[Re] Exploring the Role of Grammar and Word Choice in Bias Toward African American English (AAE) in Hate Speech Classification

First, review the README files in the "Datasets" and "Dictionaries" directories, to retrieve data and other files needed to execute the notebooks.

Then, you can run the Python notebooks in the "Notebooks" directory. We have numbered the files in the order in which they are to be run so as to replicate the exact results as our experiment.

1. 01_Reserve_Resources_on_Chameleon: Has the steps to provision resources on the Chameleon cloud. (optional - you can run this in another environment)
2. 02_Retrieve_twitter_data: Has the steps to retrieve the Twitter data
3. 03_Train_BERT_Models: Has the steps to train the BERT classifiers on both the datasets
4. 04_Train_word2vec_model_and_create_word_replacement_dictionary: Has the steps to train the word2vec models and to create the replacement dictionaries.
5. 05_Predictions_on_uncensored_and_censored_data: Has the steps to get the final predictions
6. 06_Plot_results: Has the steps to plot the results in the report.

We worked on this project as a part of the Machine Learning Reproducibility Challenge' 2022: https://paperswithcode.com/rc2022

Published at: https://openreview.net/forum?id=MjZVx7a0KX-