Bumblebee

• 

  Home page

• 

  Info page

• 

  Login page

• 

  Sign up page

• 

  Default logged in page

• 

  Feed page that uses our custom algorithm for matching

• 

  Searching page

• 

  Colonies page, displays various communities to become a part of.

• 

  Hive page, shows all current matchings

• 

  Message list page

• 

  Messages page

• 

  Profile page

• 

  Profile editing

Inspiration

We were inspired by the general disarray in modern-day communication. With so many connectivity and dating apps out there, it can feel overwhelming when trying to find someone you truly connect with. Bumblebee helps solve this problem by personalizing your feed with those who share your interests as well as implementing a community-based connectivity platform in the form of colonies. Users can move between various colonies, with each colony corresponding to a community of people with a shared interest. This way, you can easily find compatible friends or potential partners and populate your hive.

What it does

Bumblebee is an innovative dating and connectivity web app that uses your hobbies and interests to personalize your feed. We created 20 tags representing interests (e.g. cooking, the outdoors) and used AI to weigh each tag with a value from 1 to 10 based on each user's description of themselves. These weights were used to populate a 20-dimensional feature vector that characterizes each user. We then designed a custom algorithm that uses the vectors to compute compatibility scores between users. Our algorithm is unique in that it factors in how passionately someone feels about a certain interest when determining compatibility with others, not just how different their interests are. We achieved this by implementing a maximum likelihood estimator under a Gaussian likelihood function weighted by the square-root of each tag feature. This made our compatibility estimates robust and accurate.

Additionally, our colony feature allows users to explore various communities, with each colony corresponding to a specific tag/interest. This makes it easier than ever to connect with like-minded (or different) people and fosters a welcome sense of community.

How we built it

We used React to build our frontend and Node to build our backend. We connected our application to ChatGPT4o-mini using the OpenAI-API to convert descriptions to weighted tags, and user information was stored using MongoDB.

Challenges we ran into

We had a bit of trouble fully connecting our frontend to our backend. Getting the UI to accurately reflect changes took a while, but eventually we got everything to work. Designing the app's layout and exact functionality was also a challenge, since we were keeping track of a lot of information. For example, it was difficult to figure out how our colonies should be laid out, since we needed to fit 20 colonies on the screen.

Accomplishments that we're proud of

We are proud of building a creative and functional app. We believe our product adds unique features to a longstanding concept and that people could greatly improve their quality of life through using our app.

What we learned

We learned how to iteratively improve and test code. We also gained experience with fully fleshing out ideas into functional features, and how do deal with the complications of a more complex project.

What's next for Bumblebee

We plan to implement quality of life improvements such as small updates to the UI. We will also optimize our code and improve how information is stored in the database. Once we finish these refinements, we plan to deploy Bumblebee to the web.

Built With

• css
• express.js
• javascript
• mongodb
• node.js
• openai
• react