Neutralizing Biased Text

This repo contains code for the paper, "Automatically Neutralizing Biased Text".

Concretely this means algorithms for

• Identifying biased words in sentences.
• Neutralizing bias in sentences.

TODO REPLACE WITH NEW IMG

Installation

pip install -r requirements.txt

Data

Click this link to download (100MB, expands to 500MB).

Overview

harvest/: Code for making the dataset. It works by crawling and filtering Wikipedia for bias-driven edits.

src/: Code for training models and using trained models to run inference. The models implemented here are referred to as MODULAR and CONCURRENT in the paper.

Usage

Please see src/README.md and harvest/README.md.

TODO MAKE THESE HYPERLINKS