Turns out computers can distinguish between toxic and non-toxic comments after all.
Before executing the file, download and extract glove.6B.100d.txt from https://www.kaggle.com/danielwillgeorge/glove6b100dtxt
You can train the model by running toxic_comment_classification.py and implement it in your projects.
