recommender

This AI-Powered Product Recommendation System intelligently analyzes data sourced from Reddit and YouTube to deliver structured insights for informed purchasing decisions. By sifting through reviews, comments, and discussions, the system helps users comprehend product advantages and disadvantages before committing to a buy.

Key Components

The architecture of the system is built around three significant components:

1. Backend (Recommender): A Python-based service that manages data collection, AI processing, and serves API endpoints.
2. Frontend: A Next.js web application that offers an interactive user interface, enhancing user experience.
3. Nginx: A reverse proxy that efficiently routes requests between the frontend and backend.

Notable Features

• User authentication and account management services
• Integration with Reddit for gathering product discussions
• Collection of video reviews and comments from YouTube
• AI-driven analysis of product reviews providing insights
• Structured recommendation outputs featuring pros, cons, and overall guidance
• Capabilities for search history tracking and analytics

Technology Stack

Frontend

• Next.js
• React
• TypeScript
• Tailwind CSS

Backend

• FastAPI
• SQLAlchemy
• Langchain
• OpenAI API
• Async PRAW (for Reddit)
• YouTube Data API

Infrastructure

• Docker & Docker Compose
• PostgreSQL
• Nginx

API Endpoints

The backend exposes several API endpoints for various functionalities:

• GET /api/search?query=<product_name>: Fetch product reviews and analysis based on search queries.
• POST /api/register: Register a new user to access system features.
• POST /api/token: Acquire an authentication token.
• GET /api/users/me: Retrieve current user profile information.
• GET /api/reddit/status: Verify the Reddit authentication status.
• GET /api/user/recent-searches: Access a user's search history along with analytical insights.

Current Development Status

This project is a work in progress, currently facing some limitations including:

• A restricted range of product categories available for analysis
• Occasional inaccuracies in AI-generated insights
• Performance optimization required for handling larger datasets.

Invitations for contributions to enhance the system are welcome.

This project serves as a valuable tool for those seeking intelligent recommendations and insightful data, harnessing the power of AI and community discussions.