There's a dataset with comments classified to be toxic or not.
A neural network can be used to predict a sentiment.
Let's scrap recent comments from DEV.TO posts and see what gets the highest rank.
Data is being scrapped with a help of requests and pyquery.
A classifier is being trained on "Jigsaw Multilingual Toxic Comment Classification" dataset. Model architecture is a recurrent model. GPU hardware is being provided for free by kaggle platform.