (UTC)

OK same two items, next question: In the person item, should the property "image" link to the Commons image, or should it link to the item for the portrait? Thanks again in advance! Jane023 (talk) 12:22, 26 July 2014 (UTC)

And Multichill, here is another related question (working with the Rijksmuseum dataset): In Q17419089 this person item, I pick a detail of a painting to illustrate the person. I link to the detail file on Commons with property "image", and then I say that it is "part of" the item for the painting. Is this enough? Jane023 (talk) 13:50, 26 July 2014 (UTC)

What are the aims of the Wikidata project for Wikipedia interlanguage links?[edit]

I am looking for a Wikidata page that describes the aims of the Wikidata project for Wikipedia interlanguage links. What should it provide? Of course it should provide links between different languages of Wikipedia. But what characterizes two articles of two Wikipedias-languages to have an Interwiki link on Wikidata? Could anybody give me a Wikidata link to such a description? Thank you! -- Tirkon (talk) 20:32, 22 July 2014 (UTC)

Does Help:Sitelinks help you? --YMS (talk) 07:32, 23 July 2014 (UTC)
Thank you, YMS. I think one hint is missing there. Wikidata interlaguage links serves readers of Wikipedia and not the data-aspect. Readers should find informations from the article in other languages. It is not necessary that two linked articles have the equal lemma. Nevertheless you should aim this. But it is sufficient to find similar informations in the articles. Since the interlanguage links moved to Wikidata people think they should give different Ids to slightly different article-lemmata. But that is not the sense of the interlanguage links. And it was not the way interlanguage links were provided in Wikipedia. Thus the reader does not find informations in other languages. -- Tirkon (talk) 04:15, 26 July 2014 (UTC)
Tirkon, this is by design. The whole point of Wikidata is being specific, while often on Wikipedia various concepts are bundled into one article. In languages where there are two meanings for one word, you will often both meanings in one article, whereas in languages where these two meanings have two words, there are two articles. Neither language is more "correct" than the other one, but Wikidata has two items, only one of which can be linked in those languages which have "bundled" their concepts into one article. Jane023 (talk) 12:30, 26 July 2014 (UTC)
Thank you, Jane. I know the problem with not matching articles or two lemmata in one article. What do you think about the following approach in case of doubt how to group the articles together under same ID: "The aim of Wikidata interlanguage links is to provide informations in as much as possible other languages. The aim is not to group exactly the equal lemmata. That means: If grouping similar lemmata together (same ID) instead of dividing them into different groups (different IDs) produces considerable more interlanguage links in all languages than another grouping, then this grouping should be chosen. -- Tirkon (talk) 17:54, 26 July 2014 (UTC)
I think you need to just start in and see how it goes. At first I felt the same as you, but I changed my mind. I disagree with your purpose for interlanguage links. Not "as much as possible other languages" but "in other languages". Wikidata will help to start splitting articles that are now bundled in Wikipedia. I don't mean that Wikipedia should become a dictionary, but I do think we should see some of the really long articles on Wikipedia start to split into more manageable chunks that can be read comfortably on a mobile device. Jane023 (talk) 18:40, 26 July 2014 (UTC)
There are cases where every Wikipedia groups topics together but Wikidata still needs to split these topics into different items. This happens when statements about one topic are not true about the other item. If we need separate items to make statements then we create separate wikidata items, even if none of those items will ever have it's own independent wikipedia article. Filceolaire (talk) 07:14, 27 July 2014 (UTC)

Q194195 and Q2416723[edit]

Both Q194195 and Q2416723 are labeled "theme park" and seem to be described similarly. I'm sure something should be done (merge, delete, rename, etc.) but I've tried reading the Wikidata documentation and still don't know how to proceed. (I tried to merge, but the system disallowed that action.) Senator2029 (talk) 17:17, 23 July 2014 (UTC)

@Senator2029: You weren't allowed to merge because both items have links to dewiki. I've changed Q194195's label to "amusement park" and left Q2416723 alone, because a "theme park" (freizeitpark) in German is not the same as an amusement park (themenpark). --AmaryllisGardener talk 20:02, 23 July 2014 (UTC)

Lets leave fixed descriptions behind us[edit]

I have been searching for members of the European parliament in Wikidata. I needed to find them by name to identify them on an external list. I found myself hampered by the intrusive and limited fixes descriptions that we have. "British politician".. It does not mean that he or she is an MeP. It does not state that he is 19th century, it is bloody useless.

Compare this with automated descriptions; as more statements become available it becomes easier and more obvious to pick the MeP among search results..

My challenge. Provide convincing arguments why we should have fixed descriptions at all. We are better served by automated descriptions and it stops people from wasting their time adding text that is ultimately useless. Thanks, GerardM (talk) 21:04, 23 July 2014 (UTC)

A real challenge would be to provide useful automated descriptions for all kind of items. E.g.: I remember a discussion about some politician(?), which was also a football player and two other things. Because of that he had multiple instance of (P31), so which one to prioritize in a description? This would mean that in the future we need to use ranks even more to solve such cases. But yes, fixed descriptions are sometimes not really useful. --Bthfan (talk) 23:31, 23 July 2014 (UTC)
Really? He is a p31:q5 and anything else ? He may have different occupations and they are typically centred around different times in his life. Wrong example. GerardM (talk) 05:29, 24 July 2014 (UTC)
Reality is that data in Wikidata is not yet of expected quality and won't be for a long time. But you're right, this was actually a bad example. But now, let's take a look at occupation, four different occupations, which ones to display? --Bthfan (talk) 06:38, 24 July 2014 (UTC)
Reality is that the quality of statements is typically much better than what fixed descriptions offer. When I am looking for American movie actors, I want Ronald Reagan included. Exclusion means assumption and why bother? GerardM (talk) 07:22, 24 July 2014 (UTC)

My opinion is that descriptions are really only something humans should deal with, whether reading or setting them. The reason is that all machine-readable data should be standardized with claims (along with qualifiers), and bots are not good at setting anything beyond what can be described by "instance of".--Jasper Deng (talk) 23:42, 23 July 2014 (UTC)

Right.. I had to deal with garbage descriptions that were a bother and not a help. Not an opinion; observable fact. Bots are quite capable of setting all kinds of other statements based on the right definitions. So what is your point? GerardM (talk) 05:29, 24 July 2014 (UTC)
Auto-generated descriptions are a great thing and should be used more extensively, yes. But they only work with certain categories (and given that enough statements are set), e.g. humans, creative works, taxons. They are hard to model for more abstract things, like democracy or the rise of Lenin. And they will never know how detailled they have to be to really be able to distinguish the item they describe from others. "British football player" will be enough to distinguish an item from the others that are named the same in most cases, but in some cases it has to be like "British football player, born in 1978, played for Liverpool and Manchester" or something like that. Should the automated description always include quite every information that's available, just because it could be that it's needed? --YMS (talk) 06:28, 24 July 2014 (UTC)
If you are looking for duplicates then all the bits of information help. You are not addressing the point that fixed descriptions are better. When certain categories are not "helped" are they hindered? The class or instance already provide substantial information.. Please forget about English; people may not understand "Britisch football player" and, he may be from Scotland. GerardM (talk) 07:22, 24 July 2014 (UTC)
I have been doing my best to keep up with vandalism and unhelpful edits (e.g. enter discripton [sic] in [language]), and I have noticed that 90% of vandalism or useless edits are in the labels or descriptions. Perhaps other editors have been watching out for vandalism, but typically the vandalism in RecentChanges starts roughly when I had stopped looking for it, so maybe not so much... This means that the editor(s?) who care about vandalism are swamped with label/description vandalism, which takes time an attention away from data vandalism and makes it more likely that data vandalism will not be noticed. Maybe descriptions are really only something humans should deal with, but unfortunately they're not. The humans are in dire need of assistance of one form or another. For certain classes of items at least, such as of people, a formulaic description is good. Just today, I replaced a description of Tokyo High Court with "a high court in Kasumigaseki, Chiyoda, Tokyo, Japan". I think that such a description could have been added automatically. This replaced the former description, "an artist from vancouver who is less than half my age with her own degree, is robbing me and my friends at a Women's art gallery in Montreal where I have been going since 2008", which had been on there for one year, one month, and 22 days. --Haplology (talk) 13:05, 24 July 2014 (UTC)
English language descriptions don't usually contain a/an/the so I removed the "a". :) 15:59, 24 July 2014 (UTC)
I think the problem here is that we're not sure about the purpose of descriptions -- what are we ultimately using them for? In my mind, they're for disambiguation and a short explanation of what the item is about (e.g., a succinct version of the first sentence in a Wikipedia article's lead). It looks like you're trying to use them for categorization and as a summary for the entire item, but wouldn't we eventually just be able to use queries for that when it's added into the wiki-interface? As long as Ronald Reagan has "occupation = actor" in there, it would turn up in a query. However, if you were trying to differentiate between multiple people named Ronald Reagan, you would probably be looking for the "American politician" or "40th President of the United States", not "American actor" or something like that. The software probably couldn't automatically realize that him being a president is the most important, especially since we would all disagree about how much information is important (sometimes 19th century is relevant, sometimes it's not), so we need a fixed descriptions added by humans. I do agree that something very simple like "politician" translated into 30 different languages is not helpful though. Cbrown1023 talk 14:27, 25 July 2014 (UTC)

User conduct policies RfC[edit]

I have just started a massive RfC here to attempt to overhaul some user conduct policies. Translators are also needed.--Jasper Deng (talk) 22:24, 23 July 2014 (UTC)

I Symbol oppose vote.svg strongly oppose the premises under which this RFC has been drafted. See a detailed explanation here: Let us talk.--Micru (talk) 12:35, 24 July 2014 (UTC)


@Jasper Deng: As an alternative to more rules, I would like to suggest to start something similar to the Teahouse to welcome new users, to encourage them to participate, and to help them integrate into our community. Better to welcome new friends with a cup of tea than with spiky laws and regulations! --Micru (talk) 14:15, 24 July 2014 (UTC)

Strongly agree :) GerardM (talk) 16:24, 24 July 2014 (UTC)

@Micru, GerardM: You two obviously don't understand. This RfC is intended only for experienced editors. And if even experienced editors can't agree on policies, then what should we tell new users?--Jasper Deng (talk) 17:28, 24 July 2014 (UTC)

From your response it seems obvious that I am not an experienced editor. Suffering a lack of communication is therefore the ticket.. ? Does it mean that only experienced editors may express an opinion ?? Can we have a teahouse, an IRC channel that is not special purpose ... where peole can ask and gain experience ... Once they have experience, these policies start to apply ... Right ?? Thanks, GerardM (talk) 20:08, 24 July 2014 (UTC)
Where did I say that? New editors will of course have to obey it, but it's the experienced editors that need these policies. Not that new editors won't have to follow them. And frankly, I don't get how a teahouse or IRC channel has anything to do with fixing this particular problem (not that they are bad ideas). --Jasper Deng (talk) 20:13, 24 July 2014 (UTC)
@Jasper Deng: If it is for experienced good-intentioned editors, then it is even worse than I originally thought. Our goal to build a healthy community should be that everyone here can become best friends, not arch-enemies. Cultivate weapons, and you will have war. Cultivate dialogue and the wish to understand each other's opinions, and you will have peace and harmony even when disagreeing.
Yes, it is much easier to have a trigger that you pull and your problem sort-of-disappears, but it never works that way, it will start a dinamic of oppression and mistrust from which it will be very hard to escape. Have you heard about Wikipedia's editor decline and ever wondered why?
When asked to identify Wikipedia’s real problem, Moran cites the bureaucratic culture that has formed around the rules and guidelines on contributing, which have become labyrinthine over the years.
On Wikimania last year there was also a very interesting talk by a Polish researcher: Collaborative or Conflict-Driven? Conflict Trajectories on Wikipedia (there is also a youtube video). Dariusz recommends to create the role of facilitators, users who help to bring perspective and peace, in order to avoid entrenchment and hate speech.
In Wikidata we should hear to wise advice and strive for community peace as the ultimate goal. If there is any policy that we should adopt, that is the friendly space policy (I recommend watching the whole video), so that if anyone feels hurt or under harassment there will be always someone to hear them. We should act as a shelter for all our members, no matter if they are sometimes right in some issue, or wrong, because our contributors are the most valuable jewel that we have.
That RFC speaks too much of edit warring and punishment, but too little of peace and conciliation. To learn the language of peace takes time and community effort, but the fruit that will give us is much sweeter and lasting. I kindly request that this RFC is stopped and instead we draft collaboratively one from scratch where everyone can feel represented. That will be more aligned with the wiki spirit.--Micru (talk) 21:18, 24 July 2014 (UTC)
If you feel like that's a problem, then add new sections about dispute resolution. And, I repeat, blocking is not a punishment. I don't like doing this RfC, but user conduct issues are neccessating it.--Jasper Deng (talk) 22:54, 24 July 2014 (UTC)
After reading your comment more, I also will say that the main problem with policies is an excess of rigid policies that the community neglects to amend over time. I only included the minimum number of policies necessary to solve the current disagreements; for example I have not formally proposed any policies on things like citing sources in this RfC. If you don't like the proposal, you can of course oppose it.--Jasper Deng (talk) 03:56, 25 July 2014 (UTC)
I question the legitimacy of a Rfc that has been framed as a support/oppose and not as consensus building. I also question the process by which this Rfc came into existence, right after there were some conflicts, when the animosity is still high, and without initial community input. For these reasons I abstain to participate.
In my opinion this RfC only addresses "surface problems", but not root causes. When we arrive to the situation of blocking an user, that is already too late. We should instead focus on how the problems arise, and address those situations first in a patient, and friendly manner. When that fails, ok to block, but admins already can do that, this is not a new feature, just a framework to "black-and-whiten" it when otherwise it would had been morally objectionable.--Micru (talk) 11:04, 25 July 2014 (UTC)
I didn't intend to frame it as anything but consensus building. Phase 1 is intended for the preliminary discussion. The root cause is our lack of written agreement on policies. This RfC in fact requires the collaboration of the community on agreeing on them. You have so far said many things saying that "rules are bad" but you have not given any evidence for how new policies could not possibly prevent further incidents. After all, content disputes are perennial parts of a wiki and I frankly do not get how a teahouse would help preventing any of them from getting out of control. You have not answered my basic question: What do we tell users in a dispute when conflicting advice likely would arise without agreed-upon policies?--Jasper Deng (talk) 19:18, 25 July 2014 (UTC)
@Micru: We have tried to be friendly and patient in the past. It just hasn't worked to resolve recent disputes and I see no reason why it would suddenly start to. Just search the AN archives. But it does not go without saying that I worded it as "administrators may" and not "administrators must". Administrators understand that warnings and other venues such as the one you proposed below should be tried first.--Jasper Deng (talk) 18:54, 25 July 2014 (UTC)
We didn't have that page in the past, so maybe we can experiment and see if it makes a difference. It is all up to us, if we want to focus on "good" and "bad" effects, or if we want to focus on what causes those bad effects. The web is very impersonal, and that already brings a disposition towards "bad effects", and to counter that we can only make it more personal and friendly.
Now we are planting the seeds, but it will be a long time before we can see the effects of using different dynamics. I ask for a bit of patience and ideas that each one of us can use to make our virtual environment more human-friendly.--Micru (talk) 10:50, 26 July 2014 (UTC)

I'd like to see a successful RfC and a Teahouse. More policies, and a welcoming Teahouse. --AmaryllisGardener talk 17:48, 24 July 2014 (UTC)

re. an IRC channel; we have one - #wikidata. It is a channel ran by the community not the staff (sure, Lydia might be the channel contact) the chanops are all community members (mostly Rschen and I) and sure the development team piggy-back off their development part in the channel but that is because the development and community need to be one. The lack of user interaction in the channel is the communities fault for not using it. If we moved the development team into #wikidata-development, please tell me how that would boost interaction? It won't. John F. Lewis (talk) 20:18, 24 July 2014 (UTC)

John, you describe the status quo and then blame the community for not using it. Given the amount of work related traffic and given that answers to questions go unanswered, I find that I hardly ever use the IRC channel. Something new is happening all the time and it is hardly ever relevant to people not working on the Wikidata code.
When development uses its own channel and when admins talk about not so sensitive stuff in the open channel, we will get more relevant traffic to the users of Wikidata. Yes, you can BLAME others for not using the channel but now you know why. Thanks, GerardM (talk) 23:18, 26 July 2014 (UTC)
Administrators have a separate channel only for the purpose of discussing private matters such as revision deletions. We only use it for that and a way to communicate with other administrators asking for 2Os. Maybe Wikidata is a project which doesn't have to use IRC everyday then, that is fine. What is not fine is you always going around saying we don't have one. We have one, whether the community use it or not is their choice and their fault, not the development team's fault for using an empty channel instead of creating a new one. John F. Lewis (talk) 23:23, 26 July 2014 (UTC)
John, you are blaming the community again. You object to me having a point of view that is not yours. And, from my point of view the status quo is absolutely not fine. Why not have an other IRC channel how does it hurt? Call it an experiment and see what happens.. Thanks, GerardM (talk) 19:11, 27 July 2014 (UTC)
As I've said multiple times on here and on enwiki; the community can not refuse to use something, and then complain it is not there. Then when called on it, say 'it's not our fault, it's yours'. All I am going to say on the matter is, it is there. Use it, don't use it, I'm not really bothered. Just stop complaining it is not there when it is. John F. Lewis (talk) 19:15, 27 July 2014 (UTC)
I’m a bit weary of the RFC (even though I’ve translated more than half of it). I’ve barely come here and what I’m seeing is bugs and more bugs, and we’re discussing things like blocking people for reverts. If I can’t even be sure my edit is in the system how can blocks based on reverts be fair?—Al12si (talk) 02:28, 27 July 2014 (UTC)

Participation requirements[edit]

How many people need to participate in this RFC (that I do not support at all) in order to be considered valid? And why is voting permitted when it has not even considered all the possible options and the preliminary discussion is not over yet?--Micru (talk) 10:53, 25 July 2014 (UTC)

I was afraid to set a minimum because it would likely be arbitrary, but if one is needed, each should receive at least ten support votes if unopposed, or fifteen if opposed; at least 70% is required in each case. This would be enforced by the closer of the RfC.--Jasper Deng (talk) 18:54, 25 July 2014 (UTC)
Since often a piece of text isn’t really understood until it’s translated, I’ve been wondering about several things, and something that is kind of pertinent to translation is: When is the first stage considered started/finished? If anyone is going to depend on translations to make a decision, how can discussion/voting start (and looking at how much progress French is making maybe even finish) before translation is finished?
Sorry if this is a stupid question. It probably is but I am new here and things do look a bit opaque to me.—Al12si (talk) 03:04, 27 July 2014 (UTC)

New initiative to promote friendliness >> Wikidata:Lounge[edit]

Worried about the harshness that lately has been developing, I have started a new initiative to counter that by promoting more dialogue, civility, and friendliness >> Wikidata:Lounge

When you are with friends you don't need to put any rule or condition, because they are people you like to be with. If each one does a little effort to become someone who everybody likes to be with, then all harshness will disappear and Wikidata will become even more wonderful \o/ And even if we don't reach utopia tomorrow, at least we'll be on the right direction :)

I would like to invite everyone to participate in this initiative and to come up with new ways to reduce conflict before it happens. --Micru (talk) 13:25, 25 July 2014 (UTC)

@Micru: Is this like the Teahouse? --Jakob (talk) 01:32, 26 July 2014 (UTC)
@Jakec: Yes, and no. The teahouse is mainly focused on mentoring beginers, the lounge is more focused in mentoring ourselves about how to create a positive environment where everyone can feel welcome to.
The aim is not to rely so much on rules, but to create dynamics that will be more effective that any rule. That will affect beginners, perhaps not as directly as a Teahouse would, but if we manage to welcome anyone offering help, with nice words, and with guidance instead of with policies, that will be a big win for everyone.--Micru (talk) 10:12, 26 July 2014 (UTC)
A few thoughts from sister project en.wn. Sorry if I ramble a bit; Big Picture is hard to come by.
Bravo to thinking of a wiki as a self-organizing system. I firmly believe this. I've observed the most prominent rules matter for their psychological influence, but that's a lot different from thinking of a large body of rules as if it were a program to be followed (aka bureaucracy). The two most influential priniples on Wikipedia, as best I can figure atm, are Assume Good Faith and Be Bold, neither of which has primarily the effect it was intended to produce. (The best laid schemes o' mice and men gang aft a-gley.) en.wn has explicitly rejected AGF since way back (from the start as best I can figure, but I came in a few years after that), and has Never assume instead. Which for the past few years (since explicitly drafted) seems to be working pretty well; we maintain a pleasant working atmosphere much of the time — but troublemakers used to being tolerated on Wikipedia are likely to go back to Wikipedia and report that Wikinews has an unfriendly atmosphere. One of a variety of examples I've noted of social problems on Wikipedia spilling over to sister projects. That's something to watch out for: you may well find, if you trace social problems back to their source, that their source is (in one sense or another) on Wikipedia. I don't altogether know what one does about that, in general, but being aware of it may be of use.
Being more specialized than Wikipedia may be socially valuable, as it gives a stronger sense of common purpose; Wikipedia is at a disadvantage in this regard, as its goals are comparatively nebulous (though its technical challenges may be less than those of more specialized sisters). I have considerable sympathy for this social challenge facing Wikipedia; though as a practical matter that makes me no less wary of their problems spilling over.
I'll be most interested to see how your Lounge experiment develops. Good luck! :-)  --Pi zero (talk) 11:50, 26 July 2014 (UTC)
@Pi zero: Thanks for sharing your insights and your nice words! As you I also think that "bureaucracy" is not the right answer to the human condition, because bureaucracy is the assumption that there is always "something wrong" with humans that has to be "repressed" for they to fit a mold. And I totally agree with that brilliant quote, it just becomes an arms race from which it is no longer possible to escape. The only winning move is not to play.
In our case the most effective axiom that we could use is: everyone can be a perfect person, they might not know how to act as such yet. And of course never assume that you are perfect yet (I am not there either!). That shifts the perspective, because that jerk could be a very nice person, he or she doesn't know yet how to act properly. And if we want others to learn how to reach that level, first we should work on ourselves to be a model for them.
Of course we are always limited by time, and we know too little of the people that is behind the nick. Is it worth to be patient with them? And is it worth to spend time improving oneself? I think that can only be answered by getting to know each other better and improving trust in others and in our own actions. Avoiding rules is as scary as removing the training wheels. Yes, you will fall a couple of times until you learn how to do without, but without removing them you will never know what is really like to ride a bike :)
IMHO, Wikipedia is not the "root of all evils", just the consequence of living in a world full or rules and bring those "bad practices" into the digital world. Governments usually have no trust that you can be a good person, or that you can learn how to be one, or that by being in other environments it would be easier to become one.
The Lounge will be only as successful as we want it to be. If we start reading articles or books about how to improve practical wisdom, we apply those new principles into our daily actions, and learn what works or what doesn't by experience or by dialogue, all that will act as a snowball effect much more effective and satisfying that any of the thousands of rules that wikipedia has developed over the years.--Micru (talk) 14:25, 26 July 2014 (UTC)
@Micru: Oh, Wikipedia for all its faults is a great thing because of all the great people who contribute to it. The flaws in its infrastructure have had lots of time to accumulate their effects, and those on the inside of it often can't see the problems because they're on the inside. Contributing to other sister projects, very different from Wikipedia, has given me breadth of perspective, and some glimpses of insight into Wikipedia, that I wish I knew how to apply to helping Wikipedia improve its infrastructure; but that big an infrastructure has a lot of momentum, and I've got plenty on my plate as it is.
It's interesting you should say that about people perfecting themselves. It's sometimes occurred to me that Wikipedia tries to attract users and then use them largely as they are, whereas Wikinews, with its technically demanding tasks, invites users to come and learn how to contribute. My own major initiative on Wikinews over the past few years has been developing software tools to make wiki pages interactive, for the express purpose of then using interactive wiki pages to teach-and-assist users. --Pi zero (talk) 21:42, 26 July 2014 (UTC)

This is a good idea. The problem with Wikidatans is often a language issue, since I think most of us come from home wikis where we tend to hang out in one language. Commons has a similar problem, but because it's so visual, it's a different dynamic. Here the way the data links up (or not) is sometimes hard to imagine and articulate. Jane023 (talk) 12:35, 26 July 2014 (UTC)

@Jane023: Totally true! The only hope is that we can reach a critical size of people who can spread it into other language communities... or that automatic translation becomes much better...--Micru (talk) 14:25, 26 July 2014 (UTC)

I've added myself... great idea Micru! --AmaryllisGardener talk 22:34, 26 July 2014 (UTC)

Misrepresentation of edits and the respective users[edit]

@Lydia Pintscher (WMDE): User:Vlsergey did not change any contribution made by User:Tamawashi to this item. User:Tamawashi did not claim Barack Obama to be the designer of AK-47. So, why does it say "Undid revision 146609291 by Tamawashi" and looking at the prev-link Tamawashi is heavily misrepresented?

As of 2014-07-26 15:50 shows

(cur | prev) 07:13, 22 July 2014‎ Vlsergey (talk | contribs)‎ . . (20,745 bytes) (-8)‎ . . (Undid revision 146609291 by Tamawashi (talk)) (undo | thank)
(cur | prev) 01:34, 22 July 2014‎ Tamawashi (talk | contribs)‎ . . (20,753 bytes) (-2)‎ . . (‎Changed claim: Property:P279: Q177456) (undo) (restore)
(cur | prev) 20:48, 21 July 2014‎ Deaaaa123 (talk | contribs)‎ . . (20,755 bytes) (-4)‎ . . (‎Changed claim: Property:P287: Q76) (undo | thank) (restore)
(cur | prev) 20:48, 21 July 2014‎ Deaaaa123 (talk | contribs)‎ . . (20,759 bytes) (+2)‎ . . (‎Changed claim: Property:P279: Q14955393) (undo | thank) (restore)
(cur | prev) 20:48, 21 July 2014‎ Deaaaa123 (talk | contribs)‎ . . (20,757 bytes) (0)‎ . . (‎Changed claim: Property:P739: Q277462) (undo | thank) (restore)
(cur | prev) 20:47, 21 July 2014‎ Deaaaa123 (talk | contribs)‎ . . (20,757 bytes) (+12)‎ . . (‎Changed claim: Property:P1092: 75,000,000,000±1) (undo | thank) (restore)
(cur | prev) 16:04, 5 July 2014‎ Vlsergey (talk | contribs)‎ . . (20,745 bytes) (+1,060)‎ . . (‎Updated item) (undo | thank) (restore)

The prev link for the first, having text "(Undid revision 146609291 by Tamawashi (talk))" is

The prev link for the second, having text "(‎Changed claim: Property:P279: Q177456)" is

Tamawashi (talk) 15:53, 26 July 2014 (UTC)

merging occupation items[edit]

there are many duplicated occupation items which are listed in here please merge them by bot. (eg. Q2304816 and Q716711)Yamaha5 (talk) 05:47, 27 July 2014 (UTC)

Not sure if that is a good sample. Clearly those two (Q2304816 and Q716711) shouldn't be merged, --- Jura 06:00, 27 July 2014 (UTC)
Excuse me I linked by fault eg. Q14616727 and Q13381753 Yamaha5 (talk) 06:12, 27 July 2014 (UTC)

Wikidata:Requests for comment/Admin conduct policies[edit]

Since recently some admins arbitrarily, without coverage by policies, used their right of blocking a contributor, I created Wikidata:Requests for comment/Admin conduct policies. Tamawashi (talk) 14:28, 27 July 2014 (UTC)

Not a real RfC; just a way to cause trouble. The questions is poses are already policy. I recommend you close it/request it for deletion or an uninvolved admin closes it/deletes it. It's just a case of 'I was blocked, I want the admins involved to be desysop'd immediately because the community has no policy regarding it!' John F. Lewis (talk) 14:34, 27 July 2014 (UTC)

What makes an RfC a "real RfC"? Do you disagree with the initial text:

Admins should only use their specific admin rights if the usage is covered by policies.

Admins should loose their specific admin rights if they violate that rule.

The guideline(sic) named "Wikidata:Blocking policy"(sic) should be upgraded to a policy.


This It's just a case of 'I was blocked, I want the admins involved to be desysop'd immediately because the community has no policy regarding it!' is defamation. Tamawashi (talk) 14:45, 27 July 2014 (UTC)

It's not defamatory. Also the first line basically says we can't; block, protect, revision deletion etc. because we have no policies regarding them. The second line is already policy and the third point requires a dedicated RfC for it. John F. Lewis (talk) 14:52, 27 July 2014 (UTC)
  1. It is defamatory according to Google's definition [10] "damaging the good reputation of someone; slanderous or libelous.".
  2. Fixed by creating policies. Real life analogy: Give the rules to the policy and the army.
  3. Where is that policy?
  4. Agreed. Done already.
Tamawashi (talk) 15:19, 27 July 2014 (UTC)
Tamawashi, if you answer harshness with more harshness, there will be no way out of the conflict. If you feel that the other rfc was a provocation, there is no use to answer it with another rfc that also can be seen as a provocation. It just emphasizes the conflict, but doesn't offer any way out of it.--Micru (talk) 17:04, 27 July 2014 (UTC)

@John F. Lewis, Jasper Deng: - To clarify: I don't want that you both loose your admin rights. I only want that your admin actions as well as those of any other admin are bound by policies. Tamawashi (talk) 15:19, 27 July 2014 (UTC)

I also want us to resolve these conflicts; but you have to see it from our point of view. Since the block; you've been persistent about commenting on administrator's actions, wanting to make swift immediate changes basing things around our actions and claiming we're violating nonexistent policies. Please read Wikidata:Administrators. John F. Lewis (talk) 15:27, 27 July 2014 (UTC)
As a matter of peripheral interest, en.wn has an actual blocking policy, which starts
In certain circumstances it may be necessary for an admin to block a user or IP address in the best interests of the site. It is up to admins to use their discretion to decide when to block, and how long for, however for guidance: [...]
So everything else on the page is guidelines. Imho that's the sanest blocking policy around. Saves wikilawyering. If an admin is a reasonable person, reasonable dialog is possible; if not, if they're truly a problem, the community can take away their privs. --Pi zero (talk) 16:23, 27 July 2014 (UTC)
Pi zero, even not liking the word "policy", I agree that version is the sanest one we could have. I think the admins are more reasonable when they do not follow any specific criteria other than their own experience, but that also requires users to understand the responsability weight that they carry and do not force drama upon them.--Micru (talk) 17:12, 27 July 2014 (UTC)

Wikidata:Requests for comment/Blocking policy[edit]

I started Wikidata:Requests for comment/Blocking policy with the initial text:

There is a page named "Wikidata:Blocking policy" which is labeled as "guideline".

  • There are admins that claim that Wikidata has no blocking policy.
  • There are admins that claim there is one.
  • Admins refer blocked users to that page, as if it would be a policy.
  • Admins refer blocked users to that page as if it contained all rules for blocking.
  • Other admins refer blocked users to Wikidata:UCS "use common sense".

The proposal is either

  1. to make this page a policy or
  2. to rename this page to "Wikidata:Blocking guideline" or
  3. to rename this page to "Wikidata:Blocking" and mark as an essay.

Tamawashi (talk) 14:42, 27 July 2014 (UTC)


It seems that in wp:az, there can be two articles about the same topic, one in arabic alphabet and the other in latin alphabet. The result is this : Q4361548 and Q12846577. I don't know what to do in Wikidata with this; for now these two items can't be merged. Ideally the problem could be solved directly in az wikipedia. Louperivois (talk) 15:07, 27 July 2014 (UTC)

This may apply to other multi-script Wikipedias, e.g. WP:SH too. Maybe User:Lydia Pintscher (WMDE) can let the developers create linking possibilities for language and script. For az: 1) az-Latn 2) az-Arab. For sh: 1) 1) sh-Latn 2) sh-Cyrl. Tamawashi (talk) 15:23, 27 July 2014 (UTC)

Nationality of companies[edit]

Gränges (Q10509721) had the properties, country (P17) Sweden (Q34) added yesterday.

My Question is: What makes the nationality of a company?

  • Which jurisdiction it's main office/board or directors is related to?
  • Where it was founded?
  • Where is has it's main production?
  • The nationality of the owners?
  • The nationality of the company culture?

As far as I know, the main office of this company had until recently it's main office in Oslo, Norway. (Now it's in Stockholm, Sweden). The owners are Norwegian. The juridical peron of Gränges of today was founded in Norway, but the original history of the name "Gränges" is Swedish. The production is mainly in Sweden and PRC today. The result is reported in Norwegian money, NOK. The company will soon be listed in a stock market, most likely in Nasdaq, Stockholm. Thoughts? @Väsk:.

Another example is IKEA. Many think of this Dutch company as Swedish, why? -- Innocent bystander (talk) 16:50, 27 July 2014 (UTC)

I can't say I disagree with any of the points you raise. The nationality of companies can certainly be ambiguous and it has probably been debated several times before on Wikipedia. However, I think companies should be tagged with nationalities. I also think you should be able to tag companies with a nationality that isn't necessarily shared with the city where its head office is located. I. e. you should be able to tag Ikea as a Swedish company even though they have established their head office in the Netherlands and their brands are owned by a Lichtenstein-based foundation. There are lots of companies that have relocated their head office to London or the Netherlands for tax reasons, yet they were founded and established elsewhere. Modern corporate governance allows for a great variety of structures which further confuses the issue. There obviously needs to be some debate about ambiguous cases that should lead to guidelines.
As for the specific case: The reason I tagged Gränges (Q10509721) as a Swedish company was that it was categorised as such on the Swedish Wikipedia and that its history there was mostly Swedish. Locking back at that article it is clear that it is about two different companies, one historical Gränges that operated between 1896 and 2005, and a later company that changed its name from Sapa Heat Transfer to Gränges in 2013. This is not an uncommon practise, but can be confusing. Wikidata should obviously have separate entries for these companies, Gränges (Q10509721) should be about the historical company while a new entry can be created for the former Sapa Heat Transfer. Väsk (talk) 19:03, 27 July 2014 (UTC)
I'm not saying that you or anybody have done something wrong anywhere. But we need to find a way to be consistent in this matter. We have to say in what way IKEA and Gränges is/was Swedish/Dutch. Separating old Gränges from the new is one way, but it does not solve every issue. In the case with IKEA, we maybe have to separate the different parts of this company? -- Innocent bystander (talk) 12:15, 28 July 2014 (UTC)
AIUI, Wikidata is time neutral, so you can also include historical nationalities. One solution could be to use qualifiers. For example, Ikea could be classified something like country (P17) Sweden (Q34) [because: country where it was founded] and country (P17) Netherlands (Q55) [because: location of head office]. I'm not entirely certain if the terminology for that exists yet or not. I don't think it gives complete picture to say that Ikea is only Dutch or only Swedish, but I agree that we should try to be consistent. I don't think we should split up IKEA (Q54078) in this case. Väsk (talk) 17:58, 28 July 2014 (UTC)
I think that in these cases companies need to have multiple values for this property with a qualifier like Wikidata:Property_proposal/Event#specifically used to indicate in what way they are associated with each country. Feel free to support the creation of that property if you agree. Filceolaire (talk) 08:40, 30 July 2014 (UTC)

Be generous in what you accept...[edit]

When entering the value of a property, it should be possible to paste in one of these formats:

  • Prince_of_Wales's_Own_Civil_Service_Rifles

and for the software to understand that, and convert it to canonical forms. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 17:37, 28 July 2014 (UTC)

We can't accept the last one because we have no idea where it is coming from. The other two would be nice to accept but to be honest right now we have much bigger fish to fry before we can get to things like this. --Lydia Pintscher (WMDE) (talk) 19:05, 28 July 2014 (UTC)
I was thinking about the first one ( some time ago. That should be quite easy to do with a bit of javascript in the interface. Probably in the same block of code that takes "Q6942600" and shows the label and description to the user. Would love to have that. Lydia do you happen to know if we already have a bug open for that? Multichill (talk) 07:25, 29 July 2014 (UTC)
I don't think we do. --Lydia Pintscher (WMDE) (talk) 07:37, 29 July 2014 (UTC)
@Lydia Pintscher (WMDE): I think you misunderstood; where we accept Prince of Wales's Own Civil Service Rifles (without underscores), we should accept Prince_of_Wales's_Own_Civil_Service_Rifles (with underscores). Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:53, 29 July 2014 (UTC)

"Uncle" in Turkish and Bosnian[edit]

It appears that Bosnian and Turkish may have different words for an uncle who is your mother's or your father's brother. The "mother's brother" uncle links are at Q6041134 on their own but the "father's brother" links are with the other links for "uncle" in general at Q76557. Should they be separated out? And if so, what would the labels be in languages that just have one word? Perhaps "paternal uncle" and "maternal uncle"? Cheers. Delsion23 (talk) 18:52, 28 July 2014 (UTC)

Chinese also makes this distinction. In fact we make a three-way distinction: paternal uncles younger than your father (Q10912818), paternal uncles older than your father (Q10885666), and maternal uncles.—Al12si (talk) 19:10, 28 July 2014 (UTC)
The same is in Swedish, I think: morbror, farbror. Matěj Suchánek (talk) 19:25, 28 July 2014 (UTC)
True, and also in Finnish: setä and eno. --Stryn (talk) 19:31, 28 July 2014 (UTC)

Seeing as Turkish and Bosnian have articles for both, I've kept the maternal uncle articles at Q6041134 and moved the paternal uncle articles to Q17438502. I've added a statement to both that they are "subclass of = uncle". Delsion23 (talk) 21:16, 28 July 2014 (UTC)

There’s already an entry for “father’s brother” (Q13907976, the parent of Q10912818 and Q10885666). I suppose the two should somehow be merged?—Al12si (talk) 22:09, 28 July 2014 (UTC)
✓ Done Delsion23 (talk) 22:52, 28 July 2014 (UTC)

Wikidata Main page needs a banner and you can help![edit]

Hi all,

As you may know, I'm currently working on planning and implementation of a new Main page for Wikidata. One idea to already come out of the redesign process is that the Main page should be more visually appealing.

To that end, I am wanting to get together an awesome banner (or choice of banners) for the top of the Main page and am inviting you all to share your ideas and submissions. We are looking for something that represents Wikidata and the idea behind the project; is eye-catching but also easily translatable; and is freely licensed CC-BY (or equivalent).

If you think you know how to communicate what Wikidata is through images or have an eye for graphic design, navigate over to Wikidata:Portal_Redesign/Banner where I've already started brainstorming designs and to learn more about the open call for submissions. Please add your proposal (or a link to it) to the talk page.

We hope to have our banner or banners ready for use on the new Main page in two weeks from now so please have all submissions in by 19:00 UTC on August 11th 2014. Feel free to leave questions here or on my talk page.

Cheers, -Thepwnco (talk) 19:03, 28 July 2014 (UTC)
ps. thanks to those who have already provided feedback on the Main page redesign—for those who haven't, there is still time to do so!

Premierships and presidencies confused with persons[edit]

I've done a search over premiership data elements (you can also do a search for presidencies, same result). I've found elements like Q7240404, which show "alma mater", and a gender. A presidency has neither, the president has both. It has a start date and an end date, but no birth date. And it is also no member of a political party. Even if you say that "birth date" can be interpreted as "start date" of the presidency, the data are wrong. For example Q7240396. If you compare them with the dates on Q244689, you will find they are the same, so the entry for "birth date" does not show the start date of the presidency.

Non-exhaustive list of "suspects" is: User:RobotGMwikt, User:GerardM (Operator of that bot), User:GPUBot and User:Dexbot (evidence). I think the bots are doing good work most time, but their creators weren't aware of this issue. I'll contact them. Muelleum (talk) 21:01, 28 July 2014 (UTC)

I think it's because the articles themselves on enwiki aren't just using categories for people (variants of en:Template:Infobox person), but also for some idiotic reason categorized as people – en:Premiership of Tony Blair is in en:Category:1953 births and Category:Living people. This leaves the bots with little chance to understand that the item is not in fact a person. Jon Harald Søby (talk) 05:40, 29 July 2014 (UTC)
It all comes from the harvesting of data from templates like en:Template:Infobox officeholder. For example, Q7240401 contains the information in the infobox on en:Premiership of Stephen Harper. The same method would have been used to populate Q206 with the data from the infobox in en:Stephen Harper. Most of the time, of course, the bots get it exactly right. But in this case I think the bots need some sort of whitelist for the articles that appear in en:Category:Tenures in political office by individual. Another example for bots to be careful with would be articles about murders and assassinations which frequently contain infoboxes for victims and perpetrators. 08:05, 29 July 2014 (UTC)

For my bot I use categories to see the article is about a person or not. I can exclude some regexes in title. I will fix it Amir (talk) 18:51, 30 July 2014 (UTC)

Song Contests[edit]

I'm quite new to Wikidata, and find it really interesting! I have been working on different song contests, such as the Eurovision Song Contest 2013 (Q10162). Those contest should be about songs competing, but now the artists performing the songs are listed as participant (P710). How can both the song and the artist be structured under such a song contest? Example: Emmelie de Forest (Q3720656) performed the song Only Teardrops (Q3739483) during Eurovision Song Contest 2013 (Q10162). //Mippzon (talk) 21:07, 28 July 2014 (UTC)

You will need to use qualifiers for the other information about each participant. We have country (P17) and points for (P1358) but I don't think we have a property we can use for the song so you may need to propose a new property. New properties are more likely to be approved if you can show that they can be widely used.
Alternatively you could have 'Participant' link to the song and use performer/musical artist (P175) to link to the singer, or propose a new property for competitions where the entrants are not people.
To search for existing properties put P: at the start of the search term or go here
Hope this helps Filceolaire (talk) 08:22, 30 July 2014 (UTC)
ESC's intention was to be a competition of composers and song writers. I agree that listing the performers as participants (which is the current state) is wrong. Formally, ESC can also be regarded more like a competition between the members of the European Broadcasting Union. However, just as there may be more than one composer of a song, there may be more than one broadcasting organization representing a country. Consequently, to me, it would make most sense to list the countries as participants and add all detailed information as qualifiers. Proposal:
STATEMENT: participant (P710): Denmark (Q35) / QUALIFIERS: represented by (or some already existing property I am unable to find once again): DR (Q1164334), represented by: TV 2 (Q1616154) (in my opinion, the broadcasting organizations would need to be listed as those may not be the same in every year), song (yet again unable to find a correct existing property): Only Teardrops (Q3739483) (that item should feature the composer), performer/musical artist (P175): Emmelie de Forest (Q3720656)
One thing I noticed along the way is that from the item Only Teardrops (Q3739483) one could get the impression that it represents the specific combination of the composition with the performing artist (kind of saying "Emmelie de Forest's version of 'Only Teardrops'"). With such a schema, there would be no need to list the performing artist as qualifier above. However, since a composition might very well be performed by someone else as well, items should not feature specific combinations of song and artist (think of some Jazz-Standard having been performed by thousands of artists); (if Only Teardrops (Q3739483) is performed by some additional performer, (s)he should be listed in that item as performing artist as well.)
Okay, that would cover the most basic information. But what if you are asking for trouble and want to capture all the scores - the individual ones assigned by each country, not just the final scores? Well, then I would suggest to basically forget all the above and just use the following statement (which sounds quite ugly though):
STATEMENT: participant (P710): Denmark in the Eurovision Song Contest 2013 (Q4809098). (Alright, maybe there is a more appropriate property, but it is about finding such...)
Denmark in the Eurovision Song Contest 2013 (Q4809098) would act as container for all Denmark at the Eurovision Song Contest 2013 related information. Points, artist, song and whatever. In my opinion, that would be the most sane and flexible solution. Random knowledge donator (talk) 09:14, 30 July 2014 (UTC)
Yeah, I think that structure with the country as participant (P710) is a good choice. Then it's easy to add appropriate qualifiers under each country. But is it not possible to add song (Q7366) as a qualifier? //Mippzon (talk) 10:45, 30 July 2014 (UTC)
The structure with the country looks good at first sight, but if you want to track all points, e.g. how many points Sweden awarded to Denmark, you run into trouble and the second solution - simply linking to a "container" item - would be far less of a hassle - the syntax is not as expressive and obvious though...
song (Q7366) is an item which can be used as value in statements. What we need is a property. There is a property audio (P51) which has an alias "song" - but that is of data type "Commons media file". We need a property of type "Item", so an item - Only Teardrops (Q3739483) - can be linked. Random knowledge donator (talk) 11:31, 30 July 2014 (UTC)
I have been experimenting on a structure in Eurovision Song Contest 1996 (Q207815). Should the property be something like "performed song"? Like Emmelie de Forest (Q3720656) has property "performed song" and all the songs she performed listed? //Mippzon (talk) 12:53, 30 July 2014 (UTC)

Items with no interwiki[edit]

I have 2 questions:

  1. How can i find list of items which doesn't have interwiki link like Q17143521?
  2. How can i find list of items which are used with Property:P31 for example Q2176891 used with Property:P31 in Q17143521?

I want to translate these labels to use the in local wikiYamaha5 (talk) 21:17, 28 July 2014 (UTC)

Question 2: I don't understand the question because it sounds like you want to search for all items where P31:Q2176891, but in your example of Q17143521, it does not link to Q2176891 anywhere, and neither does Q2176891 to Q17143521. Did you copy and paste the right link?
In any case, probably you want to use Autolist. This is the newer version: --Haplology (talk) 06:16, 29 July 2014 (UTC)
Question 1: Special:ItemsWithoutSitelinks. If you are looking for items that don't have a sitelink and no labels and descriptions, there's User:Pasleim/Items for deletion/Almost empty. Bonus answer: If you're looking for items not having a certain sitelink (e.g. no enwiki link), there's Wikidata:Database reports/Missing links or the Terminator. If you're looking for items with only one sitelink, there's Lonely Interwiki links and Lonely items. --YMS (talk) 08:02, 29 July 2014 (UTC)
I want to find all items (like Q2176891Q3504085) which are used with P31 like P31:Q2176891 P31:Q3504085 . Yamaha5 (talk) 16:31, 29 July 2014 (UTC)
instance of (P31) Gmina Dziemiany (Q2176891) makes no sense. Michiel1972 (talk) 16:53, 29 July 2014 (UTC)
Excuse me I mean instance of (P31) :rural municipality of Poland (Q3504085) Yamaha5 (talk) 17:19, 29 July 2014 (UTC)
In that case, you want this:
Hope that helps :) --Haplology (talk) 01:05, 30 July 2014 (UTC)


I'm struggling to work out the subtle difference between division (Q4161240) and division (Q169534), can anyone who speaks the languages help please? Thank you. 08:33, 30 July 2014 (UTC)