Wikidata:Requests for comment/Close-out of statements formerly using P794

From Wikidata
Jump to navigation Jump to search
An editor has requested the community to provide input on "Close-out of statements formerly using P794" via the Requests for comment (RFC) process. This is the discussion page regarding the issue.

If you have an opinion regarding this issue, feel free to comment below. Thank you!

The property no label (P794) (English label "as", known as a "generic qualifier"), was recently removed, after a year-long collaborative process of migration to less ambiguous properties, including several created for the purpose. Ultimately, every use was migrated, and the property was duly deleted with a clear consensus. However, there remains some question of whether a couple of use cases are now using the most appropriate properties. A discussion on this was underway here, but stalled due to being archived. I'd like to get those use cases wrapped up in a way that's satisfying to all.

@ArthurPSmith, Deryck_Chan, GZWDer, Jheald, Jura1, Pasleim: I'm pinging the people who've contributed already, but of course everyone is welcome to weigh in (though I'd ask you to skim the arguments given previously). This RfC is for dealing with these specific use cases, not for re-litigating decisively-closed matters or airing grievances. Swpb (talk) 16:05, 8 January 2018 (UTC)

Wondering why don't you also ping me (@Liuxinyu970226:) here, Swpb, as I'm also supported the deletion of that property. --Liuxinyu970226 (talk) 15:28, 13 January 2018 (UTC)
Accidental oversight, I'm sorry. Swpb (talk) 14:00, 16 January 2018 (UTC)

The options here, for qualifiers that take the following items as values, seem to be:

For acting (Q4676846) and interim (Q4895105):[edit]

For de facto (Q712144) and de jure (Q132555):[edit]

  1. sourcing circumstances (P1480) (on the theory that these items are very similar to the official (Q29509043) and non official (Q29509080) that this property already takes as values; if another property is chosen, I suggest these should come along.)
  2. criterion used (P1013)
  3. determination method (P459)

Swpb (talk) 16:05, 8 January 2018 (UTC)

  • Symbol support vote.svg Support 1. here on the argument given. ArthurPSmith (talk) 16:22, 8 January 2018 (UTC)
  • criterion used (P1013) seems to be the best match from my point of view - the curators of sourcing circumstances (P1480) don't want to expand the scope and determination method (P459) is less of a good match because it isn't really a "method". Deryck Chan (talk) 17:25, 11 January 2018 (UTC)
    • To clarify: Symbol support vote.svg Support criterion used (P1013); BA candidate.svg Weak oppose determination method (P459); Symbol neutral vote.svg Neutral sourcing circumstances (P1480). Deryck Chan (talk) 17:57, 9 February 2018 (UTC)
    • @Deryck Chan: I'm dubious that we need to give much deference to the self-appointed "curators" of sourcing circumstances (P1480). Local consensus must yield to global consensus – having a particular interest in a given property does not mean one's voice should automatically carry more weight. Also, how many "curators" are we talking about here? Can we invite them to comment on the rationale I gave here (i.e., that it's not an expansion of scope)? I'm only aware of one, albeit particularly vocal, "curator" who opposed the use, and he's already been pinged here. You put more work into deprecating no label (P794) than probably anyone else; how do you feel about the appropriateness of sourcing circumstances (P1480) for this role? Swpb (talk) 14:02, 12 January 2018 (UTC)
      • @Swpb: Well, I can see a conceptual distinction between "accuracy of the source" and "accuracy of the statement". I'm fine with people wanting sourcing circumstances (P1480) to stay strictly in the "accuracy of the source" scope, e.g. "Bede was born circa 673" means that "There was an exact day on which Bede was born [accuracy of statement is much finer], but human knowledge only knows it is around 673 [accuracy of source is much coarser]". I don't feel strongly about what P1480 should represent; just pick one and we'll juggle the qualifier around to match it. Deryck Chan (talk) 11:30, 23 January 2018 (UTC)
  • Symbol oppose vote.svg Oppose first, per above; Symbol support vote.svg Support third as it has more specific meaning than second in (at least) Japanese afaik, Symbol neutral vote.svg Neutral on second. --Liuxinyu970226 (talk) 15:31, 13 January 2018 (UTC)
    • @Liuxinyu970226: Can you please expand on your oppose of #1 here? First, is "per above" supposed to refer to Deryck Chan's comment directly above this, or your own comment under the section before? Swpb (talk) 14:03, 16 January 2018 (UTC)
  • @Jura1: - You've been the most concerned about porting these statements away from sourcing circumstances (P1480). What's your preferred destination? Deryck Chan (talk) 17:57, 9 February 2018 (UTC)
    • Ditto, I'm also confused with Jura1's so-called concern about P1480, but as Swpb mentioned above, there may be has confusion that not resulted by English name, but e.g. German or French name. --Liuxinyu970226 (talk) 06:34, 8 March 2018 (UTC)

So, translations of P1480 need review[edit]

Let's review the translations of contested P1480, maybe because of some wrong translations, their native users chose this wrong one for their propose:

  • Following English (sourcing circumstances): Belarusian (акалічнасці крыніцы), Taraškievica (акалічнасьці крыніцы), Danish (kildeomstændighed), Spanish (circunstancias de la fuente), Galician (circunstancias da fonte), Macedonian (околности на изворот), European and Brazilian Portuguese (circunstâncias da referência), Russian (обстоятельства источника) and Chinese (來源情況 (hant)/来源情况 (hans))
  • Arabic: ظروف توفير المصادر (Conditions of provision of resources)
  • Bulgarian: според източници (according to sources)
    If P1013 isn't a good idea per Deryck, then how about statement supported by (P3680)? This has also "according to" aliases.
  • circumstances of origin: Bosnian (okolnosti porijekla), Hebrew (נסיבות המקור)
  • Catalan: circumstàncies de la referència (circumstances of reference)
  • Czech: nejasné informace (unclear information)
    Isn't "unknown value" provided by system good enough?
  • Welsh: manylrwydd y gosodiad (the detail of the installation)
    About software?
  • German: Qualifikator für Behauptung (qualifier for statement)
    Probably the reason that Jura1 "hates" this property, the confusion is kindly German confusion.
    changed to "Nachweisumstände" --Pasleim (talk) 09:32, 26 March 2018 (UTC)
  • Greek: ποιότητα δεδομένων (data quality)
    Greek people, consider has quality (P1552) instead please?
    No, that's definitely wrong. 1552 is about qualities in the general philosophical sense, not "quality" as in "goodness/usefulness" like 1480 is meant for. Swpb (talk) 19:56, 9 March 2018 (UTC)
  • Basque: iturriko gorabeherak (source fluctuations)
  • Persian: وضعیت منابع (Resource status)
  • French: qualité de l'information (quality of information)
    Sadly, I agree that P1552 isn't a good idea for this
  • Hungarian: bizonytalanságot jelző minősítő predikátum (uncertainty indicating predicate)
  • Armenian: աղբյուրի կարգավիճակ (source status)
  • Iloko: dagiti sirkumstansia iti panangireperensia (check this out)
  • Italian: condizioni della fonte (conditions of source)
  • accuracy of information: Japanese (情報の精度), Korean (정보의 정확성)
    Seems good as this property has alias "accuracy"
  • Georgian: წყაროს გარემოება (source of the source)
  • source quality: Bokmål Norwegian (kildekvalitet), Nynorsk Norwegian (kjeldekvalitet)
  • Dutch: waardering van bron (valuation of source)
    Edit March 29: Complex problem, see next section.
  • Polish: status w źródle (status in source)
  • Silician: pricisioni dâ fonti (precision of source)
  • Slovene: nejasen podatek (vague information)
  • Serbian: порекло околности (Cyrillic)/poreklo okolnosti (Latin) (origin of circumstances)
  • Swedish: källäge (source mode)
  • Tajik: ҳолати сарчашмаҳо (state of origin)
  • Ukrainian: обставина джерела (circumstance of source)
  • Urdu: ممکنہ طور پر (possibly)

--Liuxinyu970226 (talk) 07:15, 8 March 2018 (UTC)

Split sourcing circumstances (P1480) on Deryck's lines?[edit]

@Liuxinyu970226, Deryck Chan, ArthurPSmith, GZWDer, Jheald, Jura1:
I'm going back to what Deryck called "a conceptual distinction between 'accuracy of the source' and 'accuracy of the statement'" and the fact that 1480 has long accepted both. This is how several currently accepted values seem to break:

Qualification is about...
...the source ...the underlying nature of the parent statement
circa (Q5727902)
presumably (Q18122778)
possibly approximate value (Q21097017)
unspecified calendar (Q18195782)
misprint (Q21096955)
miscalculation (Q21096985)
official (Q29509043)/non official (Q29509080)
(de jure (Q132555)/de facto (Q712144)?)
disputed (Q18912752)
rarely (Q28962310)
often (Q28962312)
hypothesis (Q41719)
uncredited appearance (Q16582801)
proposal (Q3918409)
hypothetically (Q18603603)
expected (Q50376823)

Both the English alias "sourcing circumstances", and the other languages Liuxinyu translated for us above, seem to hew closer to the left column -- so maybe we leave only those sort of values to P1480, and create a new property for everything on the right side? Does that kind of split seem sensible? Swpb (talk) 20:34, 9 March 2018 (UTC)

This split seems reasonable to me. Would criterion used (P1013) work for the right-hand side? It doesn't seem quite right, so perhaps we do need a new property here. I note there are some similarities between this problem and the recent property proposal discussion at Wikidata:Property proposal/Attribution Qualifier (a similar qualifier scope regarding author/creators). ArthurPSmith (talk) 20:43, 9 March 2018 (UTC)
I don't think criterion used (P1013) is right. 1013 provides context, and completes statements that don't necessarily carry meaning without that context. The qualifications we're talking about don't require any context. Swpb (talk) 20:51, 9 March 2018 (UTC)
"Sourcing circumstances" has always been a particularly unintuitive and unhelpful name for the property, so it's no surprise that it's ended up as a slightly diverse collection of things -- the name is so opaque and unreferenced that it could mean almost anything. But worth noting that there are a few more values that aren't in your table above, including misassociation (Q21097088), near (Q21818619), possible (Q30230067), hierarchical link is not direct (Q50095342).
It seems odd to me to suggest attaching presumably (Q18122778) and hypothetically (Q18603603) to different properties -- surely these are both just different shades of probability?
criterion used (P1013) has quite a range of current uses tinyurl.com/y8rhwg4f, but I don't think it's quite right for things that aren't actually criteria (which is most of the entries in the right hand column).
I'm not quite sure what problem this proposal is being canvassed to solve. What would this add, over just leaving things as they are? Jheald (talk) 23:38, 9 March 2018 (UTC)
It deals with many editors' desire to keep the scope of sourcing circumstances (P1480) pure. At the heart of the issue is that one can make a distinction between the circumstances / accuracy of the sourcing, versus the circumstances / accuracy of the underlying truth. So we need to find a property (or make a new one up) to accept qualifiers that deal with the circumstances of the underlying truth. Deryck Chan (talk) 12:56, 10 March 2018 (UTC)
Qualification is about...
...the source of a factual statement sourcing circumstances (P1480) ...a prediction or counterfactual statement (either sourcing circumstances (P1480) or criterion used (P1013)) ...the condition under which the parent statement is true criterion used (P1013) ...the likelihood of the statement being true (new property "probability description")
circa (Q5727902)
presumably (Q18122778)
possibly approximate value (Q21097017)
unspecified calendar (Q18195782)
misprint (Q21096955)
miscalculation (Q21096985)
disputed (Q18912752)
hypothesis (Q41719) (hypothetical theory)
proposal (Q3918409)
hypothetically (Q18603603) (hypothetical result)
expected (Q50376823)
official (Q29509043)/non official (Q29509080)
de jure (Q132555)/de facto (Q712144)
rarely (Q28962310)
often (Q28962312)
Better properties available:
subject has role (P2868) or object has role (P3831) + uncredited appearance (Q16582801)
Deryck Chan (talk) 14:38, 10 March 2018 (UTC)
I think this four-column table is slicing the use cases too thinly and subtly to be used consistently (e.g., I think the distinction being made between hypothesis (Q41719) and hypothetically (Q18603603) is going to fly over most heads), and I really disagree with criterion used (P1013) for any of these cases, for the reason I gave above (I think 1013's aliases make it more clearly not appropriate for this). Swpb (talk) 14:35, 12 March 2018 (UTC)
@Jheald: The wedge driving this proposed split is de facto (Q712144)/de jure (Q132555), which fell out of the deprecation of no label (P794). I'd be happy to leave them with sourcing circumstances (P1480), which already accepts the extremely similar official (Q29509043)/non official (Q29509080), but a certain Jura1 wasn't having that. 1480 has attracted lots of grumpiness about acceptable values in the past, and Deryck has astutely traced the outlines of why that is. This was originally just about chasing to ground one small, niggling use case from a deprecated property, but now it feels like there's an opportunity to put a larger problem to bed. Swpb (talk) 14:50, 12 March 2018 (UTC)
@Swpb: A lot of "maximum", "minimum" and "mean" qualifiers also got parked at applies to part (P518) and criterion used (P1013) depending on topic. A new property along the lines of "applies to aspect" (which we had discussed before) could capture "de facto", "de jure" etc but this new "rarely", "often" thing will mean somebody will salami-slice it again soon. Deryck Chan (talk) 15:17, 12 March 2018 (UTC)
@Deryck Chan: I'm concerned we're losing the thread here. I think we have a lot of agreement on the primary split (source vs. underlying nature). Re any further split(s), I don't think we're close yet to agreeing where they belong, in a way that will be used consistently. I want to keep examining that, but in the mean time, I think it's prudent to cement progress where we can. Let's make an "underlying nature" property that can capture all these things that are not about the source, without precluding more refinement down the road. Swpb (talk) 16:12, 14 March 2018 (UTC)
@Deryck Chan: Swpb (talk) 18:46, 30 March 2018 (UTC)
@Swpb: Sorry I've been AFK for the past month. I'm happy to defer to your judgement on this. Deryck Chan (talk) 15:01, 17 April 2018 (UTC)
@Deryck Chan: No problem. I've created the proposal here. Swpb (talk) 12:54, 20 April 2018 (UTC)

Dutch Translations[edit]

Can't you two consider to also change the German and Dutch translations to "Beschaffungsumstände" and "inkoopomstandigheden", as just saying "qualifier/valuation of statement" can't give us any useful informations? --Liuxinyu970226 (talk) 04:42, 22 March 2018 (UTC)

I don't speak a word of German or Dutch, so I'll leave that to someone who does. Swpb (talk) 12:57, 22 March 2018 (UTC)
German one is changed by Pasleim, I'm still waiting for responds of Dutch one (@Lymantria, Mbch331, Multichill, Romaine, Sjoerddebruin, Taketa: any ideas on it?) --Liuxinyu970226 (talk) 04:46, 29 March 2018 (UTC)
I am falling in the middle of the discussion, and it is not clear which property/item is referred to for this label, but "inkoopomstandigheden" seems a weird word in this matter. Which property/label needs a translation to Dutch here? Romaine (talk) 04:58, 29 March 2018 (UTC)
Romaine, I mean that P1480 needs to alter its Dutch translation, currently it says "waardering van bron" which (sorry for using Google MT) means valuation of source, but P1480 doesn't have "valuation" alias, nor on any other properties. --Liuxinyu970226 (talk) 05:05, 29 March 2018 (UTC)
(ec) I've figured out we need to change the translation of sourcing circumstances (P1480), but I do agree with Romaine that "inkoopomstandigheden" is a strange translation. I think the current Dutch translation isn't that wrong based on the English label. Mbch331 (talk) 05:08, 29 March 2018 (UTC)
I would suggest to translate into "Kwalitatieve waardering van bron" which is more close to the French than English label. "Circumstances" in the English label (as well as translated into German) for me is slightly confusing (source is written under high pressure of time - source is written in stormy wheater conditions - those are circumstances...). That may be the case with more nonnative speakers. Perhaps a more narrow term can be found. Lymantria (talk) 06:30, 29 March 2018 (UTC)

Ready to be closed?[edit]

After the creation of nature of statement (P5102) is there still an issue left which has to be discussed? --Pasleim (talk) 16:21, 15 May 2018 (UTC)

If @Deryck Chan: wants to explore further splits, we could do that here, but otherwise, I'd be fine with seeing this closed. Good work, all! Swpb (talk) 13:22, 16 May 2018 (UTC)
@Swpb, Pasleim: I guess we still need to decide whether maximum (Q10578722), upper bound (Q21067467) and other items that qualify a numerical value should belong to criterion used (P1013), applies to part (P518), object has role (P3831), or nature of statement (P5102). Deryck Chan (talk) 13:30, 16 May 2018 (UTC)
True, and perhaps more importantly, how will we enforce consistent application of that decision? I'm agnostic as to which property is used, but I do think consistency is important. Swpb (talk) 14:15, 16 May 2018 (UTC)
I don't see maximum (Q10578722) or upper bound (Q21067467) in the tables above, this sounds like a new issue? But criterion used (P1013) probably works for most numeric values anyway. ArthurPSmith (talk) 15:04, 16 May 2018 (UTC)
@ArthurPSmith, Swpb: Try this query? It lists all uses of maximum (Q10578722) as the value of a qualifier. [2] Those instance of (P31) qualifiers are almost certainly wrong. Most of those applies to part (P518) and determination method (P459) should probably get moved to criterion used (P1013). We can put constraints on any commonly misused combinations to shepherd people in the right direction. --Deryck Chan (talk) 16:54, 16 May 2018 (UTC)
That works for me. I can't promise when I'll have time to help with those moves, but I endorse the plan. Swpb (talk) 17:31, 16 May 2018 (UTC)