# CAN 2025 AI Football Intelligence Platform
An AI-powered full-stack platform delivering intelligent question-answering and real-time statistics for the Africa Cup of Nations 2025 through automated data scraping, semantic search, and Retrieval-Augmented Generation.
## Technology Stack
### Backend & AI
### Frontend
### AI & Data Processing
### Database & Storage
---
## Architecture
## Features
### Intelligent Chatbot
- **RAG Pipeline**: Llama 3.3 70B with context-aware responses
- **Multi-Query Retrieval**: 3-5 query variations for improved accuracy
- **Real-time Streaming**: Server-Sent Events (SSE) for ChatGPT-like experience
- **Source Citations**: Transparent answer provenance with article references
### Automated Data Collection
- **Scheduled Scraping**: Celery Beat triggers every 8 hours
- **Dynamic Loading**: Selenium handles JavaScript-rendered content
- **Incremental Indexing**: Only new articles embedded to Qdrant
- **Error Resilience**: Graceful failure handling with automatic retries
### Semantic Search
- **Vector Embeddings**: SentenceTransformers (all-MiniLM-L6-v2, 384-dim)
- **Efficient Retrieval**: Qdrant COSINE similarity across 1,760+ chunks
- **Fast Performance**: Sub-second search with 2-3 second end-to-end queries
### Tournament Dashboard
- **Live Statistics**: Real-time standings, fixtures, top scorers/assists
- **Football-API Integration**: Official tournament data feeds
- **Visual Rankings**: Gold/silver/bronze badges with animations
- **Manual Refresh**: On-demand statistics updates
---
## Project Structure
```
can2025_platform/
│
├── backend/ # Django Backend
│ ├── manage.py
│ ├── backend/ # Django settings
│ ├── accounts/ # User authentication
│ ├── chatbot/ # RAG pipeline & API
│ │ ├── rag_service.py # LangChain + Qdrant
│ │ ├── views.py # REST endpoints
│ │ └── apps.py …