No Nonsense Browse Freely

Working
Extension

Inspiration

We often see that many people make homophobic comments online in various comment sections, forums etc. Sometimes it disturbs our peace of mind for rest of the day and it affects our productivity. So I thought of creating a chrome/edge extension that can help user to remove potential harmful comments using NLP and browse freely without worrying about offensive comments

What it does

The goal of the project is to create a extension that can block homophobic comments on any website. So when user loads a page the offensive comments will be removed just like how adblockers blocks the advert.

How we built it

The main task was to gather data to create a NLP model. I collected data from Hugging face https://huggingface.co/datasets/classla/FRENK-hate-en Dataset It includes around 8000+ comments that are specifically related to LGBT and the data is labelled mainly in 2 categories namely "Offensive" and "Acceptable" Then this data was used for training the model using Spacy library and the model as then saved as pickle file Colab Link - https://colab.research.google.com/drive/1NGNx2LlBc6nKTIK2ckyQwQ4KAZacKDTp#scrollTo=6qwQdmKiDEzl Colab Later this model was downloaded and Flask API was created to get result about any particular text. I have hosted api in Heroku but due to space issue I have to use locally in this demo. https://github.com/TusharAMD/pride-hacks-2-api At last chrome extension was created and all Text Elements in javascript DOM was checked and if it contains offensive text it was removed

Challenges we ran into

I have hosted on heroku and was planning to upload extension in chrome webstore but I have free tier of heroku and my app is taking memory more than allowed quota. But still I am trying to solve the issue