Authors: Priyanka Bose, Chandra Shekhar Pandey
This project validates the claims made in the paper: "Exploring the Role of Grammar and Word Choice in Bias Toward African American English (AAE) in Hate Speech Classification".
The paper claims that use of swear words impacts hate speech classification of AAE text.
To validate the same we trained two Bert Models and tested both the models on AAE tweets and stored the results. Next we made a Swear words replacement dictionary using Liwc list of swear words and the replacement of swear words would be the words which are most similar to those words by cosine similarity. Lastly we replaced all the swear words in the AAE tweets with the help of the dictionary. And then we again used Bert models to classify the tweets. We saw that there was a significant amount of reduction in the hate speech classified tweets and the results were similar to what was stated in the paper.
This project was done for Machine Learning Reproducibility challenge 2022 under the guidance of Prof. Fraida Fund at NYU Tandon School Of Engineering.
Launching this artifact will open it within Chameleon’s shared Jupyter experiment environment, which is accessible to all Chameleon users with an active allocation.Download Archive
Download an archive containing the files of this artifact.
Download with git
Clone the git repository for this artifact, and checkout the version's commit
git clone https://github.com/indianspeedster/mlr_2022.git # cd into the created directory git checkout 03ed65d7abbc5cc3a3aa24679add2c9790958327
Submit feedback through GitHub issues