User:Hsarrazin/Prénom

For the need of reflexion, I will generally use Russian as exemple of non-latin language, since I read/speak Russian, and can thus place myself and give examples in a "Russian" pov. stats on language-speakers.

Of course, all these reflections can be extended to all languages that do not use latin alphabet, and, for some parts, to some latin-alphabet languages.

Present solution

• P735 = item-type
• Property given name (P735) created on 10 july 2013, but systematic "splitting" of values a few weeks ago (very recent).
• property very widespread, almost a million (!!) items with P735 value... (on 17 October 2014)

• splitting : each written form of a given name is given a specific Item, and links are split between items
• links between all different forms through said to be the same as (P460), which does not provide linking in requests (for now)
• for people items with various written name in time or space (Popes, , use of multiple values, with "Preferred" tag
• for original non-latin form of given name (Russian born for ex.), use of cyrillic values in all labels…

Pros

• ease of statistics on each form : each item quickly provides the number of items linked for each written value…
• ease of assigning a specific value to each person, as long as there is no translation or translitteration problem

Cons

for all users

• difficult to maintain : each change on first name items will have immediate effect on thousands of items, without control...
• necessity to have multiple values for many person-items, where different written values were used in various country (Popes, Kings, artists) before WW2 (and even a little after)
• from a latin-written pov, the use of cyrillic/non-latin alphabet in latin-language labels is very difficult at use… need for many aliases just to be able to have access to those…
• absurdity of some choices… there was a recent remark on the Project chat, about "John" and all the people who certainly were not named John
• for Popes, Saints, all people with no wiki in their language, difficulty to choose a form… (for example, for Popes, the original name should probably be the la form, since it is both the official language of Vatican, and the form that is primarily chosen by new popes, then translated in other languages for communication needs… (our present Pope's name is "Benedictus", not "Benoît" or other form…)

for Russian/non-latin users

• very latin-centered pov : from a Russian pov the need for many items to translitterate the same Russian value is unreadable and preposterous
• present solution (splitting items according to written form) breaks interwiki links
• difficult to organize : necessity to have arbitrary splitting for Russian persons (for ex.), who were known in Europe under different forms of their name (all diplomats, writers, artists, politicians…) — see the "Konstantin/Constantin" discussion…

Needs

• stability in time : after installation, should not be changed on great scale, since the given name of a person is not something that changes in time, no more than one's birthdate, or so...

• interwiki links : sister-projects should be able to have iwlinks, not to be broken by each splitting/merging of given name items

• keep track of real "written value" in native language, of a person

• need for statistics, by "generic" name, by "written value"

• use of given name for string operations (building of full name, sorting items, etc.)

proposed possibilities

• keep present solution, with improvements (to be proposed)

• change type of (P735) value to monolingual or multilingual value

• use of a generic item-value (keeps iwlinks) in P735 - by merging of previously splitted written values

• with a monolingual or multilingual qualifier to keep track of "written value"...

• creation of a top item value, not to be used in person-items, but where iw-links would be put back : this would allow to keep the work that has been done in P735, and restaure iwlinks…

1. with the use of redirections from written value … - one of the problems being that for now it is not possible to add a "redirection" in an item, it is automatically replaced… - but this should evolve
2. with linking from specific written values through "subclass of (P279)" - this link is not queryable for now, but will surely be in the future, as it is the structure of describing items on wikidata - this would provide for some non-mergeable items (Etienne/Stéphane…)

my wrong... the "tree" request should allow queries :) --Hsarrazin (talk) 10:44, 31 October 2014 (UTC)