3 ms·
> about as unbiased as we're going to get. You can easily force the model to be more unbiased. Just add a filter that flips the gender of words, evaluates the
by klipt 4y ago
> about as unbiased as we're going to get.
You can easily force the model to be more unbiased. Just add a filter that flips the gender of words, evaluates the hate score for both the original and flipped version, and averages the results.
Guaranteed to give the same score regardless of the gender mentioned.
- ineptech 4y agoClever idea, but I don't think this would work very well on real posts. Consider a model that rates "typical woman driver" as hateful, because that phrase appears in a lot of argument threads with lots of downvotes. Your approach would average its score with that of "typical man driver", which will presumably be very low, not because it's less hateful but because it just rarely shows up in the training corpus.
- klipt 4y agoIf you're worried about the average score being too low, you could just take the maximum of the two scores instead?