Inspiration
Addresses the issue of skewed online ratings due to bots and sarcasm.
What it does
Calculates a "True Rating" using a 20/80 weight (20% stars, 80% sentiment).
How we built it
- Backend: FastAPI (Python), Supabase (DB).
- Frontend: React, Chrome Extension (JS).
- AI/ML: VADER (speed), DistilBERT/RoBERTa (accuracy), NLTK.
- Infrastructure: Docker.
Challenges
Solving the "Ratio Paradox" and implementing self-annealing AI.
Accomplishments that we're proud of
Creating the 3-layer architecture (Directives, Orchestration, Execution).
What we learned
Discovered that a 20/80 ratio is the "sweet spot" for stable ratings.
What's next for TrueRate.ai
Expanding to more platforms beyond Amazon.
Built With
- docker
- fastapi
- javascript
- python
- react
- supabase