Since 1969, only 279 people have become Master Sommeliers, highlighting the difficulty of the exam. Our AI sommelier, powered by RAG (Retrieval-Augmented Generation), makes it easier by giving you wine recommendations based on your taste. Whether you’re a wine expert or just enjoy a glass now and then, we’ll help you find the perfect bottle.
After watching Somm (2012), a documentary about master sommeliers and their examination process, I was inspired to create a RAG chatbot that could identify wines the user would like, just like a master sommelier. Finding a dataset that aligned with this exact inspiration was the starting point for my project.
You can watch the demo video (sommdemo.mp4) in the public directory. Alternatively you can watch it in Youtube here.
You can use the live site at Somm.
To run the project locally, follow these steps:
-
Clone the repository:
git clone https://github.com/ganbnuray/somm.git cd somm -
Run the development server:
npm run dev # or yarn dev # or pnpm dev # or bun dev
-
Open your browser: Navigate to http://localhost:3000 to see the result.
: Sampling the Kaggle wine reviews data
and
: Creating vector embeddings
: Storing vector embeddings & similarity searches
: UI components
: API calls & generating chat responses
and
: Functionalities and routing
: User login & signups
: Deployment
-
: Getting wine reviews and & sampling
Note: Given the substantial volume of data, I utilized a random sampling approach to get 100 wines to manage and use it effectively for this project.