You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hello Debiaswe Research Team, Thank you for making the code related to your paper available. This is very helpful! I am writing to seek clarification on analyzing gender bias in word vectors associated with professions.
In your paper, you suggest using cosine similarity between a given profession vector and the top PCA component. I am trying to replicate the same in the wiki context. Unfortunately, I am getting results opposite to expected.
For example, when I compute the cosine similarity between the waitress vector (or nurse vector) and the top gender principal component, I get a -ve score. However, when I compute the cosine similarity between the same profession vector and she - he vector (as you show in the example here), I get a +ve score.
I am confused about why the sign flips when using PCA and straightforward gender vector. I request your help.
Thank you!
sbs
The text was updated successfully, but these errors were encountered:
Hello Debiaswe Research Team, Thank you for making the code related to your paper available. This is very helpful! I am writing to seek clarification on analyzing gender bias in word vectors associated with professions.
In your paper, you suggest using cosine similarity between a given profession vector and the top PCA component. I am trying to replicate the same in the wiki context. Unfortunately, I am getting results opposite to expected.
For example, when I compute the cosine similarity between the waitress vector (or nurse vector) and the top gender principal component, I get a -ve score. However, when I compute the cosine similarity between the same profession vector and she - he vector (as you show in the example here), I get a +ve score.
I am confused about why the sign flips when using PCA and straightforward gender vector. I request your help.
Thank you!
sbs
The text was updated successfully, but these errors were encountered: