Wikidata:Property proposal/Refine date

refine date edit

Originally proposed at Wikidata:Property proposal/Event

   Done: refine date (P4241) (Talk and documentation)
DescriptionNarrow down the precision of the date
Data typeItem
Domainqualifiers of time properties
Allowed valuesItems corresponding to:
Example
  • Master of the Bigallo Crucifixion (Q1172702) - floruit (P1317): "second half of 13th century"
  • Cui Bai (Q709216) - floruit (P1317) "mid-11th century"
  • Sheng Mao (Q3481876) - date of birth (P569) "late 13th century"
  • Adriaen van Utrecht (Q367970) - date of death (P570) "Autumn 1652"
  • Hieronymus Janssens (Q3135313) - date of death (P570) "Summer 1693"
  • Planned useAs I am importing data stored in Commons creator templates, I will be able to import dates I can not encode using current model
    See alsosourcing circumstances (P1480)
    Motivation

    Per discussion at Wikidata:Project_chat#How_to_narrow_down_imprecise_dates. I am in the process of importing data from Commons infoboxes to Wikidata. Commons for over a decade was using c:template:other date to encode and translate dates that do not follow ISO format. I am trying to make Wikidata date storing model more compatible existing date storing model used by Commons. Jarekt (talk) 16:03, 31 August 2017 (UTC)[reply]

    Discussion
    I am fine with creation of 8 items for 4 seasons on 2 hemispheres and using those instead. --Jarekt (talk) 18:08, 31 August 2017 (UTC)[reply]
    Yes, it should be "northern hemisphere spring" etc. I think. --Marsupium (talk) 19:32, 31 August 2017 (UTC)[reply]
    I've adapted the allowed values accordingly. --Marsupium (talk) 13:16, 4 September 2017 (UTC)[reply]
    A phrase that someone died "between 1550 and 1560" might just mean that the last book by the author was finished in 1550 and in 1560 his wife was refereed to as "Widow", so we know precisely that the death occurred between those 2 dates. Phrase that someone died in the second half of of 1550s means that we do not know much but it is most probable that someone died within those 5 years. The date ranges are the same but your lever of certainty is not. Users on Commons were dealing with this issue since 2009 when the c:Template:Other date was started. They chose to support and translaate all or most commonly phrases even if there might be some overlap in their meaning. --Jarekt (talk) 04:14, 4 September 2017 (UTC)[reply]
    (For reference: indroduction diff) That someone "died in the second half of of 1550s" should be interpreted as someone "died not before 1555 and not after 1559" and this should be certain. If that is not certain sourcing circumstances (P1480)circa (Q5727902) should be used. --Marsupium (talk) 13:16, 4 September 2017 (UTC)[reply]
    •   Support, there should be defined list of possible values. Usage will be as qualifier? JAn Dudík (talk) 09:19, 4 September 2017 (UTC)[reply]
    • strong   Support. Urgently needed, to merge values in from Commons & round-trip them back to there again.
    I agree with Jarekt about the importance of preserving "first half of" from the original source -- it suggests there may be more uncertainty about the date than implied just by the numerical limits.
    On the other hand, numerical limits are very useful for sorting, so ideally the data should also be given in that form. One might want to use the referencing property inferred from (P3452) to indicate that these numbers have been calculated by us, rather than being stated directly by a source. On the other hand, for a specific reference like that to be possible, would require that the calculated numbers would have to be in a separate statement of their own. That may be overkill. Instead, one might perhaps understand the inference to be case, if earliest date (P1319) and latest date (P1326) appear on a statement in addition to this new property -- and with the expectation that if all three do appear, Commons should render text using {{Other date}} rather than numeric limits. Jheald (talk) 23:10, 4 September 2017 (UTC)[reply]
      Comment I've now found myself qualifying an inferred numerical date of "circa 1650" with object named as (P1932) "mid-seventeenth century" and a reference, which may also be a solution. I agree that numerical dates are useful for sorting (and also for queries looking for items within a date range). - PKM (talk) 20:06, 5 September 2017 (UTC)[reply]