HateWatch allows moderators to create and deploy Reddit bots and assign them to their subreddits to begin monitoring for hate speech. Comments and submissions feed into a binary classification model to detect if the content contains hate speech. We leveraged GCP's AutoML to train the model using data from hatespeechdata.com, which resulted in a 91% accuracy score.
Technologies Used:
GCP AutoML
Vue.js
Bot deployment user interface
Scanned comments and their hate speech likelihood