Hatenot | Devpost

Demo integration - Feed with warnings of hateful content
Demo integration - Writing a hateful message
Demo integration - Dashboard showing number of blocked posts
Extension blocking hateful Twitter posts
Demo login page

Inspiration

Nowadays, a large problem on the internet is the endless supply of hate speech on social networks and forums.

What it does

We trained a machine learning algorithm, that evaluates texts and resolves its hate speech level. We exposed its output to an API, which is used by our demo social network to warn potential created and seen hate speech occurrence. We also created web browser extension that uses our API, to block real world hate speech on Twitter.

Challenges we ran into

None of us had experience with creating browser extensions and we had very limited experience with machine learning techniques, especially with text processing.

Accomplishments that we're proud of

We are really proud of creating working machine learning implementation and its application to real world cases. We are also happy that we created two use cases of this API, namely our mockup social network and especially our working custom extension.

What we learned

We learned how the text processing with machine learning works and how to evaluate different datasets according to our goals. We also learned how to create plugins to web browsers and connect them to different APIs.

What's next for Hatenot

We would like to create a stable version of our web browser plugin and also create much more categorizations and statistics of the offensive posts the user sees. We want to also implement the text recognition of picture based text and also categorization of images themselves.

Built With

Submitted to

Hack Kosice 2022
- Winner DTIS Hatemeter Challenge
- Winner [MLH] Best Hack for Social Good by BlackRock

Created by

I worked on the machine learning API using Python Flask. I also helped with training and building of our machine learning model and looked for appropriate datasets.

Michal Demko
Student of Computer security
I created the demo integration social network and helped with the development of the browser extension.

Daniel Breiner
Student of Computer Science
I created the prototype of our machine learning model using Pytorch. The challenge was working on NLP problem for the first time but we learned a lot.

Daniel Forrai
I created API in Python using Flask. Then, I worked on the browser extension, which was my first experience in this area.

Ivan Havasi
Computer science student and programmer