User talk:Multichill/Archives/2018/April

From Wikidata
Jump to navigation Jump to search


Moved to User talk:FloraFlavia#Uffizi paintings, please keep conversations in one place. Multichill (talk) 21:49, 2 April 2018 (UTC)

Names from Web umenia

Hi there! Could you filter out names that end in "st." from their transfer to de, en, es, fr and nl labels/aliases? It is the Slovak/Czech wikt:starý … (I've found it at Q51549254 (Special:PermanentLink/662163875).) All the best, --Marsupium (talk) 07:21, 7 April 2018 (UTC)

It will only add the label when no label is present in that language. No aliases are added to existing items. A label to update is better than no label at all. Won't happen a lot and when it happens, it's probably a duplicate that needs to be merged anyway (like in your example). Would be interesting to run a query.
Import is mostly done, I'm just adding the missing date of birth and date of death to the items. Keep an eye on this query. It now returns 160 results and will probably shrink a lot when these items get merged. I see a couple of obvious duplicates that weren't caught by Mix'n'Match yet. Multichill (talk) 10:50, 7 April 2018 (UTC)
Thanks for your response!
  • I disagree with “A label to update is better than no label at all.” Copying labels is very easy as it is done by bots like Edoderoobot as well. Updating is comparatively difficult. It is one point more where I'd prefer higher quality for the expense of slower growth.
  • Okay, if no aliases are added to existing items are, but those names are added to newly created items and those are duplicates they will end up as aliases in target items after (hopefully accomplished) merging. It doesn't make this better in any way.
  • Obviously Marv1N has discovered the problem earlier and it perhaps applies to more parts of names. I've also spotted "ml." (Slovak/Czech wikt:mladý) at Q51548270. A refined query with this that includes aliases and excludes cs and sk as languages finds 227 names with probably wrong language. I think we should remove them. When will this be done?
  • You perform matching for bot adding creator (P170) statements also using Web umenia creator ID (P4887), so those names aren't necessary for that, or are they?
  • How many items did you bot create based on Web umenia? It has really created a lot of duplicates. I hope I can investigate the duplicates later further.
This got now a lot of questions. Thanks in advance for your response, --Marsupium (talk) 18:07, 8 April 2018 (UTC)


that was not my edit--Oursana (talk) 00:16, 11 April 2018 (UTC)

That's weird, the interface must have played a trick on me when I restored anonymous (Q4233718). Sorry about that. Multichill (talk) 09:18, 11 April 2018 (UTC)

Reminder: Share your feedback in this Wikimedia survey

WMF Surveys, 01:40, 13 April 2018 (UTC)

Albrecht Dürer

Hi Multichill, there is an item for him at Q5580. No need to make multiple ones [1].
--- Jura 17:08, 8 April 2018 (UTC)

hello Multichill ! I also merged quite a bunch of duplicates, created by your bot (see most of those from webumania. Was there a problem with M'n'M ? --Hsarrazin (talk) 08:58, 11 April 2018 (UTC)
That's where Albrecht Dürer came from. It just seemed to be the most obvious tip of the iceberg. np.
--- Jura 16:36, 11 April 2018 (UTC)
That import seems to have given some more and some unexpected duplicates. Probably a combination of records missing, not yet matched records and me firing up the bot too early. I'll do some more merging. Multichill (talk) 10:00, 11 April 2018 (UTC)
there is an easy list of merges to do (with checking first of course) here, because of Web umenia creator ID (P4887) matched by user:Spinster ;) --Hsarrazin (talk) 15:35, 11 April 2018 (UTC)
OK, most of them merged now... even if the report is slow to update ^^ --Hsarrazin (talk) 14:20, 12 April 2018 (UTC)
More show up at Wikidata:WikiProject Visual arts/Artists same name. Thanks for any help to clean the recent and older dupe mess! --Marsupium (talk) 21:09, 24 April 2018 (UTC)
What does "too early" mean? Why does the bot do what M'n'm is for at all? The bot doesn't properly check for items yet existing before creating new ones and should better leave this to humans entirely then. --Marsupium (talk) 21:20, 24 April 2018 (UTC)

Yale University Art Gallery import

Hi and thanks for doing this useful import with your bot. I know this has been raised before and you're probably working on it, but I've stumbled into some items where the final character of the name was missing, from both the label and the title field. See the histories of Peonies and Narcissus (Q49358981), Lotuses (Q49358870) and Chrysanthemums (Q49359092). I haven't looked at how many more there are. Best wishes, MartinPoulter (talk) 12:44, 20 April 2018 (UTC)

@MartinPoulter: yes, had a problem in a regex that ate up the last character for some of the imports. You might be able to find a couple more on the subpages of Wikidata:WikiProject sum of all paintings/Collection/Yale University Art Gallery, but some of them are quite hard to spot. Multichill (talk) 13:08, 24 April 2018 (UTC)

Please slow down

Hi Multichill, your bot is editing to fast at the moment and we have ongoing dispatch lag, can you make sure it goes down well under 30 edits per minute? Cheers, Hoo man (talk) 16:45, 27 April 2018 (UTC)

I doubt my single bot is causing dispatch lag and under 30 edits per minute is extremely slow. Maybe you should provide a proper solution to phab:T184948 instead of the current hack?
Are you folks even working on fixing the underlying problems? The software/infrastructure needs to be scaled up so it can handle the current amount of edits. Multichill (talk) 19:01, 28 April 2018 (UTC)