In research, we often build off the work of others. In this case we used groupings from a previous approach termed NamePrism. The NamePrism approach learns a low-dimensional representation of names. The representation was used to group countries, and we used those groupings.
-
-
Prikaži ovu nit
-
In this case, though the _groupings_ were derived from data, the terms for the groups were arbitrary. In this case, using those terms was not accurate. It also meant that we offended folks. I'm the corresponding author on this preprint and it was my fault.
Prikaži ovu nit -
We've updated the preprint to use a coding for these groups instead, and we've removed the terms which were assigned to them.pic.twitter.com/WnXy09k75k
Prikaži ovu nit -
We're also working on an expanded analysis that includes affiliation information as well. This will allow us to examine disparities at the nation level without grouping and also to examine whether where someone works affects these disparities. This will take us a few weeks.
Prikaži ovu nit -
I want to thank
@iddux@hajirasouliha@aanaqvi@MaryRBoland@michaelhoffman and@baraaorabi who gave us feedback. Also others - there was a lot of twitter traffic so I likely missed comments. We hope to continue to improve the manuscript. I apologize to everyone who we offended.Prikaži ovu nit
Kraj razgovora
Novi razgovor -
Čini se da učitavanje traje već neko vrijeme.
Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.