Content Moderation

This is a project for CSE 3000 to explore the ethics of machine learning in content moderation. In this repository you will find scripts to process and vectorize text, and then train a multinomial Naive Bayes model in order to label the text. Our team used the dataset from Kaggle's Toxic Comment Classification Challenge.

Set-Up

1. Clone the repo:

   git clone https://github.com/WillSho1/contentModeration.git
   cd contentModeration

2. Download the data set from Kaggle: Link Toxic Comment CLassification Challenge Data
3. From the download, move train.csv, test.csv, and test_labels.csv into /data/raw.
4. Run the text processing script:

   python ./scripts/textProcessing.py

5. Train the model, run predictions, and print accuracy:

   python ./scripts/main.py