User talk:ArthurPSmith

From Wikidata
Jump to navigation Jump to search

See User talk:ArthurPSmith/Archive for older discussions.

Contents

chart of nuclides[edit]

Hi ArthurPSmith, do you know why some isotopes are not displayed in the chart of nuclides? Pamputt (talk) 10:13, 2 July 2017 (UTC)

uranium-235 (Q848497) looks like a lone mistake.
Other uraniums have following claims:
uranium-235m (Q18888906), uranium-236m2 (Q18888908), uranium-237 (Q18845543), uranium-238m (Q18888909), uranium-239 (Q18845544), uranium-239m1 (Q18888910)
d1g (talk) 15:39, 3 July 2017 (UTC)
By removing the instance of (P31) statements you have cut out something like half the nuclides from this chart: https://tools.wmflabs.org/ptable/nuclides that is autogenerated based on a wikidata SPARQL query. I don't even know if what the subclass of (P279) statements you have replaced them with will allow for the same sort of chart to be autogenerated now. Plus what you have done is obviously inconsistent since things still work for some but not for others - and in a case like U-235 it wasn't findable at all after your edits. Please at least be more careful on this sort of thing. But at the moment I don't even know a practical way to solve it and get the chart working again. ArthurPSmith (talk) 16:11, 3 July 2017 (UTC)
(P31|P279) should fix everything. P31 is not the most correct property for physical classes
Where query could be located? d1g (talk) 17:32, 3 July 2017 (UTC)
The query was looking for all instances of subclasses (wdt:P31/wdt:P279*) of isotope (Q25276). Replacing P31 with P279 there will not work, because that leaves no way I can see (right now) to distinguish between an actual nuclide such as uranium-235 (Q848497) and the generic classes such as isotope of uranium (Q1369686). The ptable code is on gerrit if you want to look at it: see the instructions here - ptable is under labs/tools/ptable. ArthurPSmith (talk) 19:00, 3 July 2017 (UTC)
My understanding was that isotope can be only of some element, so we can have these relations using fixed number of P279 links.
Query below should return everything to how it was before edits.
Part on below contributes only 33 "missing" items. d1g (talk) 21:07, 3 July 2017 (UTC)

The following query uses these:

  • Properties: subclass of (P279) View with Reasonator View with SQID, instance of (P31) View with Reasonator View with SQID
    1 SELECT DISTINCT ?item ?itemLabel WHERE {
    2   { ?item wdt:P279/wdt:P279 wd:Q25276 } # suggested structure with fixed links, 4478; DISTINCT 4476
    3   UNION
    4   { ?item wdt:P31/wdt:P279* wd:Q25276 } # old one with arbitrary depth, 2391; DISTINCT 2265
    5   
    6   # BOTH UNION: 6869
    7   # BOTH UNION, DISTINCT: 4509
    8   SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
    9 }
    

Ok, so maybe it can be fixed, maybe not. Let us see. On a semantic point of view, would it be possible to use instance of (P31) : nuclide (Q108149) on each item in order to identify them clearly and to use a proper query? Pamputt (talk) 22:00, 3 July 2017 (UTC)

  • I'd be happy with that solution - it definitely seems more robust than depending on there being exactly two subclass levels below isotope. ArthurPSmith (talk) 22:06, 3 July 2017 (UTC)
  • @Pamputt, D1gggg: And D1gggg seems to be busy making the nuclides app completely worthless by continuing his P31 removal spree: Special:Contributions/D1gggg - could you at least WAIT until we have settled this discussion on how we are going to address the problem and actually had a chance to fix the code in ptable and see it working, before making further mass edits? I don't find your actions here very friendly, given that you KNOW they have already caused a problem, and now you are making it worse. ArthurPSmith (talk) 22:14, 3 July 2017 (UTC)
  • @Pamputt, D1gggg: I've been trying to rationalize our ontology a bit - this is still very much a work in progress, but I would appreciate your thoughts on my draft ontology for elements, nuclides and chemicals. Basically, I think every specific element and isotope should be treated as leaf nodes in the subclass hierarchy - even though they are abstract, I don't think it makes sense to talk about them having "instances". A "lithium atom" has instances, but there is nothing that "is a" "lithium". ArthurPSmith (talk) 18:27, 10 July 2017 (UTC)

isotopes like neon-24 are similarly NOT classes, but non-instantiable abstract concepts.

@ArthurPSmith: overall I agree, but you should word it using items, not words:
subject > instance of (P31) View with SQID < no label (Q26841207) View with Reasonator View with SQID >
claims should result in "physical object" using following statements
or
Did you mean this or something different? d1g (talk) 05:19, 11 July 2017 (UTC)

"A is a period 2" or "A si a group 18" also makes no sense - "A is a period 2 element" however would be fine.

I also think
subject > part of (P361) View with SQID < period 2 >
would be more natural, not using P31/P279. d1g (talk) 05:29, 11 July 2017 (UTC)
d1g - are you seeing some sort of difference between uranium-234 and neon-24? I don't think anything can be an instance of uranium-234. Something can be an instance of a uranium-234 atom, or a uranium-234 nucleus, and so "uranium-234 atom" for example should be a subclass of "physical object". But "uranium-234" itself should not be a subclass of anything. As to using items etc., I do intend to articulate this more concretely in terms of wikidata syntax, I haven't really gotten to that fully yet. ArthurPSmith (talk) 15:30, 11 July 2017 (UTC)
@ArthurPSmith: maybe they aren't different from standpoint of chemistry; then we need to relate other items to "physical" items.
I think that "conglomeration of mass" is what we frequently mean by "physical" "physically" in natural language.
Second definition could be "anything that explainable using laws of physics" (item similar to mathematical object (Q246672)) - i.e. something chemistry wouldn't explain in detail d1g (talk) 06:10, 12 July 2017 (UTC)

Tobias1984
Snipre
Physikerwelt
Pamputt
Petermahlzahn
Jibe-b
Restu20
Daniel Mietchen
TomT0m
ArthurPSmith
Mu301
Sarilho1
SR5
DavRosen
Danmichaelo
Ptolusque
PhilMINT
Malore
Pictogram voting comment.svg Notified participants of WikiProject Physics Saehrimnir
Leyo
Snipre
Jasper Deng
Dcirovic
Walkerma
Egon Willighagen
Denise Slenter
Daniel Mietchen
Andy Mabbett
Kopiersperre
Emily Temple-Wood
Pablo Busatto (Almondega)
Nothingserious
Antony Williams (EPA)
TomT0m
Wostr
Devon Fyson
User:DePiep
User:DavRosen
Benjaminabel
99of9
Kubaello
Fractaler
Sebotic
Netha
Hugo
Pictogram voting comment.svg Notified participants of WikiProject Chemistry

See Atom classes.svg. You don’t need any kind of abstract uninstanciable concept. At worse you could have
< hydrogen > union of (P2737) View with SQID < list values as qualifiers (Q23766486) View with Reasonator View with SQID >
of (P642) View with SQID < hydrogen pure substance >
of (P642) View with SQID < hydrogen atom >
to reflect the polysemy. You also have
< hydrogen pure substance > has part (P527) View with SQID < hydrogen atom >
. What more is needed ? where do your definitions come from to argue ? author  TomT0m / talk page 17:25, 11 July 2017 (UTC)
TomT0m that is what we have been doing until now, but I believe it to be fundamentally wrong based on the way these terms are defined by the wikipedia pages. You suggest "hydrogen" is the union of "hydrogen as a pure substance" (or "simple substance" I think is the term used here) and the "hydrogen atom". I believe it is a lot more than that. Go to en:Hydrogen; how much of that page is about either of those two meanings for "hydrogen"? A lot of the page discusses hydrogen's abundance in the universe in atomic and ionized states, chemical properties within other molecules, ions and hydrides, its combustion and production, its uses in semiconductors, its biological relevance, the isotopes also, etc. As to the diagram you post, I agree almost entirely with the left-hand side: hydrogen atom should be a subclass of atom, germanium atom should be a subclass of atom, that is correct. Rather than "tritium", the correct item there should be "tritium atom", which would be a subclass of "hydrogen atom". That is all fine - their instances are specific atoms which are physical objects. The problem is the relations between center panel and the right column. "hydrogen atom" is NOT an "instance of" "element". "hydrogen atom" is NOT the same as "hydrogen", and it is "hydrogen" that is the instance of "element". en:Hydrogen atom talks about quite different things than en:Hydrogen, as far as definitions go. ArthurPSmith (talk) 17:44, 11 July 2017 (UTC)
@ArthurPSmith: Please follow the link en:chemical element in « hydrogen is a chemical element » in en:hydrogen. You’ll see that it’s a species (a synonym of subclass) of atoms. This fits with hydrogen beeing a subclass of atom. all atoms with the same number of protons in the atomic nucleus.  »author  TomT0m / talk page 10:09, 12 July 2017 (UTC)
@TomT0m: Does "species" mean subclass? I don't think so. In the third paragraph of the article "chemical element" it states "The term "element" is used for atoms with a given number of protons (regardless of whether or not they are ionized or chemically bonded, e.g. hydrogen in water) as well as for a pure chemical substance consisting of a single element (e.g. hydrogen gas)." An element is not simply a subclass of atoms. And we don't have that in our ontology currently anyway - "neon" is not "subclass of" in any way, "atom" right now. I don't think it should be. The fact that we have separate items for "hydrogen" and for "hydrogen atom", as well as for several other elements, also implies that they are not considered synonymous. ArthurPSmith (talk) 14:38, 12 July 2017 (UTC)
@ArthurPSmith: Please please please use references. I said a thousand time that mine was https://goldbook.iupac.org/html/C/C01022.html . There is two definitions : « all atoms with the same number of protons in the atomic nucleus. » and « A pure chemical substance composed of atoms with the same number of protons in the atomic nucleus. » The first one is an element as a subclass of atoms, the second one is the one of a class of pure substance. The seciond refers to the union I proposed : the term is used for both. There is also definitions (several) of species like https://goldbook.iupac.org/html/C/CT01038.html https://goldbook.iupac.org/html/C/CT06859.html but none actually applies, the first refers to an experiment which is not relevant in the definition of an experiment, the second is about subtypes of elements and don’t apply to the element concept itself. Also note the definition of a taxonomic species : https://goldbook.iupac.org/html/S/S05786.html : « a group of closely related, morphologically and physiologically similar individuals.  » : a species is a subclass of individuals. author  TomT0m / talk page 15:57, 12 July 2017 (UTC)
I also found references to support my interpretation :
1. A species ofatoms  ; all atoms with the same number  of protons in the atomic nucleus. 
2. A pure  chemical substance  composed  of atoms  with  the same  number of protons   in the atomic nucleus. Sometimes this concept is called the elementary substance as distinct  from the chemical element as defined under 1,  but mostly the term chemical element is used for both  concepts.  
La  première  définition  est  retenue  par  les  programmes  scolaires  
français,  elle  se  traduit  par  :  «  Catégorie  d'atomes  :  tous  les atomes  
dont  le  noyau  possède le  même  nombre  de  protons  »
from http://documents.irevues.inist.fr/bitstream/handle/2042/23982/DIDASKALIA_2008_32_117.pdf?sequence=1 : translation of the text is french « the first def is used by every french teaching programs, it translates to « category of atoms, every atoms whose kernel has the same numbes of protons » wgich is exacsly what I say. author  TomT0m / talk page 16:22, 12 July 2017 (UTC)
references that provide definitions using words are unlikely to be precise enough to definitively answer these questions. We are trying to define "subclass" and "instance" (and possibly other) relationships in a way that makes logical sense. What we have now clearly doesn't. In French, does it make logical sense to say "X est un hydrogène"? What do you think that would mean? Whereas I think "X est un atome d'hydrogène" is clear, and different in meaning. So "hydrogen" "subclass of" "atom" does not seem right. "hydrogen atom" "subclass of" "atom" does. If we follow your suggestion, what would we do with the wikidata item for "hydrogen atom"? Merge it with "hydrogen" (but we can't due to the wikilinks)? ArthurPSmith (talk) 19:15, 12 July 2017 (UTC)
"X est un hydrogène" would mean « X is a kind of hydrogen », hence a subtype. It’s meaningless in other acceptations. Why would Wikidata bother about meaningless stuffs ? author  TomT0m / talk page 07:16, 14 July 2017 (UTC)
because that is what P31 means. According to Wikidata:Relation between properties in RDF and in Wikidata, P31 means the same as "rdf:type", and according to the SPARQL standard for example - rdf:type can be abbreviated 'a', as in ?x a :class is the same as ?x rdf:type :class. If "hydrogène" covers "kinds" of hydrogen, rather than just "hydrogen atoms", then a particular hydrogen atom cannot be "a" "hydrogène", and therefore "hydrogène" cannot represent the class of atoms, it must mean something else. ArthurPSmith (talk) 14:11, 14 July 2017 (UTC)

significant updates[edit]

Tobias1984
Snipre
Physikerwelt
Pamputt
Petermahlzahn
Jibe-b
Restu20
Daniel Mietchen
TomT0m
ArthurPSmith
Mu301
Sarilho1
SR5
DavRosen
Danmichaelo
Ptolusque
PhilMINT
Malore
Pictogram voting comment.svg Notified participants of WikiProject Physics Saehrimnir
Leyo
Snipre
Jasper Deng
Dcirovic
Walkerma
Egon Willighagen
Denise Slenter
Daniel Mietchen
Andy Mabbett
Kopiersperre
Emily Temple-Wood
Pablo Busatto (Almondega)
Nothingserious
Antony Williams (EPA)
TomT0m
Wostr
Devon Fyson
User:DePiep
User:DavRosen
Benjaminabel
99of9
Kubaello
Fractaler
Sebotic
Netha
Hugo
Pictogram voting comment.svg Notified participants of WikiProject Chemistry @Pamputt, D1gggg, TomT0m: Please see the latest draft ontology for elements, nuclides and chemicals - I have included both my own version described above, and what I believe TomT0m is suggesting. The actual impact of our differences is not huge, but the details do matter a bit here. Both of them imply the removal of a lot of the existing statements on elements and nuclides, and restoring some of what d1g removed. Please comment on the talk page of the draft, thanks! ArthurPSmith (talk) 20:16, 12 July 2017 (UTC)

Agree on 1.1 and 1.2; I also added (or it should be Q7946). No comments on other parts. d1g (talk) 20:33, 12 July 2017 (UTC)

Just chiming in after the storm: intuitively I would use both P31 and P279 on hydrogen (Q556), probably with different target items. The P31 would link to an element (possibly via P279*) that represent the class of all the "funny squares that you find in the Mendeleiev table" (I think they are called chemical elements but I have no idea if chemical element (Q11344) is suited for that). The P279 would link (again via P279*) to the class of all atoms, or physical particles, or whatever. But I am neither a physicist nor a philosopher. Anyway, it would be useful to be able to isolate all the "cells of the Mendeleiev table" with a simple P31/P279* query, so that we could also state with properties for this type (P1963) all the data that they are supposed to have. And I find it quite natural to treat isotopes as subclasses of these. It seems possible to reuse the best of both approaches in a consistent way. − Pintoch (talk) 16:58, 17 July 2017 (UTC)

still not working[edit]

Hi, I come back to my initial request; there are still plenty of holes in the nuclide chart. I skip the discussion above at some point and I do not understand wether something was decided. If so, could you process the nuclides that are missing in the nuclide chart? Pamputt (talk) 19:45, 23 August 2017 (UTC)

I would appreciate your weighing in on the rough plan, either one of them - essentially the solution will require reverting pretty much all d1g's changes to nuclide P31/P279 statements and while I personally think that's necessary I'm reluctant to do that without more of a weight of opinion from the community here. ArthurPSmith (talk) 20:19, 23 August 2017 (UTC)

Discussion[edit]

You may interesting about Wikidata:WikiProject Association football/Discussion about properties. Xaris333 (talk) 22:39, 4 August 2017 (UTC)

Hi Xaris333 - I've read the discussion there, I have to say I don't know very much about football or really any sports competitions in general, but it seems to me the competition structure you outline at the top makes sense (I assume those are subclass relations) and I think the rest follows logically. I don't understand what TomT0m is saying there about not using P3450, that seems like the right relationship to me. ArthurPSmith (talk) 20:30, 7 August 2017 (UTC)

Identifier classification tree[edit]

Thanks for the recent talk. Maybe this tree helps a bit. Each DE, FR are country codes, specifically ISO 2-letter codes. Each is subclass of ISO 3166-1 alpha-2 code. To inherit all superclass properties it is needed that ISO 3166-1 alpha-2 code is a subclass of country code, which in turn is subclass of UID. 213.39.164.36 23:04, 12 August 2017 (UTC)

Calling "DE" a class seems a very odd way of thinking about it. Also, what "superclass properties" do you find useful for "DE" to inherit? The subclasses can't inherit the external ids (for example the freebase ID for ISO 3166-1 alpha-2 code). There's not much else on any of these wikidata items that seems particularly relevant. More fundamentally, to my mind, is we seem to have a dispute about the meaning of "unique identifier". We can think of a class as a collection, but if everything that is a code used to identify something qualifies to be in the "unique identifier" collection, that collection becomes full of a huge number of duplicate codes that identify different things - i.e. they are very definitely no longer unique. If "DE" is a unique identifier in the collection you are calling the class "unique identifier", how do we know whether it is talking about Germany (ISO 3166-1 alpha-2) or Delaware (FIPS 5-2 US state code) or something completely different? The specific codes are ONLY "unique" within a context of a specific "identifier" - such as ISO 3166-1 alpha-2, which I would view as the collection (class) containing the specific instances of codes. I.e. to me it is very clear "DE" is an instance of "ISO 3166-1 alpha-2" which should be one way or another an "instance of" "unique identifier". But apparently there is disagreement on this. You are advocating for something that makes no sense to me at all - and property inheritance seems terribly unconvincing as a reason. ArthurPSmith (talk) 15:17, 14 August 2017 (UTC)

Strange statements[edit]

  • making ISO 639-3 code 'ara' an instance of ISO 639-2 code [1]
  • turning specific subclass of UID into less specific subclass of machine-readable data [2]
  • making language code a subclass of machine-readable data - (can humans not read language codes?) [3]
  • [4]

There are no items about instances of UIDs. The instances are printed on books, written on paper, etc.

  • I'm not going to argue that d1g is doing a good job here, I've disputed some of his edits before, he made a big mess among the isotopes that is still not cleaned up. ArthurPSmith (talk) 15:19, 14 August 2017 (UTC)

Erroneous coordinate locations added by APSbot[edit]

I was a nice sunny day of walk in this very quiet area of Tokyo, but I could not get a picture of these items, as they were not there.

Maybe the data source used by your bot is wrong, or some erroneous geocoding took place? In any case, would you mind removing the erroneous data? (for the items above and for the others that might have the same problem).

Thanks a lot! :-) Syced (talk) 08:16, 17 August 2017 (UTC)

@Syced: Hi - these and the other coordinates were stated with a reference, to the GRID database release. If you follow the GRID ID link you'll see where the GRID app maps these to. The problem is probably an uncertainty issue, which they are not providing. But it is stated information in that source directly from the json dump of their data so according to wikidata policy I don't think it's wrong to leave it there. However you might want to deprecate these entries based on your personal experience. ArthurPSmith (talk) 12:55, 17 August 2017 (UTC)

Property creation[edit]

Hi Arthur,

I think it's good you closed the other proposal. At least, this way we can move ahead. I hope you don't mind too much that I listed it for deletion.
--- Jura 06:37, 25 August 2017 (UTC)

I don't mind at all, thanks. ArthurPSmith (talk) 10:48, 28 August 2017 (UTC)

Proposal waiting[edit]

Hello Arthur,

please have a look at this proposal. It's the same as this but affects an other German state - and is waiting since one whole week. Plese tell me, if something is wrong whis my proposal. Thanks --Quarz (talk) 11:20, 13 September 2017 (UTC)

well I don't think I've ever been to Bremen, but I've been to Bayern many times! :) I did leave a comment about the id format - from the page I went to for your example, the full ID did seem to have the leading 4 0's as part of the ID. Has that not been used elsewhere? ArthurPSmith (talk) 13:41, 13 September 2017 (UTC)
OK, Bremen is not as big as Bayern. But people who came here say, that its nice. :) Thank you for your contructive comment. I have answered it. Bremen and Bremerhaven will never have 10.000 or more cultural heritage monuments. So we decided to shorten the id to the significant digits. --Quarz (talk) 14:41, 13 September 2017 (UTC)

OpenRefine demo: incoming new features[edit]

Hi Arthur,

As the OpenRefine demo is coming up next month I thought you'd be interested to know that I have been working on a few features that are relevant:

  • importing wiki tables: basically, OpenRefine now understands the wikicode of tables and can pre-reconcile the cells of such tables based on the wiki-links in the cells. The idea is that this can be useful when migrating manually-maintained tables to Listeria-based tables. It's still not as flexible as I would like (Wikitables can be very messy!) but the basic functionality is ready.
  • editing Wikidata from OpenRefine: it lets users import datasets in Wikidata with the same UI that they use to add statements manually. They can just drag and drop table columns as values in the statements they add, and these columns will be replaced by their values for each reconciled item. For now this feature relies on QuickStatements but the plan is to bypass that and make edits directly from OpenRefine.

You can try these features on the wikidata-extension branch of the OpenRefine repository. For the second one, you should get a "Wikidata" menu in the top right corner, with a "Edit Wikibase schema" action. Then, you can add statements and save the schema (hopefully it's intuitive enough). Once that is done, you can export to QuickStatements.

I'm not sure if these features will be properly released before WikidataCon, but I guess that gives some idea of what sort of things the tool is heading to.

Also, I have started Wikidata:Tools/OpenRefine to advertise a bit the tool to the community - feel free to improve if you have time.

Cheers − Pintoch (talk) 19:46, 13 September 2017 (UTC)

Hi @Pintoch: - I was just starting to try the wikidata-extension branch and hit a classLoader error:
SLF4J: Class path contains multiple SLF4J bindings.
SLF4J: Found binding in [jar:file:/Users/apsmith/src/OpenRefine/server/lib/slf4j-log4j12-1.5.6.jar!/org/slf4j/impl/StaticLoggerBinder.class]
SLF4J: Found binding in [jar:file:/Users/apsmith/src/OpenRefine/main/webapp/WEB-INF/lib/slf4j-log4j12-1.5.6.jar!/org/slf4j/impl/StaticLoggerBinder.class]
SLF4J: See http://www.slf4j.org/codes.html#multiple_bindings for an explanation.
SLF4J: Actual binding is of type [org.slf4j.impl.Log4jLoggerFactory]
SLF4J: The requested version 1.5.6 by your slf4j binding is not compatible with [1.6, 1.7]
SLF4J: See http://www.slf4j.org/codes.html#version_mismatch for further details.
Exception in thread "main" java.lang.LinkageError: loader constraint violation: when resolving method "org.slf4j.impl.StaticLoggerBinder.getLoggerFactory()Lorg/slf4j/ILoggerFactory;" the class loader (instance of edu/mit/simile/butterfly/ButterflyClassLoader) of the current class, org/slf4j/LoggerFactory, and the class loader (instance of org/mortbay/jetty/webapp/WebAppClassLoader) for resolved class, org/slf4j/impl/StaticLoggerBinder, have different Class objects for the type LoggerFactory; used in the signature
Any ideas how to fix this? How much of this is in the main branch now (which does work for me)? ArthurPSmith (talk) 13:59, 12 October 2017 (UTC)
Hmm, and something's broken with the "cell.cross" GREL action - I just get null pointer exceptions for everything now, even doing exactly the same things that used to work. ???? ArthurPSmith (talk) 15:19, 12 October 2017 (UTC)
Ok, "cross" works again, I just had to restart "./refine". Not sure why it didn't work right away, I think that was not a problem last time around. However, in investigating this I did discover the eclipse configuration seems to be broken (some jar files listed in .classpath don't exist for one thing) and running "./refine test" resulted in 1 test broken (related to URL caching). So there seem to be some things needing fixing at least... ArthurPSmith (talk) 16:11, 12 October 2017 (UTC)
Hi Arthur, sorry yes this branch is a big mess and I will not have much time to clean that up before WikidataCon. I have bitten a bit more than I can chew! I will definitely finish this work at some point. For now, everything on master should be clean (all tests should work, at least they all pass on Travis). That already includes Wikitable import. − Pintoch (talk) 19:53, 12 October 2017 (UTC)
Ok - actually the last couple of comments from me were regarding master, as I couldn't get wikidata-extension branch to run after the build so I switched back to master. Glad to hear the wikitable import bit is in, I may try that for the demo! Do you use eclipse or something else to work on this? ArthurPSmith (talk) 19:58, 12 October 2017 (UTC)
Yes I use Eclipse and yes I had issues with the config files too… I suspect the issue you had with the URL caching test could be due to a temporary network issue on your side. − Pintoch (talk) 08:59, 13 October 2017 (UTC)

Academic institutions from India[edit]

Hi!

I remember that in Wikidata:WikiProject_Universities/Scope you pointed out that we were missing many institutions from India. I have created Wikidata:WikiProject_Universities/New items which tracks the latest institutions created and a lot of them are from India! It is nice to see that there is activity where we need. Also, I am thinking about migrating Wikidata:WikiProject_Universities/Scope to Listeria-based tables, so that they get updated automatically.

Cheers − Pintoch (talk) 09:32, 27 September 2017 (UTC)

nice! I notice many of them don't have sitelinks, I wonder where they're coming from? ArthurPSmith (talk) 11:59, 27 September 2017 (UTC)
by the way I updated the first few (13) countries on the list with the latest numbers from the query; India has definitely greatly increased its count since you ran that before, although it's clearly still missing a lot. China should have at least 2000, so it's missing a lot too. And I don't really believe Poland has more academic institutions than Germany, France, or Russia, so I expect there are a lot missing from some of those countries still too. ArthurPSmith (talk) 12:47, 27 September 2017 (UTC)

GRID references[edit]

Hi Arthur. Your bot created GRID Release 2017-05-22 (Q30141628), which is used in a number of articles. In an enwp discussion it was pointed out that the references generated from this were very confusing - in particular, the URL doesn't go to a GRID URL, but to figshare, which isn't very user-friendly. Would it be possible to use a URL to a grid webpage instead for this, or otherwise improve the information in entries like this so that it can be used to generate more useful references? Thanks. Mike Peel (talk) 22:08, 29 September 2017 (UTC)

Hi Mike - I assumed including the direct GRID link in the reference was redundant since it's there in the external ID section, but I could add that, sure. ArthurPSmith (talk) 02:34, 30 September 2017 (UTC)
That would be useful, please. The DOI in the infobox also seems to point to Figshare rather than GRID, though. Thanks. Mike Peel (talk) 13:53, 30 September 2017 (UTC)
I agree that it would be good practice to put as reference the identifier the statement was drawn from. That is useful even if the identifier itself is input as a statement, for a lot of reasons. If identifier X happens to be wrong, it is easy to remove everything that was derived from X - and recursively: if another identifier Y was derived from X, we should remove all statements derived from Y, and so on… My identifier edits (adding an ISNI from a GRID, or adding a Ringgold from an ISNI) have already been reverted multiple times because the source identifier was wrong, so it would definitely be useful in these cases. Unless it has already been done, I think it would make sense to formalize that as an "essay" or something like that. − Pintoch (talk) 12:06, 1 October 2017 (UTC)
@Mike Peel: the DOI points to the permanent data repository for that edition of GRID - it happens to be on figshare, that's where you would download it. It is I think exactly what you want in a source reference: a persistent identifier pointing to the source. That source (where the GRID dataset is downloaded from) is exactly where the bot that entered these values got the information from so it really is the source of the information. However, I can appreciate people like to have more direct dereferenceable links. But I would add that only as a convenience, the source entry right now is accurate and complete. ArthurPSmith (talk) 14:35, 1 October 2017 (UTC)
@Pintoch: are you suggesting to use GRID ID (P2427) rather than reference URL (P854) to provide the link? That's reasonable I think, and it does actually add slightly to the specificity of the source reference, so it makes some sense also for future tracking as you suggest. ArthurPSmith (talk) 14:36, 1 October 2017 (UTC)
Yes, I would use GRID ID (P2427) directly as a reference (probably along with GRID Release 2017-05-22 (Q30141628)). − Pintoch (talk) 14:57, 1 October 2017 (UTC)
stated in (P248) -> no label (Q27982662), GRID ID (P2427) -> number, edition number (P393) -> GRID Release 2017-05-22 (Q30141628) would be better from my perspective, but that might be getting a bit too complex. Thanks. Mike Peel (talk) 21:19, 1 October 2017 (UTC)
@Pintoch, Mike Peel: I've pulled in about 1400 new institutions from the latest GRID release and used this method in the source fields, does it look ok? See for example Federación Alba Andalucía (Q41568673). I don't currently have a bot script to update old records along these lines but I'll look into that next. ArthurPSmith (talk) 20:03, 4 October 2017 (UTC)
I'm not sure this helps much - try {{wikidata|references|Q41568673|P571}} on enwp... Thanks. Mike Peel (talk) 23:22, 4 October 2017 (UTC)
That's an issue with Wikipedia's rendering of the Wikidata citation, we cannot do much about that in Wikidata. I think the citations are great like that, thanks Arthur! − Pintoch (talk) 08:16, 5 October 2017 (UTC)

Wikidata:Property proposal/VBL Spieler-ID[edit]

Hello there! May I have this property created now? Thierry Caro (talk) 13:29, 6 October 2017 (UTC)

I'm going through the backlog and trying to get the ones that have been hanging around the longest or which otherwise have a lot of support done first - it takes a while to do each one! I'm sure I'll get to it soon, unless somebody else does first... ArthurPSmith (talk) 14:25, 6 October 2017 (UTC)
I can deal with the declarations on the property itself and all the rest. I just need the creation per se. But OK. I guess I'll wait. We have a lot of sports properties rady if you want to take care of them. Thierry Caro (talk) 14:31, 6 October 2017 (UTC)
@Thierry Caro: Ok, see VBL player ID (P4298). ArthurPSmith (talk) 14:50, 6 October 2017 (UTC)
Hello again, and thank you. May I now get Wikidata:Property proposal/SpeedskatingResults.com ID, and possibly the few other sports properties that are ready for creation? Again, I'll take them even with absolutely no development on your behalf. Thierry Caro (talk) 13:37, 9 October 2017 (UTC)
These others that are ready are Wikidata:Property proposal/SwimSwam ID, Wikidata:Property proposal/Scottish Sports Hall of Fame ID, Wikidata:Property proposal/Eliteprospects.com staff ID. Would you mind providing me with these too? I'm ready to deal with them. Thierry Caro (talk) 16:36, 9 October 2017 (UTC)
I don't want to leave new properties hanging with no attributes though. I could work on this tomorrow (Tuesday October 10) - what is a good time (UTC) that you could do the fixing up if I create them? ArthurPSmith (talk) 20:45, 9 October 2017 (UTC)
Tomorrow is complicated. If you have time right now, I'm OK. Otherwise later in the week. Thank you whatever. Thierry Caro (talk) 20:51, 9 October 2017 (UTC)
So, OK. There are only two left and I'll be online today, Western European time, to deal with them. Thierry Caro (talk) 08:54, 11 October 2017 (UTC)
@Thierry Caro: ok, see properties 4318 and 4319. ArthurPSmith (talk) 12:41, 11 October 2017 (UTC)
OK. ✓ Done Thank you. Thierry Caro (talk) 14:31, 11 October 2017 (UTC)

Hello again! May I get Wikidata:Property proposal/DSMHOF ID as an empty property? I'm ready to deal with it. Thierry Caro (talk) 12:19, 13 October 2017 (UTC)

@Thierry Caro: Ok - DSMHOF athlete ID (P4363) ArthurPSmith (talk) 12:23, 13 October 2017 (UTC)
Thank you. I'm actually going to need the help of your URL formatting tool for this property. The # part of IDs messes things up. The URLs Wikidata generates turn this into %23. Thierry Caro (talk) 13:45, 13 October 2017 (UTC)
it should work as is just subbing in this property number and the proper URL - I've updated the formatter URL already to do this, can you check it? ArthurPSmith (talk) 13:52, 13 October 2017 (UTC)
Yes. It's now working. Many thanks. Thierry Caro (talk) 13:56, 13 October 2017 (UTC)
May I possibly also get Wikidata:Property proposal/Georgia Sports Hall of Fame ID now? This one should work perfectly straight away. Thierry Caro (talk) 15:28, 13 October 2017 (UTC)
looks like you spotted it already - done! ArthurPSmith (talk) 15:41, 13 October 2017 (UTC)

Hello again! Wikidata:Property proposal/Soccerdonna ID and Wikidata:Property proposal/LFB ID are ready and I'm ready to deal with them. Would you mind having a look at them? Thierry Caro (talk) 17:32, 16 October 2017 (UTC)

Ok! ArthurPSmith (talk) 17:45, 16 October 2017 (UTC)
Thank you. May I get Wikidata:Property proposal/Women's Collegiate Tennis Hall of Fame ID too? I can deal with it. Thierry Caro (talk) 19:51, 19 October 2017 (UTC)
Ok, see P4402 - however, I'm probably done for today. ArthurPSmith (talk) 20:04, 19 October 2017 (UTC)
Thank you. Today I can deal with Wikidata:Property proposal/NLBPA ID. Would you be OK to initiate this other property? Thierry Caro (talk) 19:05, 20 October 2017 (UTC)
OK. This has been done by someone else. Whatever, and of course if you are still OK, I'll need you a few more times before we're done with the current batch of new properties to create. I'll let you know soon again. Thierry Caro (talk) 02:15, 21 October 2017 (UTC)

Hello again. We're almost done. Next one to be created could be Wikidata:Property proposal/Ontario Sports Hall of Fame ID. May you have a look there? Thierry Caro (talk) 18:04, 22 October 2017 (UTC)

I'll also take care of Wikidata:Property proposal/New Zealand Sport Hall of Fame, of course if you can create this too. Thierry Caro (talk) 20:17, 23 October 2017 (UTC)

Quora username[edit]

Please would you review the consensus at Wikidata:Property proposal/Quora username and act according to it, as soon as convenient? I wish to include the outcome in my WikidataCon presentation. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:10, 21 October 2017 (UTC)

See Quora username (P4411). I need to work on my wikidatacon stuff so not sure I'll have time for any more this week! ArthurPSmith (talk) 20:12, 23 October 2017 (UTC)
That's great; thank you. See you in Berlin! Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 14:20, 24 October 2017 (UTC)

A new ID for photographers[edit]

Hi, I saw you created an ID specific for photographer and you are interested in science. Could you please leave a comment on en:w:Wikipedia_talk:WikiProject_Photography#A_new_ID_for_wikidata, if it interestes you. Thank you in advance.--Alexmar983 (talk) 05:50, 22 October 2017 (UTC)

Chart of nuclides[edit]

Hi Arthur, I have a new idea to improve the quality and the usability of the chart of nuclides. I think it would be really interesting to open a popup message when the user over a specific box. This popup would display the information related to the nuclides (half-life, mass, spin-parity, ...). BTW, thanks for having fix all the mess with the nuclide items. Pamputt (talk) 10:46, 31 October 2017 (UTC)

Yeah I've thought about doing something like that. I don't think it would be too hard to do... Are you offering to look into this? :) ArthurPSmith (talk) 11:15, 31 October 2017 (UTC)
Great! Actually I am not sure I am able to find time to help you on this task. I am already involved in other stuff I would like to finish first. That's said, I keep in mind this task and if you do not implement this soon, I could come back to you in few months. Note that we still need to add several nuclear data, such as atomic mass, binding energy (P2154),mass excess (P2160), ... All this data are available here. Pamputt (talk) 12:29, 31 October 2017 (UTC)

Relations[edit]

About last edits, from my set theory course book: "Any set R is a binary relation if and only if its members are ordered couples. It's called domain the set of the first elements in said couples, range or codomain the set of the second elements in the couples" (The book also reports the proofs of the existence of said sets).

So, any binary relation (Q130901) R is part of (P361) of the Cartesian product (Q173740) dom R x rng R . Conversely, a Cartesian product (Q173740) is a (subclass of (P279)) binary relation (Q130901) because its elements are ordered couples. I can't miss to notice the intricacies and apparent loops due to lack of ways to refer directly to objects in specific statements or nesting triples.

For example, n-tuple (Q600590) are a subclass of ordered pair (Q191290), since the same book above define them by induction on ordered couple. How could we model that? --Ogoorcs (talk) 14:58, 9 November 2017 (UTC)

@Ogoorcs: It's important to remember the meanings of the subclass and instance properties. "relation" is a class, whose instances are specific relations. "Cartesian product" is a class, whose instances are specific cartesian products. So while a specific relation can be described as a subset of a specific Cartesian product, that means in general a relation is not a Cartesian product, but a subset of one. If you were looking at a specific relation (for example a simple function like square (Q111124) on a specific domain, say set of real numbers (Q26851380)) then you could add a statement that in a sense this function is a subclass of the cartesian product of its domain and range. However, that doesn't seem a very useful statement to make in the specific case. And when you go up to the class level that subclass relation is not true, the relation is not a cartesian product in general, but a subset of it. As to tuple and ordered pair - you can certainly define a tuple that way, but it is a warping of definitions to make that a subclass relationship: an ordered pair is an ordered set with exactly 2 elements, while a tuple can have any finite number of elements, so the subclass relation goes the other way. ArthurPSmith (talk) 15:14, 9 November 2017 (UTC)
@ArthurPSmith: I agree that relations are not cartesian products, instead subsets of cartesian products, so binary relation (Q130901) is part of (P361) Cartesian product (Q173740) (this is what Elementi di teoria degli insiemi (Q42886736) says!). Every relation is part of a cartesian product in set theory. EDIT: of course it does not mean that every relation is is part of every cartesian product, but every relation is part of a certain cartesian products.
What do you mean as warping of definitions? In set theory, tuple ARE ordered couples defined recursively (see on enwiki); for coherence's stake that is the correct definition. Could you help me model correctly this information in tuple page? --Ogoorcs (talk) 15:57, 9 November 2017 (UTC)
The enwiki page provides several different definitions, only one of which is the ordered pair recursion one. In the computer-science sense (which is part of the meaning of tuple in wikipedia and so in wikidata also) a tuple has to be different from an ordered pair, because while one of the elements of a pair may be a list (and so allow recursion) that is a concretely different object. (1,2,3) is NOT equal to (1,(2,3)). There is an extra step involved in going from (1,(2,3)) to (1,2,3), some form of list concatenation. This is obscured in the math definition you highlight, but it is a real distinction. There is no need to model any of this confusion on the tuple page. ArthurPSmith (talk) 16:04, 9 November 2017 (UTC)
There are three definitions: one as functions (which mathematically are relations), one as ordered pairs recursion and one as sets recursion, which are equivalent and both from kuratowski, actually. I didn't understood what you said about lists and the need for not considering tuple as recursive ordered pairs. So, to you mathematics is not to be modeled in Wikidata? --Ogoorcs (talk) 16:16, 9 November 2017 (UTC)
Wikidata is about a lot more than mathematics. In any case, somebody already added defining formula (P2534) in this case which provides the definition you like - perhaps you can add a reference to that, but I don't see that any more is needed. "Ordered pair" is clearly a subclass of "tuple", so that relationship between the two is fine and already exists in wikidata. ArthurPSmith (talk) 16:22, 9 November 2017 (UTC)
It is not fine, because in set theory tuple are ordered couple and not viceversa, going the other way will soon get you in loops and omitting inclusions is not an optimal way to avoid them. Example:
you want to say that tuples are sequences? Well, ordered couples are tuple, which are sequences, which are functions, which are binary relations, which are sets of ordered couples. Loop.
If wikidata has to describe mathematics it has to use its concepts. Mathematical theories use sets and classes.
Otherwise we need other items, for tuple in elementary set theory and for tuple far each mathematics foundational theory.
This brings another point, which is that probably Wikidata doesn't have enough expressiveness, as it is now, to express many meaningful things in mathematics. -- Ogoorcs (talk) 18:20, 9 November 2017 (UTC)
Subclass and instance relations are not a proxy for the logical development of mathematics by definitions etc. In fact rather the reverse - the more abstract or general concepts (defined via definitions) tend to be the classes or superclasses of more concrete things which would be the general logical origin of any definitions. Also note what you have pointed out here is not an ontological loop because a "set of ordered couples" is not the same thing as one "ordered couple". However, are you asserting that an ordered pair is not an example of a tuple? An ordered set with two elements is surely both a tuple and an ordered pair, right? ArthurPSmith (talk) 18:47, 9 November 2017 (UTC)
Sorry, could you rephrase the first two sentences?
"Any ordered couple being a set of ordered couple" is indeed a loop, just iterate. After a few times you get that any ordered couple is a set of a set of a set of a set of ordered couples, which is false and in general it's recursive because you're describing an element using itself, which is (in some case unfortunately) now forbidden.
Anyway, I think I've found what's the problem here. With Kuratowski definition, (t+1)-uples are a subclass of tuples; in fact triples are a proper subclass of ordered couples because there are no restrictions on what can be in a couple, while triples must have a couple as first element.
So from a mathematics foundation theory point of view, since (t+1)-uples are nested in t-uple up to 2-uples, a generic t-uple element (with t not specified) and ordered couples just form the same class and are indeed the same thing.
Aaaand this is why we need at least two distinct items for ordered couples :-D --Ogoorcs (talk) 02:45, 10 November 2017 (UTC)
@Ogoorcs: Ok, a number of issues here -
  • Rephrasing: Wikidata is a structured database of information about the world. It is not a mathematical definition/proof system. In particular, the subclass and instance of (P279 and P31) properties have specific ontological meanings within wikidata, with logical entailment (which could for example be implemented via RDF + OWL) including a partial ordering on the subclass relation. That very clearly requires that if A P279 B, then A is "smaller than or the same as" B, and if A P279 B and B P279 A then A and B are the same thing. What you have done by stating "ordered pair" P279 "tuple" and "tuple" P279 "ordered pair" is to logically entail that "ordered pair" and "tuple" are identical. This is wrong. It has absolutely no bearing on how the various different entities are defined in mathematics. Mathematics proceeds from axioms (the set theory axioms being a common starting point) and defines and builds structures on top of those starting elements. Often these definitions are done by some sort of extension process (like recursion) that creates something more general out of something very specific - for example the construction of the real numbers. Real numbers are defined in terms of the natural numbers (in their constructive definitions, either via the rational numbers or decimal representations), but it is completely false to say that real numbers are therefore a subclass of natural numbers. Rather the other way around - the mathematical definition/generalization process is here creating the class or superclass from a more specific subclass or instance. This is a very general observation: the things closest to real world objects which would form the basis of defining anything else are generally at the lowest, not the highest levels of the wikidata class hierarchy.
  • Either "Any ordered couple being a set of ordered couple" makes sense, or it doesn't, there's no additional sense or nonsense added by iterating. And the "sense" of the statement comes from the path taken to get to it: let's take (9, 17) as our ordered pair. Obviously it's also a tuple, and a sequence of length 2. Sequences are functions on the natural numbers, so the function here is the one that takes the value 1 to the value 9, 2 to the value 17, and is undefined for all other values. As a binary relation that can be expressed as {(1,9), (2,17)}. Which is clearly a set of ordered pairs, as demanded. Now in what sense is (9, 17) the same as {(1,9), (2,17)}? It is the same in the sense that the latter is expressing a function that is the defining function for a sequence. That process of slightly modified meaning up the class chain is a perfectly normal process of more abstract representation. There is no ambiguity or loss in this, it works and makes sense.
  • You mention Kuratowski (and have added that named person as a reference in statements in the item for tuple) but not an actual source - please provide some sort of work, not just a name. "Stated in" a "person" is invalid as a reference (and you should note the constraint violation there).
  • You appear to indeed be asserting that "ordered pair" and "tuple" are the same thing. Therefore you should not express this via P279, but should use said to be the same as (P460) with the references you have provided for this.
  • You once again removed sequence (Q133250) as superclass for tuple, rather than adding an additional P279 statement. If you insist on references I can certainly add some for this, even though it seems obvious. More concerning is that this is your third revert of me on this in the last day, which is clear edit warring.
ArthurPSmith (talk) 14:25, 10 November 2017 (UTC)
  • Oh - one more thing: () is a tuple. Also (X). How are either of these an ordered pair? ArthurPSmith (talk) 14:34, 10 November 2017 (UTC)

Truncating aliases for elements?[edit]

Hi, I'm not trying to do any automated or semiautomated edits; some descriptions of elements in Polish had atomic numbers in it, some had periodic groups, most of elements had only 'chemical element' and per WD helpage, description field is meant only for disambiguation purpose. In this case 'chemical element' is sufficient and (now) identical for all elements (in Polish). Any other information, like atomic number or periodic group, can be (and is) properly indicated using properties. Also, I don't know how it would be possible in the future to use WD 'aliases' field in pl.wikipedia (in chemical element infobox or other templates), if there are entries in that field which are not in fact 'other names'. Chemical element's symbol is indicated in property and that's IMHO sufficient. Regards, Wostr (talk) 21:52, 11 November 2017 (UTC)

On the descriptions I guess that's ok to try to be consistent, and it is good to keep descriptions short. Aliases are however useful for search purposes (both in the wikidata UI and the API) - searching doesn't look at property values, so by removing those from aliases you may make it hard for Polish wikidata users to find the elements. ArthurPSmith (talk) 22:33, 11 November 2017 (UTC)

Survey of Scottish Witchcraft[edit]

Hi,

Could I trouble you to review Wikidata:Property proposal/Survey of Scottish Witchcraft - Person ID, please? A prompt decision was requested by the proponent, who has people waiting to carry out the data upload. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 17:04, 14 November 2017 (UTC)

Survey of Scottish Witchcraft - Person ID (P4524) is done! ArthurPSmith (talk) 18:36, 14 November 2017 (UTC)
Thank you. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 19:04, 14 November 2017 (UTC)

The terms[edit]

Hello.Please help here.Thank you David (talk) 17:28, 17 November 2017 (UTC)

Hi newbie![edit]

Stop vandalising! Not every musician is a performer! Idiot, you! 213.39.186.16 15:31, 7 December 2017 (UTC)

Thanks for changing the claims. Please take care with whom you call vandal (see project chat). It may heavily backfire. Best regards, you are able to learn, very good. Hope we can fix the classification (also of the identifiers messed up by JakobVoss). 77.180.22.68 19:18, 7 December 2017 (UTC)

Talk page vandalism[edit]

Your talk page had been vandalised: https://www.wikidata.org/w/index.php?title=User_talk:ArthurPSmith&diff=605444912&oldid=605443951 77.180.22.68 19:14, 7 December 2017 (UTC)

I hear your name is Tobias. What did you do to get yourself banned? Stuff like this? Why do you keep trying to edit here then? ArthurPSmith (talk) 19:17, 7 December 2017 (UTC)
My name isn't Tobias, you heard the wrong voice talking to you. I am not trying, I am doing. You faker. 77.180.22.68 19:21, 7 December 2017 (UTC)

Request on administrators noticeboard[edit]

Hi, there.

Let's answer about the fact that that in some cases I added informations without providing source. At my beginnings it was true; I accept I have done errors; but I have paid for this (I have been blocked for a week), and I have taken a responsible, careful and rigorous attitude, so I'm quits with it. Please keep in mind that the request is not about me, and that DDR3 is obviously attempting of create a diversion by (one more time) slandering me and spreading kilometers of bytes; and please consider that I endure this tracking from DDR3 since months. Also, I refuse categorically to get in contact with DDR3, who has a so despicable behavior regarding myself and uses so dirty methods. Do not hesistate if you have more questions. Nomen ad hoc (talk) 21:21, 8 December 2017 (UTC).

Property proposal labels[edit]

I can edit labels and descriptions in all the languages I've got on my Babel box, but I can't display any labels in languages other than English on a property proposal page. Where do you go in the preferences to enable that? Deryck Chan (talk) 21:13, 12 December 2017 (UTC)

@Deryck Chan: I believe it's set in Preferences -> User Profile -> Internationalisation, there's a Language drop-down there. At least I have mine set to "English" and everything appears in English everywhere for me, so I'm assuming that's the reason. Might want to check with a non-English user though on this if it doesn't work for you! ArthurPSmith (talk) 15:47, 13 December 2017 (UTC)

On how to handle branches[edit]

Hello! I've been meaning to come around here for a while, but I always end up distracted by something else, and then it's too late. You undid one of my edits a while ago (I really don't even remember what it was), and it called my attention for a recurrent problem I have found: that of branches, basically in languages and disciplines. So, we have English (Q1860) as a subclass of (P279) Anglic languages (Q1346342), but also as part of (P361) Anglic languages (Q1346342). As for disciplines, there is the additional issue of having them split as instance of (P31) academic discipline (Q11862829), specialty (Q1047113) or branch of science (Q2465832), which is a whole different mess, but makes it harder to keep track of them. Anyway, I have noticed a predominant use of P279 over P361, but it seems to depend mostly on personal preferences right now. What are your thoughts about this? Andreasm háblame / just talk to me 03:38, 14 December 2017 (UTC)

@Andreasmperu: I've reverted a lot of people in the last couple of weeks to try to fix problems on this list! I see you've been helping with some of those problem edits too, so thanks! On branches and P279 vs P361 - I guess I'm a little fuzzy, but my guiding principle is "can you think of this entity as a set or collection of things" (i.e. as a class) and then if the set of things that make up one of them is a subset of the set of things that make up the "bigger" one, then P279 makes sense. For a language you could think of the speakers of the language as the defining set (but we wouldn't say a speaker was an "instance of" the language) - maybe this also applies with "works" or other things that are "in the language"? Anyway, if the set of speakers (or works etc) for language A was included in the set for language B, then I think A subclass B would be appropriate. For example Swiss German (Q387066) subclass of (P279) Alemannic German (Q131339) seems fine to me. But for the subclass relationship to work both A and B need to be kind of the same thing - a "language". The problem with Anglic languages (Q1346342) is it is not a language, but a "language group". So subclass in that case seems wrong to me. But these are rather subtle questions and maybe best handled by somebody with actual expertise in the field - linguistics I guess here. ArthurPSmith (talk) 13:49, 14 December 2017 (UTC)

P159 and GRID imports[edit]

Hi, any progress on setting P159 during your GRID imports? I think it is technically possible to extract this data from GRID items and it is a pity not to do it. I think that even simple OpenRefine reconciliation may produce data for P159 with 90% accuracy. I think is easier to fix errors (it is possible to track errors using queries to compare P625 from GRID with P625 of P159 item), than doing two separate imports. If there is a technical problem, we can simply ask community for help.--Jklamo (talk) 10:21, 15 December 2017 (UTC)

@Jklamo: User:Pintoch has been working on this - see Wikidata talk:WikiProject Universities/External databases. ArthurPSmith (talk) 14:17, 15 December 2017 (UTC)
@Jklamo: OpenRefine reconciliation via GeoNames id works very well in my experience (without the GeoNames ids, there are quite a lot of ambiguous cases actually, even when refining by country…) I used to do these edits from this account but I realized that it was not appreciated to flood the recent changes and watchlists. So I have applied for a bot flag on User:PintochBot and will do the edits as soon as I get it. − Pintoch (talk) 15:00, 15 December 2017 (UTC)
@Pintoch: Great, nice to hear that. If there is anything I can help with, just let me know.--Jklamo (talk) 20:46, 16 December 2017 (UTC)
@Jklamo: the import is running. Not all the locations from GRID are imported - there are still a few thousand locations that I have not reconciled to Wikidata. Unfortunately reconciliation via Geonames is spoiled by cebwiki items so I am not entirely happy with this batch, but most of the edits look good as far as I can tell. − Pintoch (talk) 09:19, 15 January 2018 (UTC)
@Pintoch: Great, already noticed! Most of the edits looks very well, I also appreciate the parent/subsidiary imports. Just one thing, for organizations (universities, ministries, etc.) P159 is more appropriate than P131.--Jklamo (talk) 14:30, 16 January 2018 (UTC)
@Jklamo: Yes, unfortunately the types in GRID are quite uninformative so it is hard to get this right. And it is also subject to interpretation to some extent… (What about hospitals? Research labs/groups/institutes?) I just used Arthur's heuristics from Wikidata talk:WikiProject Universities/External databases but I really do not mind if someone migrates some P131 to P159 (personally I always use both in my queries because I find the distinction unreliable). − Pintoch (talk) 14:50, 16 January 2018 (UTC)
@Pintoch: Any plans to run P159 import again after recent GRID import (noticed 700+ new constraint violations).--Jklamo (talk) 09:53, 21 March 2018 (UTC)
@Jklamo: yes, I have been waiting for UK Provider Reference Number (P4971) to be created (which I have just done), so that I can import that in the same go. It should happen in the next few days. So, if I just add the headquarters location (P159) claims, a bot should migrate the coordinates in these claims, right? Is it better if I add the coordinates as qualifiers directly, or if I leave the bot add them? − Pintoch (talk) 10:02, 21 March 2018 (UTC)
@Pintoch: Yes, bot is migrating coordinates to the P159, but it is better to add the coordinates as qualifiers directly, to avoid constraint violations (and avoid spoiling property usage statistic, which are the basis for property suggesting tools like Recoin).--Jklamo (talk) 10:39, 22 March 2018 (UTC)

Ola[edit]

Olá amigo estou tentando construir uma entidade chamada a Israel Lucas Góis ele é um investidor muito conhecido aqui no Brasil o grande problema é que toda vez que cria uma entidade um administrador chamado Pasleim, excluir a página acabei brigando com ele e ele está me perseguindo não deixando o efetuar nada, Você poderia por favor me ajudar a criar essa entidade GRUPO CALIMA DIESEL (talk) 02:56, 16 December 2017 (UTC)

I was going to recommend you start with a ptwiki page rather than an item in wikidata, but it looks like that's been done - no label (Q45803705). Let me know if there's still a problem here. ArthurPSmith (talk) 13:48, 18 December 2017 (UTC)

Religions, metaclasses and instances[edit]

Hi, you recently reverted one of my edit on Taoism (Q9598) (diff: [6]), but I'm not sure to fully agree. We're trying to sort out religions within the Project Religions and the (current (weak)) consensus is to use subclass of (P279), for the reasons explained on the Ontology. Could you please read the explanations and discuss your point of view, if you feel the ontology is wrong? — nojhan () 10:07, 20 December 2017 (UTC)

Ah, nice! I wasn't aware of that - I've just been trying to clean up cases where we had both A P31 B and A P279 B, and trying to delete the one that seemed to be less common. But I'll fix this and look at some other cases too, I think you have the right approach there. ArthurPSmith (talk) 15:09, 20 December 2017 (UTC)
OK, thanks for the kind revert! — nojhan () 09:14, 21 December 2017 (UTC)

Wikidata External ID redirector for EU VAT number[edit]

Hi Arthur,

Is it possible to add EU VAT number (P3608) on your service?

The URL is like this: http://ec.europa.eu/taxation_customs/vies/vatResponse.html?memberStateCode=$1&number=$2

$1 is for the country code (first two letters) and $2 is for the digits.

For example: BE0466745984 → $1: BE, $2: 0466745984 → http://ec.europa.eu/taxation_customs/vies/vatResponse.html?memberStateCode=BE&number=0466745984

Thank you in advance. Tubezlob (🙋) 18:32, 29 December 2017 (UTC)

@Tubezlob: yes that looks doable - I'll take a look, probably next week (new year). ArthurPSmith (talk) 18:46, 29 December 2017 (UTC)
Yes of course, thank you! Tubezlob (🙋) 09:10, 30 December 2017 (UTC)
@Tubezlob: Ok this is done, and I've updated P3608 with the new URL. You may have to edit or append "?action=purge" to pages to see the new formatter URL in action. ArthurPSmith (talk) 14:37, 9 January 2018 (UTC)
OK thank you very much and happy new year! Tubezlob (🙋) 19:01, 10 January 2018 (UTC)

Textiles and fibers[edit]

I've just started trying to clean up fibers and the textiles made from them. They are very messy right now. I am using Getty AAT as a source for class hierarchy and Fairchild's Dictionary of Textiles (Q28888841) as a general reference. So far I've done "wool" and its subclasses. I'd really like your opinion on classes and instances - I always struggle with when "substances" can be instances.

Many editors seem to feel that materials like named fabrics ("tartan", "naugahyde", etc.) are <instance of> textile (Q28823). Does this seem right to you, or should <instance of> textile (Q28823) be reserved for individual tapestries, fabric samples, and other unique objects?

In the same vein, I am trying to decide whether a fiber can be an <instance of>. So far I have a class hierarchy like this:

  • fiber > natural fiber > animal fiber > wool > merino (based on AAT with a few logical additions).

I also have:

  • fiber > textile fiber > wool > merino

textile fibre (Q3071311) isn't in AAT, but it's widely used in the industry and appears in many Wikipedias. My question is, does it make sense to say wool, mohair, merino, etc. are all <instances of> "textile fiber" rather than subclassses? I can make a case either way.

I'd appreciate your thoughts on this! Thanks and best wishes for the new year. - PKM (talk) 21:40, 31 December 2017 (UTC)

@PKM: Thanks for asking! So as a general rule, P279 (subclass) is the preferred relation between a narrower and broader term when they are both abstract concepts that refer to the same general "type" of thing. If you interpose a P31 (instance) relation in the middle of an abstract sequence of that sort, then the concepts on one side of the P31 are implicitly of a different "type" than the ones on the other side; we sometimes call the broader concepts there "metaclasses", as their instances are themselves classes. It is arguably ok to do this in many cases like this one, but one needs to pick a reasonable boundary between classes and metaclasses that all could agree on. If "fiber" (or "textile fiber") can not reasonably be thought of as a metaclass, that is its instances should be individual fibers rather than classes of fibers, then you should just use subclass across the whole range of relations here. On the other hand if it seems the concept is being used in wikidata consistently in a metaclass fashion then you could make that more explicit by relabeling "fiber" as "type of fiber" or "class of fiber" etc. ArthurPSmith (talk) 15:55, 1 January 2018 (UTC)
Thanks for that excellent explanation. I think I am beginning to understand metaclasses now (the concept always confused me before, since it seems like most higher-order classes in Wikidata could be described as metaclasses). I'm still learning this stuff!
A couple of questions: If I decide that "textile fiber" should be a metaclass "class of fibers used in making yarn, fabric, or other textiles", can a class and its subclass both be instances of the metaclass? That is, can "wool" be <instance of> textile fiber and "merino" be both <subclass of> wool and <instance of> textile fiber, or is that bad ontology? And also, can a metaclass be a subclass of something that isn't a metaclass? - PKM (talk) 19:59, 1 January 2018 (UTC)
It does sometimes tax the brain to sort these things out clearly! In your example, yes I think that's fine for both "wool" and "merino" (subclass of wool) to be an instance of "textile fiber". In general a metaclass should be a subclass of another metaclass, although it could be a subclass of a variable-order metaclass or something more complicated as sometimes those things are needed. ArthurPSmith (talk) 14:56, 2 January 2018 (UTC)
Thanks again! Much to ponder here. - PKM (talk) 20:29, 2 January 2018 (UTC)
Update: after working with fibers and textiles for a few weeks, I decided to make "textile fiber" a metaclass. Thanks for your guidance. - PKM (talk) 20:32, 17 February 2018 (UTC)

ptable[edit]

Hi Arthur, I plan to push some commits for ptable. Before that, I have some question. I clone the repo and try to test the tool in local. However I do not how to it because I opened index.html or nuclides.html and the page that appears contains a lot of un-interpreted code. What is the rpocedure to have a working ptable in local? Thanks in advance. Pamputt (talk) 20:17, 19 January 2018 (UTC)

It's python - as I recall you have to run python app.py - it starts a local flask server with a particular port number you can point to and see. ArthurPSmith (talk) 20:27, 19 January 2018 (UTC)
@Pamputt: to make sure you see this! ArthurPSmith (talk) 20:27, 19 January 2018 (UTC)
It works! Now I just have to work Face-smile.svg. Thanks. Pamputt (talk) 20:29, 19 January 2018 (UTC)

Hi Arthur, I submitted a simple patch but it is not yet applied. Could you explain me what are the steps before the patch is in use? This is a really simple patch and I plan to add other feature (dislay other nuclear data in the tooltip box, (un)zoom using scroll button, ...) but I want to know all the processes before doing a big job. Pamputt (talk) 22:09, 30 January 2018 (UTC)

Maybe Ricordisamoa (talkcontribslogs) could also reply to these questions. Pamputt (talk) 22:10, 30 January 2018 (UTC)
@Pamputt: sorry I've only had a couple of days back at my office where I can work on this (due to travel and weather etc). I'll try to take a look tomorrow! I think @Ricordisamoa: has to do the actual installation and restart etc (at least I haven't done this for this app before). ArthurPSmith (talk) 22:39, 30 January 2018 (UTC)
@Pamputt: Thank you for the patch, I have merged it and look forward to the bigger ones! (You can use Phabricator to draft proposals.) And yes, the development+deployment process should really be documented (part of phab:T99847) --Ricordisamoa 23:11, 2 February 2018 (UTC)
@Ricordisamoa: thanks for the merge. For the coming patch, they are not yet ready. I cannot promise to push them soon because I have other stuff to do before. Anyway, I hope I will be able to push my commits before June. I will ping you at that time. Pamputt (talk) 08:28, 3 February 2018 (UTC)
@Pamputt: Please take your time Smiley green alien droling.svg --Ricordisamoa 10:29, 6 February 2018 (UTC)

Awards and classification, what is an award ?[edit]

Taking about https://www.wikidata.org/w/index.php?title=Q5204895&oldid=prev&diff=628083726 . I try for a while to root our « class or not » principle in a well founded manner and powerful principle. This means, for example, using the type/token distinction to identify « tokens » that are classified into (first order) « classes ». Did you read the wikipedia article on it, or User:TomT0m/Classification ?. This leads to consistency with other ontologies and predictability - several people starting from the same principles are likely to reach the same conclusion without talking to each other. Also it adds consistency in the way we treat awards themselves, so it makes querying easier.

I think the tokens at sake here, are the moment where someone (or a whole group or team) is awarded. This is what we should classify, I think, in the first place - let’s call them « awardment ». That makes any award a class of awardment, as usually there is several awardments of that type. Awardments can be classified further, such as « 2018 Nobel Price », a subclass of « Nobel price », or « Peace Nobel price », also a subclass of Nobel Price. This model works as well for a price delivered once, it’s just a class of award with a single awardment.

If there is needs that are not captured by this model, we can add a classification level and add class of classes of awards, like « recurring award » - an award class which the awardees are chosen recurringly at a timely manner, for example.

I think this model has a lot of qualities. What do you think ? What’s the principle that guides you into deciding what is a class or not ?author  TomT0m / talk page 08:27, 8 February 2018 (UTC)

An "award" or "prize" to me is inherently a recurring thing that can be given to multiple people either simultaneously or over time. There are a very small number like Orteig Prize (Q1930819) that are one-time only, but I don't think that negates the general rule here. The receipt of a particular award by a particular person at a particular point in time is an event - I don't know if there's an English word for it; let's call it "awardment" as you suggest. So such an event would be an instance of "awardment". The actual award or prize that is given is (generally) a recurring thing, and is a member of the class of "awards" just as the awardee is the member of the class of humans. It may be a member of a subclass of "awards" also. The Nobel prizes are given in different categories, so Nobel prize as a whole is a class of awards, but the individual categories (Nobel prize in medicine, for instance) is an *instance* of Nobel prize (which subclass award). These distinctions seem pretty clear to me. ArthurPSmith (talk) 15:49, 8 February 2018 (UTC)
@ArthurPSmith: «  is a member of the class of "awards" just as the awardee is the member of the class of humans. It may be a member of a subclass of "awards" also » This seems to indicate a problem in how you see thing. It’s not possible in general to be both a subclass and an instance of the same class.
« The actual award or prize that is given is (generally) a recurring thing » You have a thing that is given several times ? that would mean the last to get it transfer it to the new one … This means imho that each award is unique. What is recurring is the awardment, as an event who can recur …
Other problem : what would be a superclass of « award » in your model ? Is this an event, as the awardment, the trophy or the money that is given in awardment, an organisation as the Nobel committee ? If it’s an event, then it is easy to assimilate award and awardment, and the reason that there is no word for awardemnt in english is that … there is : it’s award https://www.merriam-webster.com/dictionary/award
In my model, there is no exception at all to the rule, it has none of the problem I exposed above. It’s clear and systematic, and do not need parasit concept. It’s also consistent with common definition in english. author  TomT0m / talk page 16:08, 8 February 2018 (UTC)
@TomT0m: What exactly is your model? P31 is never used at all for awards? You misinterpret what I said above when you claim I was saying an award is "both a subclass and an instance of the same class", that's not what I was saying at all. A specific award is a member (P31) of the class "award", or possibly a member (P31) of a subclass (like "Nobel Prize"). Note that medals such as "Purple Heart" are (in wikidata now) considered an instance of "medal" which is a subclass of "award". This makes perfect sense to me. An award is like a work, it has (generally) many manifestations in the form of individual objects obtained by individual people at particular points in time. We document these individual "awardments" with award received (P166), there's no need generally to recognize them with their own wikidata items. As to superclass, probably artificial entity (Q16686448) would be better than object (Q488383), but both are rather generic/abstract anyway. ArthurPSmith (talk) 16:25, 8 February 2018 (UTC)
(I was writing a lenghty answer but I’ll leave it for later, a remark first) The enwiki says : « An award is something given to a person, a group of people, like a sports team, or an organization in recognition of their excellence in a certain field. ». I think our model should not invent its own definition and follow the basic definitions. It follows from that definition that the « Peace Nobel price 2017 » is an instance of « award » : the « Peace Nobel price 2017 » is something that is given to a person, a group of people, like a sports team, or an organization in recognition of their excellence in a certain field. Any example of Nobel prize, like « Peace Nobel price 2017 », is also an award on that definition. It follows that « Nobel prize » is a subclass of « award ». It’s as simple as this. The same for « medals ». Medals are just specific kind of awards, so
< medal > subclass of (P279) View with SQID < award >
. That’s it. It’s hard to be less precise and conceptually economic and consistent with external definitions, we should not invent our own square wheel when it’s not needed. author  TomT0m / talk page 17:30, 8 February 2018 (UTC)
I don't think I'm inventing my own definition here, it's common sense and to my mind fully agrees with what you quote from enwiki. However, you seem to be making a distinction between "Nobel Peace Prize 2017" (which would be P31 "award" in your model?) and "Nobel Peace Prize" (which would be "P279" "award" in your model?). Do you think we should have a wikidata item for "Nobel Peace Prize 2017" (and every other year)? I don't think that's tenable - especially when you look at other cases like medals (for which a given medal may have hundreds of thousands of "awardment" events). The important thing here, just as with works, is to describe the (yes, abstract) award with its criteria, its selection committee, the monetary or concrete object associated with the award, the frequency of recurrence, etc. Every "award" has such well-defined attributes which make it distinct from every other "award", and that's why it's an instance (P31) of "award", in my view (which I think is just expressing common sense on this). This is the same approach we take with books, music, and other created entities. ArthurPSmith (talk) 20:00, 8 February 2018 (UTC)
This would be entailing we have to sacrifice some kind of rigor to represent stuffs correctly. I don’t think so. We don’t need to sacrifice rigor to implement something easy to use. We just have to make a distinction between awards type (I think I created something like a « recurring award type » item for « Nobel price » and all the alike to be an instance of, so a metaclass modelling approach) and award token, and allow « award recieved » to have as domain either an instance of « recurring award type » or an instance of « award ». Actually the « recurring » pattern is much more than just award, (I did play with this in the past, see WikiProject Recurring events. author  TomT0m / talk page 11:48, 21 March 2018 (UTC)

OpenRefine problem[edit]

Hi, I've been having a go with OpenRefine (Windows, v. 2.8 trunk) for some of the terms in the Getty AAT thesaurus (AAT ID (P1014)).

I'm finding that occasionally when I try to reconcile a column, I get a "Working..." pop-up that never goes away.

Looking at the service log screen, it seems that it's sending a "guess cells type" request for the first 10 cells, and https://tools.wmflabs.org/openrefine-wikidata/api is sending back a "Forbidden 403" error.

(Ironically, I don't even want a cell type -- the items I'm looking for are actually classes, so mostly shouldn't be instance of (P31) anything.)

At the same time, it seems to handle other columns from the same project without any trouble. But sometimes there's a column it just doesn't seem to like, nor even facets from it.

Is this something you've ever encountered; any idea what might be causing it; or any idea of a fix? Jheald (talk) 15:13, 23 February 2018 (UTC)

@Jheald: @Pintoch: is the real expert on it - however I have seen problems similar to what you describe; it may be a memory issue? Also note that the project is under active development; there might have been a change in the version you are using that broke something, so you could try updating to the latest git version or going back a few days to a more stable version. The version I've mostly been using recently (which seems to work fine) is dated February 9th, if that helps. ArthurPSmith (talk) 15:31, 23 February 2018 (UTC)
I was just using the official release 2.8 from 19 November. Is there much new, fixed, or otherwise worth having more recently ?
I'd be surprised if it were a memory issue -- the request seems quite small, and it's the wmflabs service that seems to be rejecting it, rather than OpenRefine itself failing. Plus it's this initial query that's falling over, not a later stage. My assumption was that there might be a value in one of the fields that was breaking it. But I haven't yet found a way to work round it. Jheald (talk) 15:39, 23 February 2018 (UTC)
@Jheald: sorry, yes I am aware of this bug: it is tracked here: https://github.com/wetneb/openrefine-wikidata/issues/19. If you can find any reliable way to reproduce it I would be very interested. I will try to find the time to improve these error messages to make this debugging easier. − Pintoch (talk) 15:45, 23 February 2018 (UTC)
@Pintoch: Most strange. It seems in this case it was specifically a cell containing the string "vector graphics" in the first 10 that caused this. When I blank that single cell, the reconciliation works. When I type back in the string "vector graphics", it fails. Jheald (talk) 16:06, 23 February 2018 (UTC)
Adding "vector graphics" at the top of a column in a different project also made its reconciliation fail, so this may be something that is reproducible. Jheald (talk) 16:33, 23 February 2018 (UTC)
Indeed, a project that only has one cell, which contains the text "vector graphics", will trigger the problem. But "tor" or "tor gra" don't, so I'll be fascinated to know what it is about this string that causes it to be rejected. Jheald (talk) 16:51, 23 February 2018 (UTC)
@Pintoch, Jheald: - looks like the problem is because vector graphics (Q170130) is P31 "unknown value". ??? ArthurPSmith (talk) 16:56, 23 February 2018 (UTC)
Yes indeed! I just fixed that. https://tools.wmflabs.org/openrefine-wikidata/en/api?query=vector%20graphics works fine now. Thank you so much to both of you for your help! − Pintoch (talk) 17:03, 23 February 2018 (UTC)

Wikidata:Property proposal/Aicte Institute ID[edit]

Thankyou for supporting. What next level. What all process/time goes on before creating the Property. Jinoytommanjaly (talk) 07:33, 27 February 2018 (UTC)

@Jinoytommanjaly: properties cannot be approved before 1 week after initially proposed. For ID properties like yours, just one or two supporting comments is generally sufficient, so I expect it will move ahead in the next day or two. ArthurPSmith (talk) 20:04, 28 February 2018 (UTC)

OpenRefine feedback[edit]

Hi!

I have polished up the wikidata extension for OpenRefine that I had asked you to test a while ago - it should compile fine (and hopefully even work) this time. If you have some time to check it out, I would really like to know what you think. The discussion is happening here: https://github.com/OpenRefine/OpenRefine/pull/1530.

All the best − Pintoch (talk) 20:46, 4 March 2018 (UTC)

Reply edit summary[edit]

Sorry, the edit summary on edit 653212109 on WD:Property proposal/hearing date was supposed to say "reply: yes you're right". The apostrophe is so dangerously close to the enter key... esbranson (talk) 01:49, 20 March 2018 (UTC)

Share your experience and feedback as a Wikimedian in this global survey[edit]

WMF Surveys, 18:57, 29 March 2018 (UTC)

edition number (P393) in references[edit]

Hey Arthur, I see that your bot added almost 30k edition number (P393) reference claims to GRID ID (P2427) claims in 2016/2017 (query), but this property is no longer supposed to be used in references. It probably never was, but meanwhile the constraint property constraint (P2302): used for values only constraint (Q21528958) explicitly states that. The values of those reference claims are apparently dates in format YYYY-MM-DD. What does this mean? Would it be possible to keep the value, but move the reference property from edition number (P393) to publication date (P577) or retrieved (P813)? —MisterSynergy (talk) 21:54, 5 April 2018 (UTC)

@MisterSynergy: GRID identifies the version of the database by the release date, in that format. See their download page. However, I've been using a new format for these citations recently, with "stated in" pointing to the specific version; I guess it would be fine for a bot to fix up the old ones but I'm not sure the best way to do this. ArthurPSmith (talk) 22:30, 6 April 2018 (UTC)
Yes, the new format seems suitable. Since the situation with the violations is rather simple, I suggest to move the old references to the new model as well. It is more a matter of time than of complexity, due to the large amount of affected items with the same problem. One would just have to loop over all affected references and replace the edition number (P393) claim with a stated in (P248) claim. Do you have such code, or could develop it? —MisterSynergy (talk) 05:42, 7 April 2018 (UTC)

FYI: I’m now about to do this so that all references follow the new style (stated in: specific GRID edition item). New items for missing releases were already created, and very soon I’ll have the repair bot code as well. Cheers, MisterSynergy (talk) 08:17, 18 May 2018 (UTC)

Oh! Thanks, I've been worrying about when I would get a chance to get on this, so I appreciate you fixing it. Go ahead! ArthurPSmith (talk) 15:55, 18 May 2018 (UTC)
Code is ready, but there are questions in relation to this sample edit:
  • There are at least occasionally references to several GRID releases. Should we keep all of them?
  • There are sometimes plain links to GRID releases with reference URL (P854). Should we keep them as well if there is another reference to another GRID release in the same statement?
Both phenomena can be observed in the sample edit. I’d just do it as you prefer. —MisterSynergy (talk) 21:32, 18 May 2018 (UTC)
@MisterSynergy: The reference URL (P854) entry there was added by me via quickstatements, that was all I knew how to do at the time. The other two were bot actions. I don't think we have edition items for those early GRID releases; it's ok with me to just leave them as a reference URL link. Alternatively it would also be ok with me to just keep the most recent "stated in" reference and remove earlier links to GRID altogether. I don't think it hurts to keep all of them as it is slightly more informative about when the organization was present in the database, but if you think it's better to just have a single reference I have no strong feelings on this. ArthurPSmith (talk) 13:53, 21 May 2018 (UTC)
Okay, I’m now just doing the replacement of references containing edition number (P393), no removals. —MisterSynergy (talk) 14:15, 21 May 2018 (UTC)

This has just finished. I also fixed the bare URL references mentioned earlier to the new “stated in: specific GRID release” format. Still no removals, so some GRID IDs now have references to different GRID releases. —MisterSynergy (talk) 08:15, 22 May 2018 (UTC)

Wikidata:Requests for comment/Privacy and Living People[edit]

You are receiving this message because you commented at the above RFC. There are additional proposals that have been made there that you are welcome to comment on. MediaWiki message delivery (talk) 03:50, 9 April 2018 (UTC) (for Rschen7754)

Reminder: Share your feedback in this Wikimedia survey[edit]

WMF Surveys, 01:40, 13 April 2018 (UTC)

Your feedback matters: Final reminder to take the global Wikimedia survey[edit]

WMF Surveys, 00:50, 20 April 2018 (UTC)

Possible WikiProject for year discussion[edit]

I have inquired at Wikidata talk:WikiProject Calendar Dates#Expand to cover year items? to see if that project would like to be a more enduring place to discuss and document the year-related discussions currently going on at Wikidata:Project chat. Since you have participated, I wanted to invite you to the discussion. Perhaps we shouldn't advertise the WikiProject in Project chat until a consensus emerges about whether the WikiProject wants to take on years or not. Jc3s5h (talk) 21:06, 30 April 2018 (UTC)

Municipality vs administrative territorial entity[edit]

Thanks for pointing this out to me [7]. I was trying to fix the constraint error in London Borough of Newham (Q208139)local dialing code (P473)  "020" and similar constraint errors in other boroughs in the UK. Where should I make the change? Deryck Chan (talk) 14:33, 7 May 2018 (UTC)

@Deryck Chan: sounds like the property constraint on local dialing code (P473) should be changed to allow any "administrative territorial entity", not just a municipality. Assuming a borough of London cannot be classed as a municipality? ArthurPSmith (talk) 14:36, 7 May 2018 (UTC)

Accidentally undoing your edit[edit]

Hi Hi. Sorry for accidentally undoing your edit on Lexeme:L298, I was actually intending on undoing edits on linked Items and Properties. Adam Shorland (WMDE) (talk) 10:09, 26 May 2018 (UTC)

@Adam Shorland (WMDE): Thanks for fixing it. I notice you also reverted several property statements that linked to lexemes (these were generally the property examples which is important for people to know how the property is intended to be used). I understand there are database issues or something along those lines requiring these to be gone for the moment, will you be able to restore those when it's ok to have them again? ArthurPSmith (talk) 18:50, 26 May 2018 (UTC)
We will re asses the situation today and see if we can re add the statements to Properties. As for on Items, https://phabricator.wikimedia.org/T195615 will be the ticket to follow. Adam Shorland (WMDE) (talk) 08:22, 28 May 2018 (UTC)
Due to a US holiday today (Monday) this will be pushed back to Tuesday. Adam Shorland (WMDE) (talk) 18:00, 28 May 2018 (UTC)

Eurovision Song Contest[edit]

It isn't a script, but manual edit. Sorry --ValterVB (talk) 18:28, 28 May 2018 (UTC)

Wikidata:Database reports/Constraint violations/P5094[edit]

Hi Arthur. Is there a tweak that could be made to the formatting regex of HPIP ID (P5094) so that two-digit numbers aren't flagged, please? I suspect that the unique value constraint isn't 100% applicable here, but we can see how that goes. Thanks. Mike Peel (talk) 18:58, 28 May 2018 (UTC)

  • @Mike Peel: I tweaked so this should be ok now - is there a possibility of 1-digit or 5-digit numbers though? ArthurPSmith (talk) 13:25, 29 May 2018 (UTC)
    • It looks like 1-digit IDs exist, see commons:Category:Oceana Building (Beira). I don't think the list is long enough to have 5-digit IDs though. Thanks. Mike Peel (talk) 13:44, 29 May 2018 (UTC)
      • Well, I just generalized it to any valid positive integer; I don't think it's terribly useful to have overly strict constraints like that, in general. ArthurPSmith (talk) 13:55, 29 May 2018 (UTC)
        • That should work. Thanks! Mike Peel (talk) 14:14, 29 May 2018 (UTC)

Sounding like Tobias[edit]

Hi. I'm not amused by what you said about me in project chat. I don't know what is the exact user name or IP adresses of this particular blocked user that you are referring to. I don't know what similar subject this user edited/discussed in a similar manner that I did, or whatever makes you think I'm this user. I also don't feel like I have to prove that I'm not a camel. I believe the norm is to provide the behavioral evidence at appropriate venue (instead of discrediting me at some generic topic), should you carry on insisting this. 90.191.81.65 18:59, 14 June 2018 (UTC)

(A) I don't see that I "discredited" you in any way, and (B) the reason for my saying you "sounded like Tobias" was because you are using editing via an IP address rather than an account, and yet making assertions that sound like you should be an authority on wikidata. I'm sorry, without an account and a history, you simply cannot argue from "authority" here. You may have a good argument, and I think I pointed out to Chris there what the issue was, but you can't force your opinion on others. ArthurPSmith (talk) 21:05, 14 June 2018 (UTC)

Télescope géant[edit]

Hi,

You reverted a modification on Wikidata. My point is that "Extrêmement grand télescope" is as good French as "Telescope extremely large" is good English. Besides, the French equivalent is "Télescope géant" (literally "giant telescope"), as is used by the ESO (http://www.eso.org/public/france/teles-instr/elt/?lang) and the French article.

However, I acknowledge that for some reason (unknown to me), wikidata seemed unable to match "Télescope géant" to its article.

Therefore, I invite you to help me correcting this problem. Otherwise I'll just revert your revert. Padex (talk) 09:13, 27 June 2018 (UTC)

@Padex: I left your label change alone, that was fine. What I reverted was your change to a statement - instance of (P31) which previously pointed to a separate item extremely large telescope (Q2665569). That item is a generic "extremely large" category, as opposed to the specific telescope with that name. It sounds like the French label on extremely large telescope (Q2665569) should be fixed - please go ahead and fix it! But note that is an abstract item, not specifically about the ELT. ArthurPSmith (talk) 14:18, 27 June 2018 (UTC)

Thanks :) Hope I did it right this time. To answer your worries : yes, "télescope géant" seems to be the generic equivalent of "extremely large telescope" (not specifically the ELT). For instance the ESO does not translate ELT as a name. Padex (talk) 16:25, 27 June 2018 (UTC)

Yes, that looks right, thanks for improving Wikidata! ArthurPSmith (talk) 18:09, 27 June 2018 (UTC)

peninsula of location[edit]

Hi ArthurPSmith. I created Wikidata:Property proposal/peninsula of location similar to Property:P5130 (island of location). Can you support it? 91.227.222.7 15:10, 7 July 2018 (UTC)