Mitigating Racial Biases in Toxic Language Detection with an Equity-Based Ensemble Framework

1 code implementation • 27 Sep 2021 • Matan Halevy, Camille Harris, Amy Bruckman, Diyi Yang, Ayanna Howard

While previous work has focused on a single fairness criteria, we propose to use additional descriptive fairness metrics to better understand the source of these biases.

Descriptive Fairness