This project was created for the 2024 Collide Hackathon hosted by Digital Wildcatters. Not only did we win 1st place, but we also won the Most Creative and Best UI/UX awards, and our team was awarded $1,750 in prize money!
- View our Devpost submission here.
Create a content ranking engine for Collide. Data includes posts, comments, and fake user profiles.
Users on Collide need recommendations. The goal for this challenege is to create a recommendation engine that shows a user posts and content tailored to their role and interests.
- Use the data from the user profiles, comments, likes/dislikes table to create a a summary statement about the user.
- We can use this to add context to a system prompts in Collide.
- Follow recommendations based on location/interests/engagement.
- This will be a discovery mechanism for Collide.
Important
Ensure sure both source/client/example.env and source/server/shared/example.env are renamed to .env and are properly configured before hosting locally.
cd source/client # If not already in the client directory.
npm i -y # Install dependencies.
npm start # Start the client.cd source/server # If not already in the server directory.
python3 -m venv venv # Create a virtual environment.
source venv/bin/activate # Activate the virtual environment.
pip install -r requirements.txt # Install dependencies.
python3 app.py # Start the server.- Obtain more analytics for entire user base.
- Add trending posts/topics as recommendations.
- Optimize loading speeds on the client by using memoization.
- Add a skeleton loading animation while the content is being fetched.
- Create a more sophisticated algorithm that displays different (but still relevant) user and post recommendations on every refresh.