I finally made the batch upload I was planning for a while. I’m the kind of person who likes to try out a new technology first and then figure out the problems… (I think I should have made this test batch smaller). My data had some issues–I ended up removing all of the external links that I added as official websites. You can see some of the uploads on my Wikidata contributions page.
My batch had about 2400 reconciled items. We had matched 711 through mix’n’match; I used the property for the MLCAD as a proxy for mix’n’match reconciliation… I think? I also tried to reconcile using birthdates as a way to match items. That means that some 1300 items were ones that I matched using the reconciliation tool in OpenRefine. Some of these, at least several hundred, were incorrect. Other Wikidata users reverted some of my batch edits because they were reconciled to the wrong person. I would like to update my data with their reversions. Is there a way to download which items were reverted and un-reconcile them in my openrefine file?
One thing you could do is go to the EditGroups page for your batch (which you should be able to find here) and then export the list of edits to CSV using the button at the bottom.
In that CSV table there will be a “reverted” column which indicates whether the edit has been reverted so far.
A few words of caution on this:
- edits reverted “manually” will not be detected as reverted. For instance, if your batch added a statement, and another editor deleted that statement without using the undo button in Wikibase, then your edit will not be detected as reverted
- similarly, edits reverted using the “rollback” functionality are not detected as reverted
- if an edit is reverted, it stays reverted in EditGroups even if the reverting edit was reverted itself
I hope it helps!
I am still working on cleaning up the mess I made. Right now, rather than worry about roundtripping data, I just want to check the matches that I made with OpenRefine which also ended up with edits to Wikidata. Even though I included all the matched items in my potential batch edit, not all matched items ended up having enough data to create an edit. Is there a way to facet based on if my account edited the wikidata item via OpenRefine?
Yes you could! To do so, I would:
- create an OpenRefine project from the list of edits you made: https://editgroups.toolforge.org/api/b/OR/c2886fa10ff/edits.csv (that’s for a single batch - you could import more of them in the same project though). Let’s assume this OpenRefine project is named “My Wikidata edits”.
- in your original project, create a new column based on the column where your edited items are. You can use the expression
cross(cell.recon.match.id, 'My Wikidata edits', 'title').cells['reverted'].value.
- you can then create a facet on that column. It contains “True” when an edit was made and was already reverted, “False” when an edit was made and not reverted yet, and will be blank for items that have not been edited yet
You can read more about this
cross function, which lets you access data from other projects here:
thank you!! this is amazing and exactly what I needed ! And now I have a new OpenRefine superpower. Is there any way to display more than 1000 items, or should I facet around that?
(If anyone else uses your code, they should replace the single quotes with double quotes)