Page MenuHomePhabricator

Add a « Gregorian mean year » ( Q87193822 ) unit to the Wikidata conversion table to normalize values that uses this unit
Open, MediumPublic

Description

The annum unit is used in Wikidata statements to represent half life of isotopes, this is shown by this query. 1 annum = 1 gregorian mean year = 31 556 926 s The value is also cited by the NASA in this document as used currently for calendar calculation by them.

This unit is used in the dataset used to source these half-life statement, as shown in this document (if I’m not wrong) in the name « tropical year ». The number of seconds indicated in the item and of the « gregorian mean year » used in the document corresponds.

Amongst the different units used to represent half life statements of isotopes, it’s basically the only one that is not normalized as shown by this other query

This basically blocks the possibility to use normalization in this query that is a part of a work in progress work to build a Wikipedia Module that uses Wikidata data to maintain a frwiki table of isotope from Wikidata data. (The workflow of which would be, generate a json dataset from the query service, use listeria bot to import this in an frwiki, use Lua to load the json file from this page. This works, as a prototype, actually — see the module and a page where it’s used). It’s a comfort in the query to use the normalization possibility not to have to deal with the unit. Currently the isotopes with high life life are not colored in the sandbox page due to this.

There is a difficulty with the « tropical year » unit, it’s that in the broader sense it’s a variable quantity. But it’s not a problem if we use several items with different length in seconds and decide, as it’s done in the paper cited above, that the number of seconds is that one. Other items can eventually represent a different number of seconds for different tropical years (like the « tropical year 1900 » mentioned in the paper).

Of course it’s possible to convert without normalization one way or another, but considering the legitimate use in datasets, the importance of the unit, I think it’s worth adding it to the conversion table anyway as another variant of « year », even if the number of uses is not that high.

Request : add the item gregorian mean year to the translation table to be able to use this unit. This is NOT the « annum » unit used in the statements that motivated the query because the linked Wikipedia articles cite different possible lengths for the year definition, so a fresh item was created.

Event Timeline

@Ladsgroup Do you know where the Wikidata conversion table is stored and how to add new entries?

@Ladsgroup Do you know where the Wikidata conversion table is stored and how to add new entries?

I wondered a while ago, the answer is unitConversionConfig.json and I added the fact on the wikibase doc on mediawiki.

I think this would be rather easily doable.

Indeed. I don’t know if a unit has previously be added to the translation table after the first initialization, but the problem seem to be that creating a ticket and pinging everyone for such simple task seems like an overkill procedure. I don’t know if someone less community involved or a newbie would even try to do something like this.

Can we imagine a procedure with more visibility with a simple set of criteria on wether or not a unit is eligible for inclusion in this translation table ?

In fact the unit corresponds to the « mean gregorian year », and I created a fresh item for this, to be sure there is no pollution from inconsistent datas on Wikipedia articles.

TomT0m renamed this task from Add a "tropical year" unit to the Wikidata conversion table to normalize values that uses this unit to Add a « Gregorian mean year » ( Q87193822 ) unit to the Wikidata conversion table to normalize values that uses this unit.Mar 6 2020, 5:02 PM
TomT0m updated the task description. (Show Details)

I stopped reading after the first paragraph of the "Description" because it is false. I quote it here, in case it is edited, so my comments will remain clear.

The annum unit is used in Wikidata statements to represent half life of isotopes, this is shown by this query. 1 annum = 1 gregorian mean year = 31 556 926 s The value is also cited by the NASA in this document as used currently for calendar calculation by them.

Based on comments at https://www.wikidata.org/wiki/Wikidata:Contact_the_development_team#%C2%AB_annum_%C2%BB_conversion_? I surmise that 31 556 926 s is based on this paper by Holden et al. In that paper the length of the tropical year near AD 2000 is asserted to be 31 556 925.445 SI seconds. But the mean Gregorian year is not the same as the tropical year.

As the NASA document (although intended for K-12 education) correctly states, the mean length of a Gregorian calendar day is 365.2425 calendar days. If we define a second to be 1/86400 calendar day, this is equal to 31,556,952 seconds. The length of these days has been measured in various ways, initially by simply observing the rising and setting of the Sun; later with astronomical observations of the Sun and stars to calibrate mechanical clocks in observatories. Today, atomic clocks are used but leap seconds are introduced at irregular, unpredictable intervals to keep calendar days in alignment with the observed rotation of the Earth on it's axis. So no precise SI equivalent can be given; it all depends on what period of time you decide to average over, and how much the rotation of the Earth on it's axis disagreed with atomic clocks during the averaging period.

Looking a bit further, Holden et al. give 31 556 925.9747 as the number of ephemeris seconds (a predecessor to the SI second) as the mean length of a tropical year near 1900. When rounded to the nearest second, this agrees with the length of a year adopted by Audi et al. on page 6.

Gehel triaged this task as Medium priority.Sep 15 2020, 8:06 AM