Re: Bug#1014908: ITP: gender-guesser -- Guess the gender from first name
Hi Philip (2022.07.19_20:51:36_+0000)
> > In other words, if you don't pick a culture, the _global_ dataset (ie, the
> > default one) must not assume either. It takes a lot of work to prepare
> > such a database, that's why this package is good to have.
>
> What exactly is it supposed to be good for?
Seems perfect for statistics from large data sets. The data in those
kind of things is always messy as hell. Full of mistakes and
misalignment to whatever lines you are trying to draw through it.
But, on aggregate, it does give you results, if you can control for the
problems.
A lot of research is done like this, "data science" they call it.
> From what I can tell, at best it produces harmless nonsense, and at
> worst it will cause people to be misidentified in ways that will vary
> between mildly humourous to significantly hurtful.
On an individual level, but on a large scale, you'd expect things to
average out. Unless it has a systemic bias towards any particular
output.
SR
--
Stefano Rivera
http://tumbleweed.org.za/
+1 415 683 3272
Reply to: