https://mix-n-match.toolforge.org recently got a new or improved import function. It's described at m:Mix'n'match/Import.
Maybe it's possible to improve the csv you provide for Digital Dictionary of Surnames in Germany ID (P6597) values.
Mix'n'match could attempt to run a re-import on a weekly basis.
Yes, I also saw the update via twitter! I manually updated the catalogue on Monday one week ago, it worked well. (Though I prematurely selected “update weekly”, so now every Monday an update job runs with static data.)
I’m also interested in getting a fully automated update job to run, but I have to talk to a colleague about the changes needed for that. I’m sure we’ll get around to it, but not with highest priority. In the meantime, I’m sure I’ll run manual updates at least occasionally.
(I just did a manual update again. The recurring job, which wasn’t ready yet, is deactivated again. I also purged the automatches, because none of them were with instance of (P31) family name (Q101352).)
I had tried to set up an automated imported at https://mix-n-match.toolforge.org/#/import based on the urls provided on the property talk page, but the tool didn't seem to find column headers.
BTW, maybe you want to include a column "P282" with the value "Q8229" in the file.
I will try to create some of the remaining ones. I noticed some entries I had created previously, but someone tried to correct my spelling ;)
Also, some entries in MxM don't have a type defined, maybe this lead to mismatches.
BTW, default description in English is "family name" (all lowercase).
Thank you for the item creations! I don’t keep a record of exact numbers, but that must have been around 6000 items?
Thanks also for your suggestions for the import file. Yes, the CSV table we provide currently has no header; we used the format as it was specified by Magnus in 2019. Adding P282 seems a good idea. Regarding the description, I faintly recall that having an English description is a mismatch to the catalogue main language, which is German. (I just never got to following up on that until now.) In my view, having German as the main language of the M’n’M catalogue seems appropriate – then, the description probably should be »Familienname«.
Also, good catch regarding the missing types. I hadn’t noticed that. The type is missing on ca. 9000 items, which is the number added last week with the new import tool. I’ll try to set the type (again?) in the catalogue editor, or if necessary for each item individually in the import.
I mean to test the changes in the import table first with manual uploads. The next chance to test new data will be March 1. Once the details are settled, I’ll look into if/when my colleague can get around to changing the format permanently.
I tried to run an update on the existing MxM catalogue, but it didn't seem to work. Maybe it's because I didn't create it or because there is some bug. For existing names, now that most are matched, it doesn't matter that much. It could be interesting to have new ones automatically loaded into MxM (when the format and the import function work), but the current approach gives a similar result.
"Chart by item creation date" on the property talk page shows when the items were created. Tricky ones are still where we have some existing incomplete or incorrectly merged one.
At Wikidata:WikiProject_Names/reports/given_names/Italian, I set up a few checks for given names. I can try to adapt more as complex constraints for the property.
I also added the newly items to a few persons as family name (P734) values. This chart shows when the items for these people were created.
Yes, automating the M’n’M import fully is what I’m also keenly interested in. The DFD is set to continue publishing around 400 new articles every two weeks for the next 14 years. Having each publication automagically in M’n’M would be nice. I’ll certainly try to make progress on this when time allows :-)
At #4200, I did another attempt with MxM. With correctly defined columns, import seems to work ;) Feel free to experiment with the catalogue.
Tabbed import file was:
id name description type P282 note 1009673 Melnykov family name Q101352 Q8229 actual data from MxM 10103 Hussein family name Q101352 Q8229 actual data from MxM 1021719 Melnyczenko family name Q101352 Q8229 actual data from MxM 103306 Tay family name Q101352 Q8229 actual data from MxM 104370 Lichteblau family name Q101352 Q8229 actual data from MxM 106519 Gollob family name Q101352 Q8229 actual data from MxM 9999999 Testing family name Q101352 Q8229 fake name without id 9999998 Aaker family name Q101352 Q8229 existing item without id 1 Müller family name Q101352 Q8229 sample from property 1095930 von Poppen family name Q101352 Q8229 sample from property
mul:"Testing"Also, I finally managed updating https://mix-n-match.toolforge.org/#/catalog/2844 adding "writing system" to entries. Somehow I couldn't update the type. Maybe simply using P31 instead could help. Accordingly automatches at https://mix-n-match.toolforge.org/#/list/2844/auto still match people or disambiguation pages.
BTW, there seems to be some bug with native label (P1705), I wouldn't add it to all entries in MxM yet. See Topic:W3oxnq257tke6xi4.
Thanks for your investigations! I followed your examples for a test run on catalog 4200 with the newest entries, which looked good. Therefore I updated the main catalog 2844 with the current complete list. I skipped the native label column for now, as per your caveat.
There’s currently no progress on our part to change the lists we provide to the new format and fully automate the update. However, now that I’ve tried it out and it works, I think I could do the update manually from time to time — as a temporary stopgap. I hope we’ll get around to this eventually.
Magnus would have to do some changes to make it feed into MxM automatically as well.
I did some more stats at Property_talk:P6597/numbers.
Géopatronyme ID (P3370) was populated last week.