Cbasile06

Please don't remove referenced statements

Even if they are wrong such as here. Please just deprecate them with reason for deprecated rank (P2241) Thank you! --Vojtěch Dostál (talk) 05:26, 4 June 2020 (UTC)

Apologies; I changed it in that case while explicitely keeping the reference because the source is inconsistent and uses both dates (the anachronistic 1975 in the biography and the - presumably - correct 1757 in the header). The 1775 date should still have the same reference (a lot of the pages on Wikidata:Database_reports/items_with_P569_greater_than_P570 actually come from this problem). It felt wrong to have two different dates with the same reference, but maybe that is the preferred way? --Cbasile06 (talk) 08:33, 4 June 2020 (UTC)

Yeah, I guess having two statements with the same reference (one deprecated) won't hurt anything. The National library will probably use a SPARQL query to identify dates which we identified as wrong (deprecated) and hopefully fix them. They have used Wikidata in past to identify other inconsistencies like this. Anyway, the header usually only gives years, not full dates, so maybe it's technically wrong to put together (presumably correct) day and month from one part of the source and year from another part of the source. Better to keep everything there including mistakes. Thanks for working on dates though! :) Vojtěch Dostál (talk) 12:21, 4 June 2020 (UTC)

Conflation

Hi, you have edited J.E. de Sturler (Q86739702) in a way I don't seem to understand properly. You have deprecated human (Q5) by stating the item itself is conflated; and then you split J.E. de Sturler (Q86739702) into J.E. de Sturler (Q96001436) and Jacques Edouard de Sturler (Q96002245). Can you explain why you have split the item into two new items instead of keeping the original item and creating just a second? Your way of keeping the 'old' item but instead creating two new items leads to problems in several databases who link to the 'old' item. Thanks, DanielleJWiki (talk) 07:54, 7 July 2020 (UTC)

I have done this following the guideline from Conflation_of_two_people to split into two separate people, because the conflation was not recent. Furthermore, there were identifiers associated with J.E. de Sturler (Q86739702) which referred to either J.E. de Sturler (Q96001436) or Jacques Edouard de Sturler (Q96002245). So there was no good answer to know which to keep. If there was a better course of action here I would like to know. Cbasile06 (talk) 17:20, 9 July 2020 (UTC)

You can keep both as far as I am concerned. Say there was item A that included two people 1 & 2; make item A for person 1 and create a new item B for person 2. In that way for example cultural heritage institutions can themselves adjust their databases to point to either item A or B.

You have instead created item B and C and left item A to just wither in a corner. Databases now don't get any trigger that their databases are pointing to the wrong item (A) and that they should change the db to point to either B or C instead.

Is it possible to find out which items you have split like this in two new items? Thank you for your answers, DanielleJWiki (talk) 09:40, 10 July 2020 (UTC)

But if item A included people 1 & 2, why would I keep item A for person 1 instead of person 2? Some databases were linking to person 1 and others to person 2, both are wrong to link to this conflated version. I would like to understand why this scenario is worse for outside database than the one you propose. Is it because I merely deprecated the identifiers on the initial conflation instead of removing them completely? The only other entry split by me this way I believe is Malcolm Cross (Q16206224) split into Malcolm Cross (Q96021879) and Malcolm Cross (Q96021878). Cbasile06 (talk) 04:31, 11 July 2020 (UTC)