User talk:Multichill

From Wikidata
Jump to navigation Jump to search
Welcome to my talk page. Use it; don't send me e-mail.

I reply to messages left on my talk, on my talk page. If I left a message on your talk page, I will reply there (unless you specify otherwise).

Bot creates artist duplicates[edit]

"Creating artist based on RKD: Painter with works in RKDimages and date and place of birth known"-style item creations produce duplicates: no label (Q55792459). I'd like to ask you not to perform them. If we want to have any automatic artist/person item creations we need consensus before, at least a WD:RFP/BOT. I'd be happy to have that, but in a very careful implementation that reduces dupe creation to the absolute minimum that can be achieved. Unfortunately, manual checking identity of two items consumes a lot of time and we already have a huge backlog of pairs to be checked and probably many dupes more that aren't even detected by current methods.
That said, I'd be glad to help in fine-tuning mechanisms for an artist item creation bot that uses all means possible to create items for artists actually not yet existing! Thank you, --Marsupium (talk) 13:10, 31 July 2018 (UTC)

RKDartists ID (P650) is used extensively (71.000 links at the moment according to the talk page). Every once in a while I create some of the missing painters that have passed the notability criteria (in the edit summary). I've been running it for several years and usually the number of duplicates are low and easy to find because they pop up in reports like Wikidata:WikiProject sum of all paintings/Creators same name. In this case I ran it because I kept running into missing painters at Wikidata:WikiProject sum of all paintings/Missing creator with RKDimages link. I haven't changed the notability criteria for a while so only new or updated entries in RKDartists might get imported. To get an idea of what is coming in new, have a look at Wikidata:WikiProject sum of all paintings/RKD to match/Recent additions. More matches in mix'n'match would prevent some of the duplicates, but the number of suggestions might be high, the number of good matches are quite low. Tons of records in RKDartists are quite empty, so you don't know for sure if it's the same person. These records to make mix'n'match big, but wouldn't be created by my robot (not enough metadata available). I wouldn't worry about it too much about this. Multichill (talk) 16:05, 31 July 2018 (UTC)
Thank you for your long reply and sorry for responding only know properly! It is good to know that it is code used before and with the same criteria! Then I'm ok with it. :-) I saw a bunch of dupes bouncing after the creations. You know how many items you created in that batch all together? I'm not sure if we find all duplicates with current methods, but also those we find are a problem because in masses the identifying and merging is a lot of work. Perhaps the script could check for identical labels or aliases and if one is found it rather logs it so that it can be put somewhere here and checked before creation? Also if not, we can perhaps live with the current system. I hope we will get some automatic person creating system in the future that does all checks that are possible, but it has time. Best, --Marsupium (talk) 12:29, 12 August 2018 (UTC)

Give your feedback about changes to Special:Block[edit]


You are receiving this message because you are a top user of Special:Block on this wiki. Thank you for the important work that you do. There is a discussion happening about plans to improve Special:Block with the ability to set new types of blocks. To get the best design and new functions added, it is essential that people who use the tool join the discussion and share their opinions about these changes.

Instead of a full site wide block, you would be able to set a Partial Block. A user could be blocked from a single page, multiple pages, one or more namespaces, from uploading files, etc. There are several different ways to add this feature to Special:Block. Right now Important decisions are being made about the design and function.

Please review the page on Meta and share your feedback on the discussion page. Or you can reach me by email Also, share this message with anyone else who might be interested in participating in the discussion.

I appreciate any time that you can give to assist with making improvements to this feature. Cheers, SPoore (WMF) (talk) , Trust and Safety Specialist, Community health initiative (talk) 01:59, 10 August 2018 (UTC)

  • Apologizes for posting in English.

Tate paintings height and width[edit]

It seems height and width got mixed up for example for The Spiritual Form of Nelson Guiding Leviathan (Q28537451) (cf. file). And it seems to be a wider problem see this query. What can be done about it? I'd propose to sort them if there is one image (P18) value and the ratio of one of the two possible identifications of height and width matches the ratio of the file approximately and to remove the statements otherwise. Perhaps those tests would be good to do on future uploads? What do you think? --Marsupium (talk) 12:55, 12 August 2018 (UTC)

@Marsupium: The data is from the (now defunct) , from to be exact. I wonder if the dimensions field is consitent height x width. If that's the case, the data could be extracted from that one, compared with the current fields and if it's swapped, fixed. Multichill (talk) 12:00, 8 September 2018 (UTC)

Our friend is back[edit]

Over to you. --Tagishsimon (talk) 00:31, 21 August 2018 (UTC)

Although, I've listed this on the admin noticeboard; you're taking a wikibreak, I think. --Tagishsimon (talk) 02:53, 21 August 2018 (UTC)
Our friend is blocked. Sorry to have disturbed you. --Tagishsimon (talk) 11:06, 21 August 2018 (UTC)
Just a bit different priorities these days. Multichill (talk) 11:55, 8 September 2018 (UTC)

Import of so many English labels... not in English[edit]

Hello! I've seen many entities imported from your bot in 2015. Thank you. Unfortunately it seems that you imported a lot of wrong English labels e.g. (look at your edit) in Q19362350, Q19362313, Q19362543, Q19362432, Q19362381, Q19362383 and so on. Did you know that? --Valerio Bozzolan (talk) 14:50, 8 September 2018 (UTC)

I think I added the missing ones for Beat. Most of the labels are given names and it's always the question to translate these or not, see for example Berlin Hauptbahnhof (Q1097). Usually I import the given names and leave it up to users if they want to translate it or not. I don't think these labels are wrong, but good to see you're interested in improving them. On this list you can see what a user with the language set to English will see. You might want to add or update the English labels. Multichill (talk) 17:11, 8 September 2018 (UTC)
Trust me. It's very difficult to understand how we can help an English Wikidatian displaying stuff like Im Sand-Felsenau, Teil der spätrömischen Rheinbefestigung, Jüppe, Teil der spätrömischen Rheinbefestigung, Axalp, mittelalterliche /neuzeitliche Alpwüstung, Oberes Bürgli, Teil der spätrömischen Rheinbefestigung, Riesi, spätbronzezeitliche Seeufersiedlung., Schaltenrain / Grossholz, hallstattzeitliche Grabhügelgruppe, Heimenholz, Teil der spätrömischen Rheinbefestigung, Kleiner Laufen, Teil der spätrömischen Rheinbefestigung, Schweizer Radio und Fernsehen, Schweizer Radio DRS, D+A (vgl. auch Basel und Zürich)..., and so on. On the other hand, it's very easy to understand that now it's nearly impossible to query which of all the monuments I have to handle need a translation and which not. --Valerio Bozzolan (talk) 18:39, 8 September 2018 (UTC)