bigbang
bigbang copied to clipboard
datactive
handling for 'unknown' gendered names
The current gender detector script will label many names as being of 'unknown' gender.
[ ] @nllz has requested a feature that would, when using the detector over a large set of names, export the gendered lists of names, and especially the 'unknown' list.
That would open the list to inspection to see if any names could be labeled, or the gender detector otherwise improved.
[ ] A related feature might support the use of an additional, custom data file assigning names to (known) genders.
I've been doing this by outputting a CSV file (after entity resolution and gender estimation), manually making changes (to add/correct known gender, or to manually consolidate names/addresses that I know to be the same person) and then importing that CSV back into a dataframe.
I do think it would be possible and potentially useful to researchers who are studying overlapping communities (like IETF and W3C groups), to be able to import a CSV file with email addresses known to the researcher along with information the researcher manually annotates (gender, nation of origin, etc.).
